Energy is a scalar quantity that describes the capacity of a system to do work or to cause change. In physics it is conserved in isolated systems, transferred between objects and fields, and converted from one form into another. The concept links motion, heat flow, light, chemical bonds, and mass, so it appears in almost every branch of science and engineering. Because it is so widely used, the word also carries metaphorical and everyday meanings that differ from its precise technical sense.
The entry first sets out what energy means and how the word developed, then describes its main forms and the laws that govern it. It covers the mechanical, thermodynamic, statistical, quantum, electromagnetic, relativistic, and nuclear descriptions of energy, and it follows the concept into chemistry, biology, Earth and cosmic systems. The later sections address units, energy resources and conversion technology, economics and policy, the role of energy in computation, and unresolved questions such as the nature of dark energy.
Meaning and etymology of the concept
The English word energy comes from the Greek enérgeia, which is built from en ("in") and ergon ("work"), and is usually rendered as "activity" or "actuality." Aristotle used the term in the 4th century BCE in his discussions of potentiality and actuality, in works including the Physics, the Metaphysics, and the Nicomachean Ethics. In his usage, energeia was contrasted with dynamis, the capacity a thing has to act. A seed is potentially a tree; a mature tree is actually one. The concept was qualitative and broad. It could describe a thing's activity in the world, a living being's flourishing, or even a kind of happiness, and it had no numerical content. Modern scholars have therefore tended to treat the Aristotelian sense as a separate philosophical concept from the physical one, although some argue that the older idea of "being at work" still sheds light on what energy is.
The physical meaning arrived much later. The Scottish physicist and physician Thomas Young introduced the word in its modern quantitative sense in 1807, in lectures he gave at the Royal Institution on collisions and related topics. He applied the term to the product of a body's mass and the square of its speed, a quantity that had previously been called vis viva ("living force"). Young's quantity is twice what is now called kinetic energy. The factor of one half was introduced by Gaspard-Gustave Coriolis in 1829, who also introduced the concept of work in its modern form. Young's choice of word was not immediately adopted by everyone, and the term took decades to become standard in mechanics, heat, and electrical science. Its acceptance was helped by the growing recognition that a single quantity could be tracked across these fields.
In modern physics, energy is defined operationally through the processes that change it. The most basic definition is that energy is the capacity of a system to do work or to produce heat, with both quantities being ways that energy is transferred between systems. Work is the transfer that occurs when a force moves its point of application, and heat is the transfer that occurs because of a temperature difference. Energy itself is not directly observed in the way that a length or a mass is. It is inferred from how it changes and from how much work or heat a system can deliver. Energy is a state function: the change in a system's energy depends only on its initial and final states, not on the path between them. This property is what makes energy useful as an accounting quantity. A system that has returned to its starting state has the same energy it started with, regardless of what happened in between.
Energy is a scalar quantity, which means it has magnitude but no direction. It is measured in joules in the International System of Units, with one joule equal to one newton-metre. Because kinetic energy depends on the square of speed rather than on velocity itself, reversing the direction of motion leaves the energy unchanged. This property lets energy be added up across the parts of a system without vector bookkeeping. The conservation of energy is expressed as a balance equation. For any region or system, the change in the amount of energy stored inside it equals the energy that enters minus the energy that leaves, whether as work, heat, radiation, or matter transport. In an isolated system no energy enters or leaves, so the total stays constant even though it may move between different stores, such as kinetic and potential forms, and between mechanical and thermal ones. This balance equation is the basis for the rest of energy physics, from the first law of thermodynamics to the conservation principles that appear in later sections of this entry.
Forms of energy
Energy appears in many forms, and physics distinguishes them mainly by where and how the energy is stored. The most basic is kinetic energy, the energy of motion: for a body of mass moving at a non-relativistic speed , it equals . Its counterpart is potential energy, which is stored in the position of a body or charge within a force field. Near the surface of a planet, gravitational potential energy is approximately for a height above a reference level. Between two electric charges, electric potential energy depends on their separation through the electrostatic force between them. Potential energy can be assigned only to conservative forces, whose work depends only on the starting and ending positions, so a raised weight can release its stored energy as kinetic energy when it falls.
Internal energy is the total of the microscopic kinetic and potential energies of all the particles in a body. It includes the translational, rotational and vibrational motion of molecules, along with the energy of their mutual interactions and of the bonds that hold them together. The part associated with the disordered, random motion of particles is the thermal energy, which increases as the temperature rises. For an ideal gas, internal energy depends only on temperature, because the interaction energy between particles is negligible. Internal energy cannot be observed directly. Its changes, however, can be measured through heating, mechanical work, or the transfer of matter.
Chemical energy is stored in the arrangement of electrons and nuclei within molecules and in the strength of the bonds between atoms. It is released when a chemical-reaction lowers the total electrostatic energy of the system. Burning methane, for example, releases about 890 kJ per mole. Nuclear energy is stored in the atomic-nucleus, where the strong interaction binds protons and neutrons together. The binding energy per nucleon is higher in medium-mass nuclei than in very heavy ones, so fission of uranium-235 releases roughly 200 MeV per event. This yields more than a million times the energy per kilogram of fuel that burning methane does.
Electromagnetic energy is carried by the electric and magnetic fields themselves. In vacuum, its energy density is , and light and other radiation transport this energy across space. At the quantum level, the energy of radiation comes in units of carried by each photon. Elastic energy is stored when a solid is deformed. A spring displaced by a distance from equilibrium, with stiffness , stores , and this energy is returned when the spring relaxes. At the molecular scale, the elastic energy of a solid is itself electromagnetic in origin, since it comes from the electrostatic forces between atoms.
The division into these forms is a matter of description rather than a fixed list of fundamental categories. The same physical process can be described in several ways depending on the scale. The energy of a chemical bond is electrostatic potential energy among electrons and nuclei, and the energy of an elastic solid is also electrostatic at the atomic scale. Whether a given energy counts as internal or external depends on where the system boundary is drawn. For the particles inside a gas container, their motion and mutual interactions form the internal energy, but for a container treated as a single body, its motion is kinetic energy. A body at rest also carries rest energy, given by mass-energy equivalence, which does not fit neatly into any of the everyday categories above. The total energy is conserved regardless of how it is labeled. Only the bookkeeping of forms changes with the chosen scale and system boundary.
Mechanical energy and work
Mechanical energy is the portion of a system's energy tied to motion and position, and its accounting starts with work. In physics, work measures energy transferred by a force that moves its point of application. Work is the transfer of energy by a force and can be calculated as the product of the force and displacement along the line of action of the force.
Work and kinetic energy
Work connects force to the motion of a body. Starting from Newton's second law, one can show that the work done on a body over a displacement equals the change in its kinetic energy, . This is the work–energy theorem: , where is the work of all forces acting together. The work-energy theorem holds for any force, constant or not, as long as W_net is the total work done by the net force. A positive net work raises the speed of a body, and a negative net work lowers it.
Gravitational potential energy is the complementary quantity for a body near a planetary surface, where it is approximately for height above a reference level. Its value depends only on position, not on the path by which the body arrived there, which is what allows it to be treated as part of the mechanical budget (see potential energy).
Conservative and dissipative forces
Some forces are special in that their work depends only on the starting and ending points. Some forces—like gravity and spring forces—have a special property: the work they do depends only on the starting and ending positions, not the path taken. Such forces are called conservative, and for them a potential energy function can be defined. For an ideal spring stretched by , the stored energy is . Because the force is conservative, this work is stored as potential energy (PEs) in the spring, and it can be fully recovered.
When only conservative forces do work, the sum of kinetic and potential energy stays constant, so . This quantity is commonly known as mechanical energy. A falling body therefore trades potential energy for kinetic energy without any net change in the total.
Dissipative forces break this bookkeeping. Friction is the standard example: it does work that cannot be recovered as stored potential energy. Friction and air resistance are non-conservative and do not have a defined potential energy. Instead, friction converts ordered motion into disordered motion of the atoms in the contacting materials. As the object slows down because of friction, its coherent kinetic energy decreases, and the internal random kinetic energy of both the object and the surface increases, manifesting as a rise in temperature. Energy removed from the mechanical budget in this way is not destroyed; it is carried by the internal, or thermal, energy of the system, as discussed in the sections on internal energy and heat.
The conversion is effectively one-way. Friction, for example, is a typical dissipative process in which mechanical energy is converted into thermal energy and cannot be completely converted back into mechanical energy. This is why a sliding block eventually stops on a table, and why mechanical energy alone is seldom conserved in practical situations.
Conservation laws and symmetry
The principle of energy conservation states that the total energy of an isolated system does not change, although it can move between forms and between parts of the system. Its standing as a law rests on repeated experimental checks. In collisions, the total kinetic energy of the bodies before and after impact agrees within measurement error once any energy converted to heat or sound is accounted for. A simple pendulum with a fixed pivot and negligible air drag swings back to the same height on each side, trading kinetic energy for gravitational potential energy and back again. For a body in a Keplerian orbit around a much heavier one, the sum of kinetic and gravitational potential energy stays constant, and that constant fixes the size of the orbit. None of these checks is a proof, however. The deeper justification comes from the link between conservation and symmetry described below.
In 1918, Emmy Noether proved that for every continuous symmetry of a physical system's action there is a corresponding conserved quantity. This result, now called Noether's theorem, gives energy conservation a specific origin. The relevant symmetry is time-translation symmetry: the laws governing a system do not change when the same experiment is performed at a later time. Other symmetries yield other conserved quantities. Invariance under spatial translation gives momentum conservation, and invariance under rotation gives angular momentum conservation. On this view, conservation laws are consequences of the structure of physical laws rather than independent postulates. Noether's theorem applies only to continuous symmetries of the action, so the statement that every symmetry produces a conserved quantity is an overgeneralization. Noether's work also raised questions about how energy should be defined in general relativity, which are treated in the relativity section.
The formal development of this idea in classical mechanics runs through the Lagrangian and Hamiltonian formulations. If the Lagrangian of a system, which is the difference between kinetic and potential energy, has no explicit dependence on time, a particular function of the coordinates and velocities is constant along every trajectory. This function is the energy. In the Hamiltonian formulation, the Hamiltonian is itself the energy function, and Hamilton's equations describe how the system moves through phase space. In this picture, is the generator of time translations: the flow it defines carries the system forward in time, and its value does not change along the flow when has no explicit time dependence. In quantum mechanics the same role appears in the Schrödinger equation, where an isolated system's state evolves through the operator . For a time-independent Hamiltonian, the energy eigenstates evolve only by a phase, which is the reason they are called stationary states.
Energy conservation becomes less straightforward when the system is not invariant under time translation. Examples include a particle in a potential that varies with time, a pendulum whose support is driven back and forth, and an atom in an oscillating light field. In these cases the Hamiltonian depends explicitly on time, the symmetry is broken, and the energy of the subsystem generally changes. The energy is exchanged with whatever external agent produces the time dependence, so the subsystem is not isolated. If the driving agent is included in the description, total energy is again conserved. In quantum mechanics, time-dependent perturbation theory treats this situation by noting that energy is not conserved and that transitions occur between the unperturbed eigenstates. A related case arises when a parameter of the system changes very slowly. The energy then changes, but a closely related quantity, called an adiabatic invariant, is approximately conserved. On cosmological scales, the expansion of the universe stretches light and reduces the energy of individual photons, which is often cited as a case where a global time-translation symmetry does not hold.
Thermal energy and the first law of thermodynamics
Energy enters or leaves a thermodynamic system in two principal ways, as heat and as work. Heat is energy transferred because of a temperature difference, moving by conduction, convection or radiation and carried by microscopic molecular motion. Work is energy transferred through an organized mechanism, such as a force moving a piston or a torque turning a shaft. Neither quantity is a substance that a body contains. Both describe transfers and depend on the path taken. The quantity a body does store is the internal energy U, the total of the microscopic kinetic and potential energies of its particles. U is a state function.
The first law of thermodynamics is energy conservation applied to thermodynamic systems. It states that the change in internal energy equals the energy added as heat minus the work the system does on its surroundings. With Q the heat added to the system and W the work done by it, the law reads
Chemists often use the IUPAC convention, ΔU = q + w, in which w is work done on the system. The two forms are equivalent, but the sign of the work term differs between them, which is a frequent source of confusion when textbooks are compared. For a cyclic process that returns the system to its starting state, ΔU is zero even though Q and W individually need not be. The first law sets no restriction on the direction in which a process runs. That role belongs to the second law, discussed in a later section.
Enthalpy
Most chemical reactions run in open vessels at roughly constant atmospheric pressure. Under those conditions the expansion work is P ΔV, and the bookkeeping is simplified by the enthalpy,
For a process at constant pressure with only expansion work, the change in enthalpy equals the heat exchanged, ΔH = q_p. A negative ΔH marks an exothermic reaction that releases heat, and a positive ΔH marks an endothermic reaction that absorbs it. The standard enthalpy of combustion of methane is about −890 kJ/mol, and the formation of liquid water from hydrogen and oxygen is about −286 kJ/mol. Because enthalpy is a state function, the enthalpy change of a reaction can be obtained by summing the enthalpies of formation of products and subtracting those of reactants. This is Hess's law, and it is why tabulated thermochemical data are so useful. Enthalpy changes are the basic quantities in reaction energetics.
Classification of processes
Thermodynamic processes are commonly grouped by which quantity is held fixed. In an isothermal process the temperature stays constant. For an ideal gas, internal energy depends only on temperature, so ΔU = 0 and in a reversible isothermal expansion the heat absorbed equals the work done, W = nRT ln(V₂/V₁). In an adiabatic process no heat is exchanged, so ΔU = −W. A gas that expands adiabatically cools, and a gas that is compressed adiabatically warms. Diesel engines use this effect, since rapid compression of the air heats it enough to ignite the fuel. In an isobaric process the pressure is constant, so the work is P ΔV and the heat exchanged equals ΔH. In an isochoric process the volume is constant, so no expansion work is done and the heat supplied equals ΔU. These four cases are idealizations, but many real processes approximate one of them, and the classification determines which simplifications of the first law apply.
Heat capacity
Heat capacity relates the heat supplied to the resulting temperature change through C = δQ/dT. For a body of mass m with specific heat capacity c, the heat needed for a temperature change ΔT is Q = mcΔT. Liquid water has a specific heat of about 4.18 J/(g·K), so raising 1 kg of water by 10 K requires about 42 kJ. The value of heat capacity depends on the constraint applied. The heat capacity at constant volume is C_V = (∂U/∂T)_V, and at constant pressure it is C_P = (∂H/∂T)_P. For an ideal gas, C_P − C_V = nR. A monatomic ideal gas has C_V = (3/2)R, about 12.5 J/(mol·K), and C_P = (5/2)R, giving a ratio γ = C_P/C_V = 5/3. Diatomic gases near room temperature have C_V ≈ (5/2)R and γ ≈ 1.4. For solids and liquids the two heat capacities differ only slightly. The reasons for these values are traced to the number of degrees of freedom that can store energy, a point developed in the statistical section.
Entropy, the second law, and energy quality
The first law of thermodynamics says that energy is conserved, but it does not say which energy transformations can actually happen. The second law fills that gap. Its early formulations came in the 1820s and 1850s, and Rudolf Clausius gave it its mature mathematical form. In 1850 he stated that heat does not pass spontaneously from a colder body to a hotter one. Between 1854 and 1865 he showed that for any cyclic process, the integral of heat exchanged divided by absolute temperature cannot be positive. For reversible cycles it is exactly zero:
This result implies that a quantity exists which depends only on the state of a system, not on the path taken to reach it. Clausius named it entropy in 1865, and defined its change for a reversible process as . The name was built from a Greek root meaning "transformation," and he chose it because it sounded close to "energy." His 1865 summary of the two laws put it in compact form: the energy of the universe is constant, and the entropy of the universe tends toward a maximum. The statement that entropy must not decrease applies most cleanly to adiabatically isolated systems, and later textbooks extended it to isolated systems in general.
Sadi Carnot's earlier analysis of heat engines had reached a related conclusion without the concept of entropy. An engine working reversibly between a hot reservoir at temperature and a cold reservoir at has an efficiency that depends only on those two temperatures, not on the working substance:
Temperatures must be expressed on the absolute scale. This expression is an upper bound: no real engine operating between the same reservoirs can exceed it, and real engines fall short because of friction, finite-rate heat transfer, and other irreversibilities. The bound is a direct consequence of the second law, which is why the Carnot cycle became a reference standard in the theory of heat engines. Raising or lowering increases the attainable efficiency, which explains the engineering interest in high-temperature steam and combustion processes.
Not all energy is equally useful. Exergy, also called available energy or availability, is the maximum useful work that a system can deliver as it relaxes to equilibrium with a specified reference environment, usually taken as ambient temperature and pressure. Energy that is already at ambient conditions has zero exergy, even though it contains energy. Exergy is not conserved: in any irreversible process some of it is destroyed, and the amount destroyed is proportional to the entropy generated. If is the entropy produced and the environmental temperature, the lost work is . This is why a heat exchanger with a large temperature difference wastes more useful work than one with a small difference, even when no energy leaves the system. The concept, sometimes described as a measure of energy quality, is central to modern analysis of power plants, refrigeration, and chemical processes, where exergy analysis identifies the stages that waste the most potential for work.
The second law also supplies a direction to time. Processes such as the mixing of two gases, the flow of heat from hot to cold, and the dissipation of friction are effectively irreversible: their reverse does not occur spontaneously. In an isolated system the total entropy does not decrease, so these processes tend to proceed in one direction, which is what physicists call the arrow of time. A common heuristic describes high entropy as energy being more dispersed and low entropy as energy being concentrated. The heuristic is useful, but it should not be taken literally, because entropy is defined through heat and temperature rather than through a notion of "disorder." The statistical explanation of why entropy increases, developed by statistical mechanics, is treated in a later section.
Statistical interpretation of energy
Macroscopic energy is an aggregate. A litre of air at room temperature contains roughly 10^22 molecules, each translating, rotating, and vibrating while exerting forces on its neighbours. The gas's internal-energy is the sum of these microscopic kinetic-energy and potential-energy contributions. Temperature and pressure are statistical averages over the same enormous population. Fluctuations about these averages scale roughly as 1/√N, so for N ≈ 10^22 they are of order 10^-11 of the mean, far below anything measurable. This is why the macroscopic description of energy in terms of heat, work, and the first-law-of-thermodynamics works without tracking individual particles, and why statistical-mechanics can link it back to microscopic dynamics.
The central tool is the boltzmann-distribution. For a system in thermal equilibrium at absolute temperature T, the probability of finding it in a microstate of energy is
where k is the boltzmann-constant (1.380649 × 10^-23 J/K, fixed exactly since the 2019 redefinition of the SI) and Z is the partition function. Higher-energy states are exponentially less likely, and the energy scale that matters is kT. At low temperature the system concentrates in its lowest states; as T rises, more states become accessible. For two levels separated by 1 eV at 300 K, the population ratio is about e^-38.7, roughly 10^-17, so the upper level is effectively empty. The same counting of microstates underlies Boltzmann's expression S = k ln W for entropy, which connects the energy distribution to the second-law-of-thermodynamics. The Maxwell–Boltzmann speed distribution of gas molecules is one application of this result.
The equipartition-theorem is the classical shortcut to the same physics. In thermal equilibrium, each quadratic term in the energy, or each degree of freedom, carries an average energy of . A monatomic ideal gas has three translational degrees of freedom, so each atom averages and the internal energy is . A diatomic gas adds two rotational degrees of freedom, and classically its vibrational mode adds one more kT because it has both kinetic and potential terms. The classical total is therefore per molecule. Measured heat capacities of hydrogen near room temperature match about of the gas constant, showing that the vibrational mode is frozen out. Quantum mechanics explains this: when level spacings are comparable to or larger than kT, the higher states are seldom populated, and the mode contributes little (energy-quantization, quantum-mechanics, heat-capacity). Solids show a similar failure. The Dulong–Petit value of about 3R per mole is reached at high temperature but falls at low temperature, a behaviour that Einstein's and later Debye's quantum models reproduced. The most dramatic failure concerned radiation. Applying equipartition to electromagnetic modes in a cavity predicted unbounded energy at short wavelengths, the so-called ultraviolet catastrophe, which Planck resolved by introducing energy quanta (blackbody-radiation, planck-constant).
For an ideal gas, the mean translational kinetic energy per molecule is . This depends only on temperature and not on the molecular mass or the identity of the gas. Temperature is therefore best understood as a measure of the average energy per degree of freedom, not as energy itself. The mass does matter for speed: the root-mean-square speed is
For nitrogen molecules (m ≈ 4.65 × 10^-26 kg) at 300 K, this gives about 517 m/s. The relation holds exactly for an ideal gas with negligible intermolecular interactions. In liquids, solids, and dense or strongly interacting systems, potential energy between particles contributes significantly, and temperature is then defined more generally through entropy, as 1/T = ∂S/∂U.
Quantum energy
By the end of the nineteenth century, classical physics treated energy as a continuous quantity that could take any value. Experiments on thermal radiation exposed the limits of that picture. The Rayleigh–Jeans law, derived from classical equipartition, predicted that a blackbody would radiate without limit at short wavelengths, a result known as the ultraviolet catastrophe. In December 1900, Max Planck found a formula that matched the measured spectrum by assuming that the oscillators in the cavity walls could exchange energy only in whole multiples of a minimum packet, , where is an integer, is the frequency and is the Planck constant, fixed at J·s in the modern SI. Planck himself regarded discreteness as a formal device for counting states rather than a property of the radiation, so the idea that energy comes in quanta was not yet a physical claim about light.
Light quanta and the photoelectric effect
In 1905 Albert Einstein took the step Planck had avoided. In a paper he called heuristic, he proposed that light itself carries energy in localized packets, each with energy . This accounted for the photoelectric-effect: electrons are ejected from a metal only when the light's frequency exceeds a threshold, the maximum kinetic energy of the ejected electrons rises linearly with frequency, and it does not depend on intensity, which changes only the number of electrons. The relation , where is the work function of the metal, summarizes these observations. Robert Millikan's experiments of 1916 confirmed the linear relation quantitatively, though he had expected it to fail. The quanta of light were later named photons, and Einstein received the 1921 Nobel Prize in Physics for his work on the photoelectric law.
Discrete levels in bound systems
Quantization is most striking in bound systems, where a particle is confined by an attractive force. In 1913 Niels Bohr applied the idea to the hydrogen atom: the electron can occupy only certain energy levels, and it emits or absorbs a photon when it moves between them, with the photon energy equal to the difference between the two levels, . For hydrogen the levels are , where eV is the electronvolt, so the ground state lies 13.6 eV below the energy of a free electron at rest. The resulting spectral lines, such as the Balmer series in the visible range, match the observed frequencies. Bohr's orbit model was an ad hoc construction, and it was superseded by full quantum mechanics. Free particles, by contrast, have a continuous range of energies.
Eigenvalues and zero-point energy
In 1926 Erwin Schrödinger wrote down a wave equation whose solutions give the allowed energies of a system. For a stationary state, the equation takes the form , where is the Hamiltonian operator. Its allowed values are eigenvalues. For a particle in a one-dimensional box of length , these are with . For a quantum harmonic oscillator of frequency , the levels are . The lowest level is not zero. The ground state retains an energy , known as zero-point energy, which follows from the uncertainty in position and momentum. Its effects are observable: helium-4 stays liquid at atmospheric pressure down to absolute zero and must be pressurized to solidify, and zero-point fluctuations of the electromagnetic field give rise to the measurable Casimir force. The role of vacuum energy in cosmology is treated under dark-energy.
Electromagnetic energy and radiation
In classical electromagnetism, fields themselves carry energy. In vacuum or a linear medium, the energy density (energy per unit volume) of the electric and magnetic fields is
where and are the electric and magnetic field strengths, is the vacuum permittivity and is the vacuum permeability. The result is in joules per cubic metre. The formula follows from the work needed to build up the fields, and it is the field-level form of the energy bookkeeping set out in energy density and electromagnetic theory. A charged capacitor is a familiar case: its stored energy resides in the electric field between the plates, and in a propagating wave the electric and magnetic terms contribute equally.
Energy also moves through space. The energy flux, or energy transported per unit area per unit time, is described by the Poynting vector
measured in watts per square metre. Its direction is perpendicular to both and , so it points along the direction of propagation. For a plane wave in vacuum, the flux equals the energy density multiplied by the speed of light . The same relation shows that electromagnetic waves exert radiation pressure: a fully absorbing surface experiences a pressure equal to the flux divided by , and a perfectly reflecting surface experiences twice that. This pressure is small in everyday settings but is measurable, and it is the classical counterpart of the momentum carried by each photon. Sunlight at the top of Earth's atmosphere delivers roughly 1.36 kW per square metre, a value set by the Sun's luminosity and the Earth–Sun distance, and this flux is the starting point for the Earth's energy budget.
The thermal emission of matter is the most important case of electromagnetic energy exchange. An idealized body that absorbs all incident radiation, a blackbody, emits a spectrum that depends only on its temperature. Planck's law gives the spectral radiance as
where is the Planck constant, is frequency and is the Boltzmann constant. The derivation required quantized exchange of energy, which Planck introduced in 1900 to resolve the ultraviolet catastrophe of classical theory. Two consequences follow from the curve. Wien's displacement law gives the peak wavelength as . A body at about 5800 K, such as the Sun's photosphere, peaks near 500 nm in the visible range, while a body at about 310 K, such as the human body, peaks near 9 µm in the infrared. The total power radiated per unit area is given by the Stefan–Boltzmann law, , with . Real surfaces radiate less than this ideal value, and the ratio is called emissivity.
Radiation is also a distinct mechanism of heat exchange, and it does not require a material medium. Conduction and convection transfer thermal energy through matter, but radiant transfer works across a vacuum. This is why the Sun warms the Earth across about 150 million kilometres of empty space, and why a thermos flask uses a silvered, low-emissivity surface to suppress heat loss. Between two surfaces, the net radiative exchange depends on the fourth powers of their absolute temperatures, so the process is strongly nonlinear: doubling a temperature does not simply double the radiated power. This nonlinearity matters in planetary energy balance, where the outgoing infrared emission of the Earth must roughly match absorbed sunlight, as discussed in the greenhouse effect and radiative forcing.
Across the full range of frequencies, from radio waves to gamma rays, electromagnetic radiation is classified by the energy carried by each quantum, . The electromagnetic spectrum therefore describes not only how waves differ in wavelength but also how their energy per photon varies. Low-frequency radiation carries small quanta and is produced efficiently by thermal sources at low temperatures, while high-frequency radiation requires far higher temperatures or atomic and nuclear processes. These relations connect the classical field description, the energy flux, and the quantum picture of light, and they form the basis for the thermal and spectroscopic applications of electromagnetic energy.
Mass–energy equivalence and relativity
Before 1905, physics treated mass and energy as quantities with separate accounts. Mass was conserved in chemical reactions, while energy was tracked through mechanics, heat, and electromagnetism. Einstein's special theory of relativity brought the two together. In the process, it changed the meaning of both special relativity and mass–energy equivalence.
Einstein's 1905 result
Einstein published the core of special relativity in June 1905. In a short sequel that September, he asked what happens to a body at rest that emits light. He imagined a body sending out two equal light pulses in opposite directions, each carrying energy . Requiring the result to hold in a moving frame, he found that the body's inertial mass drops by . From this he concluded that a body losing energy as radiation loses mass . Written for the energy a body holds at rest, this gives the rest energy , where is the rest mass and is the speed of light. The famous form appears in the 1905 paper only in this simplified sense; Einstein's general argument was stated in other terms.
The derivation has been debated since its publication. Planck (1908) objected that Einstein's argument relied on an approximation that was valid only at first order. Laue (1911) criticized the treatment of extended, non-spherical bodies. In 1952 Ives argued that the reasoning was circular. Historians also note that earlier writers had proposed mass–energy relations, so the priority of the idea is contested, even though Einstein's paper was the first to attempt a general proof.
Experiments later confirmed the idea clearly. In nuclear reactions, the total mass of the products is slightly less than that of the reactants, and the difference appears as kinetic energy of the fragments. The 1932 disintegration of lithium by protons, carried out by Cockcroft and Walton, is a classic example. Similar mass deficits underlie nuclear binding energy, which explains why fission and fusion release energy. Pair annihilation, in which an electron and a positron convert into photons, shows the reverse process.
The energy–momentum relation
For a free particle, energy, momentum, and rest mass are linked by
Here is relativistic momentum, . The total energy is , where is the Lorentz factor. The kinetic energy is the difference . For speeds much smaller than , a binomial expansion gives , so , the familiar Newtonian expression. Rest energy is only a special case of the general relation. A massless particle such as a photon has and no rest energy.
Frame dependence and conservation
Energy and momentum combine into a four-vector, with energy as the time component. Its value therefore changes from one inertial frame to another. The energy of a moving car or a gas depends on the observer, and the rest mass does not. The quantity , called the invariant mass of a system, takes the same value in every frame. This matters in particle accelerators: the energy available to create new particles in a collision is set by the invariant mass, not by the energy in the laboratory frame alone.
Energy conservation survives in relativity, but in the form that the total four-momentum of an isolated system stays constant. A collision that conserves momentum and energy in one inertial frame does so in all of them, even though the share of kinetic energy changes.
Energy conservation in general relativity
In general relativity, the status of energy is more limited. As Noether's theorem shows, a conserved energy follows from invariance under time translation. In general relativity, time-translation invariance requires a suitable symmetry of the spacetime, expressed as a timelike Killing vector. Stationary or asymptotically flat spacetimes, such as the field around an isolated star, admit a well-defined total energy. In expanding cosmological spacetimes, such a symmetry is absent, so there is no global energy conservation law for matter, radiation, and gravity together. The Einstein field equations still imply that the energy–momentum tensor is locally conserved. However, the gravitational field itself does not supply a local energy density that is unambiguous in every coordinate system. This is one reason why defining energy in gravitational settings remains a subject of active research.
Nuclear energy
Nuclear energy is the energy bound up in the atomic nucleus, and it is released through the same mass–energy relation that governs all energy changes. A nucleus holding Z protons and N neutrons weighs less than its separate nucleons would. The difference, called the mass defect, is the mass that was converted into the energy holding the nucleus together. The nuclear binding energy equals . Binding energy per nucleon rises steeply from hydrogen through the light elements and peaks at iron-56, at about 8.8 MeV per nucleon. Whether a nuclear reaction releases energy therefore depends on which way it moves nuclei along this curve. Joining light nuclei and splitting heavy ones both raise the average binding energy per nucleon, so both can release energy. Per event, the amounts are very large: the energy of a single fission is on the order of tens of millions of times the energy of one carbon atom burned in fossil fuel.
Fission occurs when a heavy nucleus, most commonly uranium-235, absorbs a neutron, becomes unstable, and splits into two lighter fragments while emitting neutrons and gamma rays. A typical event releases about 200 MeV. Most of this, roughly 170 MeV, appears as the kinetic energy of the fragments, which lose it to the surrounding fuel as heat. Published totals vary by a few MeV depending on whether delayed decay energy, neutron capture and neutrinos are included. Because fission releases on average about two and a half neutrons, the reaction can sustain itself. A nuclear-chain-reaction is critical when, on average, exactly one neutron from each fission goes on to cause another. Slow, or thermal, neutrons are far more likely to split U-235, so reactors use moderators to slow the neutrons. Only a small fraction of neutrons, about 0.66 percent for U-235, are delayed. They come from the radioactive decay of fission products, and they give operators enough time to adjust the reaction rate. Enrico Fermi achieved the first self-sustaining chain reaction in 1942.
The same physics yields two very different applications. In a nuclear-power reactor the chain reaction is held in a delayed-critical state, and the heat from fission is used to raise steam for turbines. Fission products keep decaying after shutdown, so reactors must continue to remove heat for some time afterwards. In a nuclear weapon, the chain reaction is prompt-supercritical, and the energy is released uncontrolled in roughly a microsecond. Fission and the weapons it enabled were first demonstrated in combat at Hiroshima and Nagasaki in 1945, and the technology has since been governed by a mix of safety regulation and arms-control agreements.
Fusion is the joining of light nuclei, and it powers stars. In the Sun's core, the proton–proton chain converts hydrogen into helium and dominates in stars up to about 1.3 solar masses. Heavier stars rely more on the CNO cycle, which uses carbon, nitrogen and oxygen as catalysts. Fusing four protons into one helium-4 nucleus releases about 26.7 MeV. The obstacle is electrostatic repulsion. Two protons must approach within a few femtometres, facing a barrier of roughly 1 MeV, while the thermal energy in the solar core is only about 1.3 keV per particle. Fusion proceeds because of quantum tunnelling, and the enormous number of collisions in a stellar core makes even a tiny tunnelling probability produce a substantial reaction rate. The process is so slow that the Sun will take more than ten billion years to exhaust its core hydrogen.
On Earth, gravity cannot confine the plasma, so fusion research relies on much higher temperatures and on deliberate confinement. The most studied fuel is a mixture of deuterium and tritium, which fuses at lower temperatures than hydrogen does. Two approaches are magnetic confinement, in devices such as tokamaks and stellarators, and inertial confinement, in which lasers or other drivers compress a small fuel capsule. Laboratory experiments have produced more fusion energy than the laser energy delivered to the fuel, but commercial fusion power has not yet been demonstrated.
Radioactive decay is the spontaneous process by which unstable nuclei move toward more stable configurations. In alpha decay, the nucleus emits a helium-4 nucleus, so its atomic number drops by two and its mass number by four. The alpha particle typically carries a few MeV and stops quickly in matter. In beta decay, a neutron converts into a proton, emitting an electron and an antineutrino (β−), or a proton converts into a neutron, emitting a positron and a neutrino (β+). The atomic number changes by one, but the mass number does not. Because the energy is shared among the electron, the neutrino and the recoiling nucleus, the electron's energy spectrum is continuous. In gamma decay, a nucleus that is left in an excited state after alpha or beta decay drops to a lower state and emits a high-energy photon, without changing its number of protons or neutrons. The total energy released in any decay is set by the mass difference between the initial and final states, and this energy is shared among the decay products.
Chemical and biological energy
Chemical energy is the part of a substance's internal energy that is released or absorbed when atoms are rearranged into different bonds. Breaking bonds requires energy input, while forming stronger or more numerous bonds releases it. A chemical reaction is exothermic when its products have lower enthalpy than its reactants, so heat is released, as in the combustion of hydrocarbons. It is endothermic when the products sit higher, so heat is absorbed, as when ammonium nitrate dissolves in water. Energy diagrams show this by plotting potential energy against the progress of the reaction. Reactant and product levels differ by the enthalpy change, and a barrier rises between them. The height of that barrier is the activation energy, which governs how fast the reaction proceeds, not whether the overall energy balance is favourable.
Catalysts lower the barrier by providing an alternative pathway with a lower peak. In living cells, the catalysts are enzymes. A catalyst speeds up the forward and reverse reactions equally, so the equilibrium position and the net energy change are unchanged. The rate dependence on the barrier is captured by the Arrhenius relation , where a lower gives a larger rate constant at the same temperature. This is why enzymes can make reactions that would otherwise be too slow for life proceed at body temperature.
Living systems do not usually run on the heat of combustion directly. They use ATP as a short-term energy carrier. ATP consists of the base adenine, the sugar ribose and three phosphate groups. Its hydrolysis to ADP and inorganic phosphate has a standard biochemical free-energy change of about −30.5 kJ/mol, though the value in cells is considerably more negative because the ATP/ADP ratio is kept far from equilibrium. The energy is not released by "breaking a high-energy bond," a common misconception. It comes from the greater stability of the products. Cells use this energy through coupling: an unfavourable step, such as the phosphorylation of glucose, which requires roughly +14 kJ/mol, is linked to ATP hydrolysis so that the combined Gibbs free energy change is negative. Membrane pumps such as the sodium–potassium pump use the same principle to move ions against their gradients.
Photosynthesis is the main route by which biological systems capture external energy. Light-dependent reactions in the chloroplast thylakoid membranes absorb photons with pigments such as chlorophyll. The absorbed energy drives electron transfer, which builds a proton gradient and produces ATP and the reducing agent NADPH. The Calvin cycle then uses these to fix carbon dioxide into sugars. The overall process, in simplified form, converts six molecules of carbon dioxide and six of water into one glucose molecule and six of oxygen, with light supplying the energy that is stored in the sugar's chemical bonds. Glucose has a heat of combustion of roughly 2.8 MJ per mole, which is the reservoir that respiration later draws on.
Cellular respiration releases this stored energy in stages. Glycolysis in the cytoplasm yields a net of two ATP per glucose. The citric acid cycle in the mitochondrial matrix oxidizes the resulting acetyl groups and transfers electrons to the carriers NADH and FADH₂. The bulk of the ATP is then made by oxidative phosphorylation. In the electron transport chain embedded in the inner mitochondrial membrane, electrons pass through protein complexes, which pump protons into the intermembrane space. The resulting gradient drives ATP synthase. By commonly used textbook estimates, about four protons return through the synthase for each ATP made, so one NADH yields roughly 2.5 ATP and one FADH₂ roughly 1.5. Oxygen serves as the final electron acceptor and is reduced to water. Adding the earlier stages, a eukaryotic cell typically obtains about 30 to 32 ATP per glucose. The theoretical maximum of 38 is not reached because of proton leakage and the cost of transporting molecules across membranes.
Taken together, these pathways form a continuous energy cycle. Light energy is converted into chemical bond energy, that bond energy is transferred to ATP and then coupled to work, and the remainder is dissipated as heat. Only part of the energy from each step is captured in usable form, a pattern consistent with the general limits discussed under the second law of thermodynamics.
Energy in Earth systems
Climate can be read as an energy account. Averaged over the whole planet, sunlight reaching the top of the atmosphere amounts to roughly 340 W/m², and about 30 percent of that is sent straight back to space. Clouds and the atmosphere account for most of the reflection, with the surface contributing less, so only about 240 W/m² remains to be absorbed. The Earth's energy budget is the accounting of these flows, and it is the reason the Sun rather than the planet's interior governs the surface climate. Earth's internal heat flow, roughly 47 terawatts, is only a few hundredths of a percent of the energy the planet receives from sunlight.
In its simplest form, the budget balances when the absorbed shortwave radiation from the Sun equals the longwave infrared radiation the planet emits. Because the planet is warm, it radiates in the infrared, at wavelengths from about 4 to 100 micrometres for almost all of its outgoing flux. Emitting this radiation is the only way the planet loses energy to space, so the rate of emission relative to absorption determines whether the climate is warming or cooling. Satellites such as NASA's CERES instruments measure the top-of-atmosphere flows directly, and the difference between them is known as the Earth's energy imbalance.
Greenhouse effect and radiative forcing
Greenhouse gases absorb outgoing infrared radiation and re-emit part of it downward, so the surface must become warmer before the emitted flux again matches the absorbed sunlight. Climate scientists quantify this shift with radiative forcing, defined as the change in net irradiance caused by an agent such as a change in gas concentrations. An increase in incoming radiation and a decrease in outgoing radiation both raise the forcing. The IPCC's best estimate for a doubling of atmospheric carbon dioxide is about 3.7 W/m², and the forcing from CO₂ grows roughly logarithmically with concentration, so each successive doubling adds a similar amount. Forcing is not temperature itself: the warming it produces depends on climate sensitivity and on feedbacks such as changes in water vapour, ice and clouds.
The measured imbalance has grown over time. NASA-derived CERES data cited by one summary rose from about +0.42 W/m² in 2005 to about +1.12 W/m² in 2019, and more recent analyses report that the imbalance has roughly doubled in recent decades. The increase reflects both rising greenhouse gas concentrations and secondary changes such as reduced ice cover and cloud cover in some regions.
Albedo and absorbed solar energy
The fraction of sunlight a planet reflects is its albedo, and on Earth the Bond albedo is about 0.30. Surface properties strongly control the local value. Fresh snow, ice and desert sand reflect much more sunlight than open ocean or forest, so changes in ice cover alter how much energy the surface absorbs. Clouds are the largest single contributor to planetary reflection, but their net effect depends on their altitude, thickness and type: thick, bright low clouds tend to cool by reflecting sunlight, while high thin clouds mainly trap outgoing infrared and warm. Because the reflecting surfaces are often the same ones that respond to warming, albedo changes can amplify or dampen the initial energy imbalance, which is why they feature prominently in feedback analyses.
Energy transport by atmosphere and ocean
The Sun heats the tropics far more than the poles, so the atmosphere and oceans must move energy poleward to keep the latitudinal gradient from growing without limit. In a satellite snapshot from September 2008, equatorial regions absorbed about 200 W/m² more than they reflected or radiated, while polar regions gave off about 200 W/m² more than they absorbed. Winds, storms and the general atmospheric circulation carry heat and moisture from the tropics toward the mid- and high latitudes, and ocean currents do the same over longer timescales. Surface water warmed in the tropics is carried by currents toward cooler regions, and colder water returns toward the equator at depth.
The ocean also stores much of the excess heat that the greenhouse effect produces. Estimates converge on roughly 90 percent of the accumulated excess heat being taken up by the oceans, a share that reflects water's high heat capacity compared with air. Ocean heat content, measured from the surface down to 2,000 metres, has been rising since about 1960, and a NASA estimate put the total heat gained since 1955 at about 372 zettajoules as of late 2024. Averaged over the whole Earth surface, the heat gain of the global ocean in 1993–2024 corresponds to roughly 0.7 W/m². Heat taken up by the ocean is not lost. It is redistributed through currents and later returns to the rest of the system by melting ice shelves, evaporating water or reheating the atmosphere, so it can keep warming the planet for decades after it was absorbed.
Units and measurement
The joule (symbol J) is the SI unit of energy. It is defined through mechanics: one joule is the work done when a force of one newton moves its point of application one metre in the direction of the force, so . In SI base units, . The same quantity appears in electrical terms, where one joule is the energy dissipated when a current of one ampere flows through a resistance of one ohm for one second. The unit honours James Prescott Joule, whose experiments on the interconversion of mechanical and thermal energy helped establish energy as a single quantity. Because the joule is algebraically identical to the newton-metre, which is also the unit of torque, many authorities discourage using N·m for energy to avoid confusion. Larger quantities are written with SI prefixes such as kilojoule, megajoule and gigajoule. The older centimetre-gram-second system used the erg, and one joule equals ergs.
Several other units remain in use, largely because they were defined in practical contexts before the joule was adopted. The calorie was originally tied to water: it was the heat needed to raise the temperature of one gram of water by one degree Celsius. Because the water reference temperature was not fixed, several definitions coexisted. The International Table calorie is 4.1868 J, while the thermochemical calorie is 4.184 J. In nutrition, the food "Calorie" is in fact a kilocalorie, equal to 4,184 J. A daily intake of about 2,000 food Calories therefore corresponds to roughly 8.4 MJ.
The kilowatt-hour (kWh) is the energy delivered by a power of one kilowatt over one hour, equal to 3.6 MJ or 3,600,000 J. It is the standard billing unit for electricity supplied to households and industry, which is why electricity bills rarely show joules. The electronvolt (eV) is a much smaller unit, defined as the kinetic energy gained by an electron crossing a potential difference of one volt, and it is equal to . Atomic, molecular and particle physics use it because typical bond energies and photon energies are tiny when expressed in joules. Chemists often quote reaction energies per mole rather than per particle, so the choice of unit depends on the scale of the process.
The British thermal unit (BTU) originates from the imperial system and was defined as the heat needed to raise the temperature of one pound of water by one degree Fahrenheit. The International Table BTU is about 1,055.06 J. In the United States it remains common in heating, ventilation and air-conditioning, where equipment is rated in BTU per hour, and national energy statistics sometimes use the quad, which is BTU. One kilowatt-hour corresponds to roughly 3,412 BTU, so converting between the two systems is a simple multiplication.
Energy and power are connected by time. Power is the rate at which energy is transferred or converted, so , and the watt is one joule per second. A heater rated at one kilowatt that runs for one hour transfers one kilowatt-hour, or 3.6 MJ. The name "kilowatt-hour" contains the word "power", but the unit measures energy, and this distinction is a frequent source of confusion. Conversion factors among the units above are fixed and exact in principle, although the calorie and BTU have several variants that differ by a fraction of a percent. The diversity of units reflects the history of separate traditions: calorimetry and chemistry, imperial engineering, the electricity trade and atomic physics each adopted units suited to their own measurements. The SI joule provides a common reference, and most scientific and engineering work now converts other units into joules as a standard step.
Energy resources and conversion technology
Primary energy sources
Primary energy is the energy contained in natural sources before any conversion. Fossil fuels still supply most of it. According to the Energy Institute's Statistical Review as summarised by Energy Factbook, fossil fuels provided 86.2% of world primary energy in 2025, with oil at 33.5%, coal at 27.7% and natural gas at 25.1%. Non-fossil figures depend on how a source is classified, and different compilations give different totals. A separate Statistical Review summary gives renewables only about 6% of primary energy, because nuclear, hydropower and traditional biomass are counted separately in that breakdown.
Electricity is a different picture. The IEA's Global Energy Review 2025 reports that renewables and nuclear together supplied two-fifths of global electricity in 2024 for the first time, with hydropower at 14%, wind at 8%, solar PV at 7% and nuclear at 9%. In 2025, renewables overtook coal as a share of world electricity, at roughly 34%. Hydropower, which converts the gravitational potential energy of falling water, is the largest renewable source of electricity, while wind and solar are growing fastest.
The main sources are as follows. Fossil fuels are chemical energy stored in hydrocarbons and released by combustion. Nuclear power releases energy through nuclear-fission of heavy nuclei such as uranium. Solar energy is radiation from the sun that is captured directly or indirectly, since wind is driven by uneven solar heating of the atmosphere. Geothermal energy draws on heat from the Earth's interior, and biomass draws on energy stored in plant material.
Secondary carriers and the electrical grid
Secondary carriers are produced from primary sources and are used to move energy to where it is needed. Electricity is the most important, although it accounts for only about a fifth of final energy use. Most of the remaining demand is for transport fuels, industrial heat and building heating, which remain far more dependent on fossil fuels than the power sector.
The electrical-grid delivers electricity from generators to consumers through high-voltage transmission lines and lower-voltage distribution networks. Because electricity is difficult to store in bulk, the grid must match supply and demand continuously, using frequency as a signal of imbalance. Hydrogen is another carrier. It can be made by electrolysis using surplus electricity, but producing it always costs more energy than it returns when it is later burned or used in a fuel cell.
Conversion devices and their efficiency
The steam-engine, developed by thomas-newcomen and improved by james-watt, was the first widely used device that turned heat into mechanical work. Modern electricity generation still relies on the same principle. A boiler or nuclear reactor produces high-temperature steam, which drives a steam turbine, and the turbine spins an electric-generator. The efficiency of this chain is limited by the temperatures involved. The Carnot bound on any heat engine is , where and are the absolute temperatures of the hot and cold reservoirs. A plant with steam at 800 K and cooling water at 300 K therefore cannot exceed about 62.5%. Real coal plants typically achieve roughly 35–40%, and combined-cycle gas plants can reach around 60%.
Other conversion devices have different efficiency profiles. Large generators and hydroelectric turbines often exceed 90% efficiency, because they convert mechanical energy directly into electricity without passing through heat. Photovoltaic cells convert sunlight directly into electricity. Commercial silicon modules typically convert around 20% of incident sunlight, and the theoretical limit for a single-junction cell is about 33%, set by the band gap and the spectrum of sunlight.
Energy storage and intermittency
Storage is needed because wind and solar output varies with weather and time of day. The main grid-storage technologies are pumped storage hydropower and batteries. Pumped storage moves water between two reservoirs at different elevations, storing energy as gravitational potential energy and releasing it through turbines when demand rises. Its share of total storage depends on the metric used. Estimates from the China Energy Storage Alliance, as reported by Stanford's energy site, put lithium-ion batteries at 56% of installed storage power capacity in 2025 and pumped hydro at 43%, while pumped hydro still holds most of the stored energy (GWh). Hydropower-sector estimates, by contrast, put pumped storage at over 94% of installed storage capacity, which reflects a different measure and an earlier period.
Batteries are well suited to fast response and short-duration balancing. Pumped hydro suits long-duration storage, and it provides services such as inertia, frequency control and black-start capability. Other options include thermal storage, such as heating molten salts or storing hot water, and compressed air. The main challenge of intermittency is that generation and demand do not coincide. Without storage, surplus renewable output must be curtailed, and additional firm capacity is needed for calm, cloudy periods. Storage therefore affects the cost and reliability of a system that depends heavily on variable sources.
Economics, policy, and the energy transition
Energy markets set the prices of fuels and electricity through supply and demand, but the mechanisms differ by carrier. Oil, gas, and coal are traded internationally or at regional hubs, so their prices respond to production decisions, inventories, weather, and geopolitical disruption. Electricity is hard to store and must be balanced continuously. In many wholesale markets, plants are dispatched in order of their marginal cost, and the most expensive unit needed to meet demand often sets the clearing price. Retail tariffs add network charges, taxes, and subsidies, which is one reason household bills move more slowly than wholesale prices. Market shocks can be abrupt: in 2022, primary energy intensity declined 2.1 percent, more than four times the weak improvement of 2021, as the energy crisis drove sharp falls in demand across many regions.
Energy intensity as a measure
Energy intensity expresses the energy an economy uses per unit of economic output. It is commonly written as , where is total energy use and is GDP, so a falling value means more output is produced for each unit of energy. Changes in intensity reflect efficiency improvements, shifts toward services, and the structure of industry. The global figure has improved steadily but unevenly. In 2025, global energy intensity improved by nearly 2%, in line with its long-term average from 2010 to 2019. That pace was faster than the recent trend of around 1.3% per year between 2019 and 2024. China's record is central to this: its improvement slowed sharply from nearly 4% per year between 2010 and 2019 to just 0.6% per year from 2019 to 2024, before rising again above 3% in 2025.
Progress is still judged against targets. Since 2015, global energy intensity has improved by 1.6 per cent per year on average, which is still short of the 3.2 per cent annual rate now needed to reach Sustainable Development Goal 7.3. The benchmark itself is contested in detail: the SDG 7 custodians kept the target fixed at 2.6 percent per year even though due to data revisions, the baseline annual improvement stands at 1.2 percent. The IEEJ outlook notes that the COP28 goal of doubling the annual efficiency improvement rate, from 2% to 4%, is unlikely to be met, with an expected pace of 2.1% annually from 2023 to 2030.
Carbon pricing and climate policy
Carbon pricing has become the main market-based tool of climate policy. According to the World Bank's 2026 survey, there are now 87 carbon pricing policies globally, increasing by seven since last year. These comprise 47 carbon taxes and 40 emissions trading systems, and just over 29% of global greenhouse gas emissions worldwide are now covered by direct carbon pricing. The average price is still low: the average carbon price is now nearly $21/tCO2e, and prices vary widely by region, from roughly US$68 in Europe and Central Asia to about US$11 in East Asia and the Pacific.[17-4] Revenue has grown with coverage; carbon pricing revenues have tripled over the past decade, reaching more than US$107 billion in 2025, and emissions trading systems account for more than 70 per cent of total revenues globally.
Attribution of falling emissions intensity is disputed. One assessment argues that this declining carbon intensity has not been caused primarily by climate policy, while also noting that climate policies have also played an important role, driving down emissions intensities in developed economies, particularly in Europe.
Access, finance, and the SDG 7 gap
The Sustainable Development Goals set a 2030 horizon for modern energy, and the 2026 progress report judges the world not happening fast enough for the SDG 7 targets to be achieved by 2030. Access remains the sharpest gap: in Sub-Saharan Africa, over 560 million still lack electricity, and 970 million have no access to clean cooking, and one estimate holds that the pace of electrification must triple to achieve universal access by 2030. Financing lags as well: international public financial flows to developing countries in support of clean energy were marginal, rising from USD 24.4 billion in 2023 to USD 24.6 billion in 2024.
Energy security and the shape of the transition
The energy mix changes slowly. In 2025, the energy mix remained dominated by oil (42%), followed by gas (23%) and nuclear (14%), while renewables contributed 5.7EJ (11%). Electricity is moving faster than the wider mix: 2025 marked the first year since the Covid-19 pandemic in 2020 without an increase in electricity generation from fossil fuels.
Energy security, the reliability and affordability of supply against shocks, drives much of policy. Import dependence, price volatility, and the resilience of networks all shape investment decisions. Historically, the energy efficiency measures enacted in the wake of the oil price shocks of the 1970s were an early example of security concerns reshaping energy use. Today the investment agenda centres on grid expansion, energy storage such as batteries and pumped hydro to manage the intermittency of solar and wind power, and the continuing decline in the cost of key technologies, which has made renewable generation competitive in many markets. Whether these changes proceed quickly enough depends on finance, the design of carbon and market instruments, and the political durability of commitments.
Historical development of the concept
The modern notion of energy was not invented in one step. It emerged in roughly the middle of the nineteenth century, when engineers trying to improve steam engines and physicists trying to explain heat began to treat heat, motion, and chemical and electrical effects as different manifestations of one measurable quantity.
Steam engines and the efficiency problem
By the early 1800s steam engines powered mines, mills, and transport, yet most of the fuel's heat was wasted. Engineers improved machines by trial and error, and there was no systematic science connecting heat to mechanical work. In 1824 the French military engineer Sadi Carnot published Reflections on the Motive Power of Fire, which asked whether an upper bound on the work obtainable from heat exists, and whether it depends on the working fluid or only on the engine's temperatures. He answered with an idealized reversible cycle operating between a hot and a cold reservoir, showing that its efficiency depends only on those temperatures. Carnot still worked within the caloric theory, which treated heat as an indestructible fluid, and his premise was later shown to be wrong. His efficiency limit, however, survived the change of theory and still bounds real heat engines.
Heat, work, and the conservation principle
The decisive step came in the 1840s. In 1842 the German physician Julius Robert Mayer argued from theory and from others' experimental data that a quantity is conserved while it converts between forms, and he estimated the mechanical equivalent of heat from the warming of gases. In 1843 the Danish engineer Ludvig August Colding presented a similar memoir in Copenhagen. Between 1839 and 1850 the English brewer James Prescott Joule carried out a long series of experiments, using falling weights, electrical currents, and compressed gases, to show that mechanical work and heat are quantitatively interchangeable. His measured value for the mechanical equivalent of heat was far more accurate than Mayer's.
In 1847 Hermann von Helmholtz presented the general principle in the memoir Über die Erhaltung der Kraft ("On the Conservation of Force"). His term Kraft corresponds to what is now called energy. Helmholtz treated conservation as an empirical generalization supported by experiment, not as a philosophical doctrine, and he showed how mechanical, thermal, electrical, and chemical processes could all be expressed in the same accounting. Early workers used a variety of names, such as "living force" and "fall-force", for quantities that were later unified under the single word energy. In 1854 Helmholtz publicly credited Mayer with priority for the principle.
Naming and reformulating the concept
The word itself had been used earlier in a quantitative sense by Thomas Young in the early 1800s, but it was not widely adopted for decades. In the 1850s William Thomson (later Lord Kelvin) and William John Macquorn Rankine reformulated thermodynamics in terms of energy. Rankine, in 1853, recast Helmholtz's conservation of force as a universal statement that the total energy of the universe is constant, and he distinguished "actual" energy in moving bodies from "potential" energy stored in configurations, a split that made the term potential energy standard. Thomson and Tait replaced "actual energy" with "kinetic energy" in 1862. Rankine defined energy as the capacity to perform work, a definition that quickly became the general standard. Rudolf Clausius, meanwhile, showed in 1850 how Carnot's result could be reconciled with Joule's experiments by treating part of the heat as converted to work and the rest as transferred to a colder body; this reconciliation led him to the concept of entropy. Thomson's work on irreversible processes added the idea that natural processes tend to degrade energy into heat.
Consolidation and the twentieth century
By the late nineteenth century energy had become a common language for mechanics, heat, electricity, magnetism, and chemistry. Maxwell's electromagnetic theory described field energy, Gibbs and Nernst applied energy and entropy to chemical equilibria, and Stefan's work on radiant energy linked thermodynamics to radiation. Energy thus functioned as an organizing principle across subfields rather than a property of one kind of phenomenon.
In the twentieth century the concept was extended further. In 1900 Max Planck proposed that energy is exchanged in discrete packets, a hypothesis that led to energy quantization and to the quantum theory of matter. In 1905 Albert Einstein published his special theory of relativity, which identified mass and energy as interchangeable through (mass–energy equivalence). The same year, his explanation of the photoelectric effect used light quanta, anticipating the concept of the photon. In 1913 Niels Bohr linked atomic spectra to discrete energy levels. In nuclear physics, measured mass defects revealed the energy stored in nuclear binding. The discovery of nuclear fission in 1938 and the subsequent explanation of stellar fusion made nuclear energy a practical and political issue. General relativity, published in 1915, later showed that the conservation of energy becomes subtler in curved spacetime.
Energy in computation and information
A theoretical floor on the energy cost of computing was proposed in 1961. The reasoning starts from the premise that a computer is a physical device, so its operations must obey thermodynamic law Rolf Landauer believed that computation is a physical process; thus, it must obey the laws of physics and, first and foremost, the laws of thermodynamics. The central result of Landauer's principle is that erasing one bit of information must dissipate at least as heat, where is the Boltzmann constant and is the absolute temperature the Landauer bound W=kBTln2, where T is the temperature of a thermal reservoir used in the process and kB is Boltzmann's constant. Because the bound scales with temperature, it is about 0.0175 eV per bit at room temperature Landauer limit of 0.0175 eV (2.805 zJ) per bit erased. A later generalization suggested that the entropy cost could in principle be carried by another conserved quantity, such as angular momentum, rather than by heat the principle was generalized to show that while information erasure requires an increase in entropy, this increase could theoretically occur at no energy cost. Instead, the cost can be taken in another conserved quantity, such as angular momentum.
Charles Bennett's formulation places the cost in a different location. Logically irreversible operations, such as merging two computation paths, require a matching rise in entropy somewhere in the hardware or its surroundings, so the entropy resides in the physical apparatus rather than in the logical bit the entropy thus resides in the physical hardware and its surroundings, not in the bit as a logical entity. This is why the principle is framed around irreversibility. Operating below the bound would therefore require computation that avoids irreversible steps, which is the aim of reversible computing It is hypothesized that energy consumption below this lower bound would require the development of reversible computing.
Experiments have begun to approach the bound. At 300 K, flipping a bit took about 0.026 eV, only about 44% above the Landauer minimum Flipping the bit required about 0.026 eV (4.2×10−21 J) at 300 K, which is just 44% above the Landauer minimum.. A 2018 experiment performed erasure at 1 K on an array of molecular magnets, with each nanomagnet encoding one bit A 2018 article published in Nature Physics features a Landauer erasure performed at cryogenic temperatures (T = 1 K) on an array of high-spin (S = 10) quantum molecular magnets. The array is made to act as a spin register where each nanomagnet encodes a single bit of information.. Commercial hardware sits far above this floor: For the present generation of silicon-based integrated circuits, the energy dissipation per logic operation is about a factor of 1000 larger than the Landauer limit. The gap is mostly attributed to other losses in switching and transmission rather than to erasure itself, so the bound is not the practical constraint on today's chips.
Physical hardware dominates the actual energy budget of AI. A single H100 accelerator has a 700 W thermal design power, and an eight-GPU server can draw between 5.6 and 10.2 kW An H100 GPU has a TDP of 700W, and an 8-GPU server draws 5.6-10.2 kW depending on configuration.. In one node measurement, GPUs accounted for roughly 56% of total node power The GPUs account for roughly 56% of total node power.. Denser racks have raised the stakes: moving from eight to 72 GPUs per rack pushes requirements toward 150 kW Racks have gone from each having eight GPUs to 72 starting two years ago, requiring around 150kW of power.. Cooling is a major share of facility electricity, reported as high as 40% The main source of electricity consumption (as high as 40%) at data centers is the high cooling needs during operation. At the facility scale, In 2024, data centers consumed approximately 415 TWh of electricity, representing approximately 1.5% of total global demand. By 2030, this figure is projected to increase to nearly 945 TWh. Operators can trade throughput for energy: one MIT team limited processors to 150 or 250 W, roughly 60% to 80% of their rated power "Rather than letting them go to 100%, we limit usage to 150 or 250 watts [about 60% to 80% of their total power] depending on which processor we're using,".
Training and inference are the two main energy costs of large language models. One estimate puts GPT-3's training at about 1,287 MWh GPT-3 consumed approximately 1287 megawatt hours (MWh), while another large language model, Gopher, consumed approximately 1066 MWh. In contrast, the average annual energy consumption of a U.S. household in 2022 was estimated to be around 11 MWh.. Per-query figures are less settled. One estimate gives about 3 Wh per GPT-3 query, while a later analysis suggests about 0.3 Wh per GPT-4 query Alex de Vries' widely cited paper estimates about 3 Wh per GPT-3 query. However, a more recent paper by epoch.ai suggests that GPT-4, OpenAI's flagship model, consumes about 0.3 Wh per query, ten times less., while a news site gives a much higher value of 0.3 kWh, so sources disagree by orders of magnitude. The variation depends on chip efficiency, the share of parameters activated per query, and assumed query length It takes into account the improved efficiency of modern chips, the fact that ChatGPT probably activates only a fraction of its parameters for each query, and the assumption about the typical query length.. Measurement methods differ in scope: some read hardware counters, such as metrics obtained from NVML and RAPL to calculate the energy consumption of the CPU and the GPU, while others estimate facility-level use across development, training, and inference we calculate (i) electricity consumption, (ii) carbon emissions, and (iii) water consumption at three points in the machine learning pipeline: early model development, training of the main model, and inference.
The link between information and thermodynamics rests on a shared mathematical form. Claude Shannon's information entropy measures uncertainty, and in physics the term "information" in this context refers to uncertainty rather than certainty the technical term "information" here refers to uncertainty rather than certainty.. Landauer's argument applies Boltzmann's relation to a bit with two states, which gives an entropy increase of per erased bit The equation can be deduced from the Boltzmann's entropy formula S=k_B ln W, considering that W is the number of states of the system, which in case of a bit is 2. In this sense, a Shannon entropy measured in bits corresponds to thermodynamic entropy multiplied by . The correspondence is conceptual rather than an identity: information entropy describes a probability distribution over messages, while thermodynamic entropy describes physical microstates. The principle itself remains debated; one review notes that The Landauer principle remains highly debatable.
Philosophical and popular meanings
Aristotle's energeia and metaphysical debates
The Greek energeia (ἐνέργεια) combines en (in) and ergon (work), and Aristotle set it against dynamis, potentiality. In his Physics and Metaphysics, a block of marble has the potential to become a statue, and the act of shaping it realizes the form. This was a qualitative notion about actuality. It had no numerical measure and no conservation law. The quantitative concept that became physics emerged later, in the late seventeenth-century dispute over vis viva. Leibniz's measure of living force, proportional to , was set against the momentum of Descartes.
Debates about what energy is continued into the modern era. Around 1900, Wilhelm Ostwald's "energetics" proposed that energy was the fundamental substance from which matter and all other phenomena derived. Atomistic physicists such as Ludwig Boltzmann rejected this program. Contemporary philosophers of physics mostly treat energy as a magnitude defined by its role in dynamical laws, not as a material stuff. Energy depends on conventions such as the choice of zero for potential energy, and it is defined only in certain spacetimes, so many argue it is better understood as a relational or structural property than as a thing. The debate remains open within philosophy of science and metaphysics.
Everyday meanings
In ordinary English, "energy" refers to vitality, enthusiasm, or drive ("she has a lot of energy"), to the atmosphere of a place ("a tense energy in the room"), and to personal qualities ("an energetic approach"). These uses are qualitative and unmeasured. Someone can be "drained" or "recharged" without any quantity being transferred in the physical sense. The physical concept is precise. It is a quantity with a unit, the joule, defined through work and heat, and in an isolated system it is conserved. The two senses overlap only metaphorically, although they meet in nutrition, where food labels report the energy a body can extract from food in kilojoules or kilocalories. That is a physical measurement, but it says nothing about how alert or cheerful a person feels.
Common misconceptions
The most persistent misconception is that energy is "used up" or "consumed." Because energy is conserved, a lamp does not consume it. It converts electrical energy into light and thermal energy, and the total stays the same. What is lost is usefulness. Energy spreads into the surroundings as disordered thermal motion, where it can no longer be easily turned into work. This is the content of the second law of thermodynamics and of entropy. Phrases such as "energy crisis" or "saving energy" therefore usually refer to available energy or to resources, not to a shrinking total. A second misconception treats energy as a fluid stored in objects, as in "a battery holds electricity." Energy is better understood as a bookkeeping quantity that changes through work and heat, and "flow" is a convenient metaphor rather than a literal description. Historically, a similar confusion appeared in the caloric theory of heat, which treated heat as a substance; the energy-based account replaced it in the nineteenth century.
Pseudoscientific claims
Pseudoscientific claims about energy include over-unity machines that supposedly output more energy than they take in, perpetual motion devices, "free energy" generators, and healing or "life energy" fields. Some also invoke zero-point energy as an inexhaustible power source. Physics distinguishes these claims from established theory on several grounds. A device producing net work over a closed cycle without an external input contradicts the first law of thermodynamics. Zero-point energy is the lowest energy of a quantum system, so under standard quantum mechanics it cannot be drained continuously for net work. The term "free energy" also has a precise thermodynamic meaning, and its vague use in popular claims often hides this. More broadly, established physics makes quantitative predictions that can be tested, reproduced, and falsified, in line with the criterion of falsifiability associated with Karl Popper. Pseudoscientific energy claims often lack these features and resist disconfirmation. A historical parallel is vitalism, the idea that living matter requires a special force. It lost ground as organic compounds were synthesized and metabolism was explained in chemical and thermodynamic terms. Claims of energy healing or "vibrational" fields have not been supported in controlled tests, and the scientific consensus holds that they have no basis in physics.
Open questions and frontiers
Dark energy and the cosmological constant problem
Observations of distant Type Ia supernovae in 1998 showed that the expansion of the universe is accelerating. The simplest explanation is a component with negative pressure whose energy density stays roughly constant as space expands. In the standard CDM model this component is dark energy, usually identified with the cosmological constant . Recent cosmic microwave background analyses put it at about 68% of the universe's total energy content, compared with about 27% for dark matter and about 5% for ordinary matter. Its physical origin is unknown.
The central puzzle is the vacuum energy. Quantum field theory assigns a nonzero energy to the vacuum from the fluctuations of every field, which are described through vacuum expectation values and virtual particles. Simple cutoff estimates of this energy exceed the observed value by roughly 120 orders of magnitude. This mismatch is the cosmological constant problem. Proposed responses include unknown cancellation mechanisms, dynamical fields (such as quintessence) whose energy density changes over time, and modifications of gravity itself. Measurements of the equation-of-state parameter are a key test. So far it is consistent with , the value expected for a true cosmological constant, within measurement uncertainty. A statistically significant departure would favor dynamical models.
Energy in general relativity and quantum gravity
In special-relativity, energy conservation follows from time-translation symmetry in a fixed flat spacetime. General relativity lacks this simplicity. The Einstein field equations define a covariant energy-momentum tensor for matter and fields, but the gravitational field's own energy has no covariant local density. Attempts to assign one, through so-called pseudotensors, depend on the choice of coordinates. Total energy can be defined only in special settings, such as asymptotically flat spacetimes, where it is given by the ADM mass. In an expanding cosmology the time-translation symmetry is broken, and the energy of a redshifting photon is not conserved in the naive sense, because its wavelength stretches with the universe.
Quantum gravity aims to combine general relativity with quantum mechanics, and it faces open questions about energy. Its natural energy scale is the Planck energy, about GeV, or roughly 2 GJ. That is about a quadrillion times the energy of a single particle at the Large Hadron Collider. Physicists do not yet know whether energy remains a fundamental, locally defined quantity at this scale, or whether it emerges from more basic structures. Black hole thermodynamics, with its Hawking temperature and Bekenstein–Hawking entropy, is one of the few areas where these ideas have been tested in theory.
Energy in the early universe
Cosmic inflation is a proposed period of accelerated expansion in the first fraction of a second after the origin of the universe. It is generally thought to be driven by the potential energy of a scalar field, the inflaton. During inflation this potential energy is nearly constant and dominates the total energy density. When inflation ends, the field decays, and its energy is converted into particles that thermalize. This reheating process is taken to start the hot phase of the Big Bang.
The energy scale of inflation can be estimated from the amplitude of primordial gravitational waves, which would imprint a faint pattern on the polarization of the cosmic microwave background. Current bounds on this signal, from BICEP/Keck analyses, imply an energy scale on the order of GeV. That is well below the Planck energy, but far above anything reachable in laboratories. Since inflation is the proposed link between quantum fluctuations and the large-scale structure of the universe, its energy scale is a central target for future observations. The total energy of the universe, including gravitational contributions, remains a subtle quantity that is not easily defined in cosmology.
Energy requirements and limits of future technologies
Two open questions concern what physics permits and what society will require. On the energy side, global primary energy use is on the order of 600 exajoules per year. The solar flux intercepted by Earth is about W, so the resource is large, but converting it to usable form is limited by photovoltaic efficiency, the thermodynamic bounds of heat engines described by the Carnot cycle, and the cost of energy storage to handle intermittency. Designing a sustainable system therefore depends on engineering, materials, and grid integration, not only on the total amount of energy available. The Earth's energy budget, driven by energy from the Sun, sets the outer limit.
On the computational side, Landauer's principle gives a fundamental minimum. Erasing one bit of information at temperature dissipates at least , where is the Boltzmann constant. At 300 K this is about J per bit. Present-day computers, including the graphics processing units used to train artificial intelligence systems, dissipate many orders of magnitude more energy per operation, so the bound is not an immediate constraint. As energy use for large models grows, however, the gap between practical and theoretical limits becomes a question of engineering and of how information is processed. Reversible computation, which avoids erasing information, is one proposed route, though it has not yet been realized at scale.
These frontiers connect cosmology, quantum gravity, and applied energy technology through the same question: how energy is defined, conserved, and limited under different conditions. Many answers remain unresolved.
References
- Energylink.springer.com
- Energyen.wikipedia.org
- History of energyen.wikipedia.org
- The history of the concept of energy and workspark.iop.org
- What does energy really mean?physicsworld.com
- 0% found this document useful (0 votes)scribd.com
- Why Work Equals the Change in Kinetic Energy — The Work-Energy Theoremphysicsfundamentals.org
- Conservative Forces and Potential Energyopenbooks.lib.msu.edu
- Conservative Forces and Potential Energycourses.lumenlearning.com
- 3.4: Energy Accounting with Conservative Forces: Potential Energy - Physics LibreTextsphys.libretexts.org
- Physics Study Guide: Potential Energy & Conservationpearson.com
- [FREE] Why is friction considered a dissipative force? - brainly.combrainly.com
- What is meant by dissipation of energy?tec-science.com
- Kinetic Energy and the Work-Energy Theoremcourses.lumenlearning.com
- Time-translation symmetryen.wikipedia.org
- Noether's Theorem in a Nutshellmath.ucr.edu
- How Noether’s Theorem Revolutionized Physicsquantamagazine.org
- Noether’s Theorem and Newton’s Lawspubs.aip.org
- Noether’s Theorem: A Complete Guide With Examplesprofoundphysics.com
- Energy and time evolution Conserved quantities in ...phas.ubc.ca
- Copyright c⃝2020 by Robert G. Littlejohn Physics 221A Fall 2020 Notes 5bohr.physics.berkeley.edu
- Investigating Chaos by the Generalized Alignment Index (GALI) Methodarxiv.org
- Energy evolution in time-dependent harmonic oscillatorarxiv.org
- Time dependent Hamiltonians: features • Physics Forumsphysicsforums.com
- Time-Dependent Perturbation Theorygalileo.phys.virginia.edu
- Hamiltonian analogs of combustion engines: a systematic exception to adiabatic decouplingarxiv.org
- Invariants for time-dependent Hamiltonian systemsarxiv.org
- Entropyen.wikipedia.org
- Clausius and the Second Law of Thermodynamicsebsco.com
- Lectures on Probability, Entropy, and Statistical Physicsarxiv.org
- A History of Thermodynamics: The Missing Manualmdpi.com
- Bluff your way in the second law of thermodynamicsarxiv.org
- On the (Boltzmann) Entropy of Nonequilibrium Systemsarxiv.org
- Entropy and Frenesy: the arrow and the bustle of timearxiv.org
- Carnot Efficiency - an overviewsciencedirect.com
- Carnot Cyclestudy.com
- Van der Waal's gas equation for an adiabatic process and its Carnot engine efficiencyarxiv.org
- Exergy/Availability - an overviewsciencedirect.com
- Exergy - an overviewsciencedirect.com
- Internal Exergy Loss - an overviewsciencedirect.com
- Reversible work and irreversibilityfiveable.me
- Understanding Availability and Exergy in Thermodynamics - Engineering Cheat Sheetengineeringcheatsheet.com
- September, 1905: Energy and Mass are Equivalentaps.org
- The Equivalence of Mass and Energy (Stanford Encyclopedia of Philosophy)plato.stanford.edu
- Did Einstein prove E=mc2? - ScienceDirectsciencedirect.com
- 20.5 Energy of Nuclear Reactions and Nuclear Binding Energy - Chad's Prep®chadsprep.com
- Iron-56en.wikipedia.org
- What Is Nuclear Binding Energy? Mass Defect and the Stability Curvephysicsfundamentals.org
- Nuclear fissionebsco.com
- Physics of Uranium and Nuclear Energy - World Nuclear Associationworld-nuclear.org
- Uranium-235 (U-235)britannica.com
- Nuclear chain reactionen.wikipedia.org
- Nuclear fissionen.wikipedia.org
- Uranium-235en.wikipedia.org
- Uranium 235 Fissionnuclear-power.com
- Describe nuclear fusion in stars (proton–proton chainvarsitytutors.com
- Most important nuclear reaction - proton-proton chainnuclear-power.com
- Proton-Proton Chain (Astronomy) - Overviewstudyguides.com
- graphsearch.epfl.chgraphsearch.epfl.ch
- Cellular respirationen.wikipedia.org
- ATP Yield From Cellular Respirationjove.com
- 9.1.4: Oxidative Phosphorylation - Biology LibreTextsbio.libretexts.org
- 7.6: ATP as Energy carrier - Chemistry LibreTextschem.libretexts.org
- Adenosine triphosphateen.wikipedia.org
- Coupled Reactions: AP® Chemistry Reviewalbert.io
- Earth's energy budgeten.wikipedia.org
- Outgoing longwave radiationen.wikipedia.org
- Climate and Earth’s Energy Budget - NASA Sciencescience.nasa.gov
- Climate Science Investigations South Florida - Energy: The Driver of Climateces.fau.edu
- Earth’s Radiation Budget: Balancing Incoming and Outgoing Energy • Environmental Studies (EVS) Instituteevs.institute
- Earth's energy imbalance has doubled—here's why that mattersphys.org
- Halfway to doubling of CO2 radiative forcing Article Accepted Versioncentaur.reading.ac.uk
- Radiative Forcing of Climate Change Co-ordinating Lead Author V. Ramaswamyipcc.ch
- Radiative forcingen.wikipedia.org
- Ocean Warming - Earth Indicator - NASA Sciencescience.nasa.gov
- Heat Storagemarine.copernicus.eu
- Climate Change: Ocean Heat Contentclimate.gov
- Climate change: ocean heat contentclimate.us
- Overheating a Water Planet: Warmed Oceans Will Not Be Ignored - Union of Concerned Scientistsblog.ucs.org
- Ocean Heat Content Risesncei.noaa.gov
- Joulebritannica.com
- Jouleen.wikipedia.org
- British thermal uniten.wikipedia.org
- Energy Converter — Joules, kWh, BTU, calories, eV & morecalculatordashboard.com
- Energy Converterovacen.com
- Energy Converterilovewebtools.online
- Joule (J): Definition, Formula, Units, Conversions & Examplesenergymanageredu.com
- Global electricity mix 2025, by energy sourcestatista.com
- Global Energy Review 2025iea.blob.core.windows.net
- 6 facts about how global energy production is changingpewresearch.org
- What Percentage of the World's Energy Comes From Fossil Fuels? (2025)energyfactbook.com
- What Percentage of the World's Electricity Comes From Fossil Fuels? (2025)energyfactbook.com
- Renewables Supplied Just 6% of Global Primary Energy in 2025clintel.org
- Energy Storageunderstand-energy.stanford.edu
- The world’s water battery: Pumped Storage Hydropower and the clean energy transitionhydropower.org
- Pumped storage hydropower explained: how it works and why it mattersPumped storage hydropowerhydropower.org
- Pumped Hydro Storage Market Size, Share, Growth, Analysis, 2034straitsresearch.com
- For Storing Electricity, Utilities Are Turning to Pumped Hydroe360.yale.edu
- Photo: © Prathan Chorruangsak / Moment / Getty Images CHAPTER 4 ENERGYtrackingsdg7.esmap.org
- Global trendsiea.org
- Goal 7sdgs.un.org
- 2. Energy demand IEEJ Outlook 2026 1 2. Energy demandeneken.ieej.or.jp
- State and Trends of Carbon Pricing 2026worldbank.org
- State And Trends Of Carbon Pricing 2026 Shows Carbon Pricing Expansion, Higher Revenues, And Rising Demand For Quality Carbon Creditsccarbon.info
- Global carbon pricing revenues top US$107bn as ETS markets strengthen - H2 Bulletinh2bulletin.com
- Global Energy Outlook 2026: How the World Lost the Goal of 1.5°Crff.org
- 2026 Energy Progress Report Finds Renewables Supply Over 30% of Electricitysdg.iisd.org
- 2026 Energy Progress Report: SDG 7 Progress Falls Short of 2030 Goal - News and Statistics - IndexBoxindexbox.io
- Energy Institute Statistical Review of World Energy 2026: Trends, Security, and Transitionslideshare.net
- Global Electricity Review 2026ember-energy.org
- Energy intensity of GDPyearbook.enerdata.net
- World Bank State and trends of carbon pricing 2026 - Vertisvertis.com
- Tracking SDG7: The Energy Progress Reportseforall.org
- Hermann von Helmholtz (Stanford Encyclopedia of Philosophy)plato.stanford.edu
- The History of the Word "Energy"home.uni-leipzig.de
- The Source of Solar Energy, ca. 1840-1910: From Meteoric Hypothesis to Radioactive Speculationsarxiv.org
- Mayer, Julius Robert (1814–1878)encyclopedia.com
- The law of conservation of energyspark.iop.org
- Conservation of Energy: Readings on the Origins of the First Law of Thermodynamics. Part IIscielo.org.mx
- Sadi Carnot Publishes Treatise on Heat Enginesaps.org
- The Carnot Engine and Thermodynamics - Kronecker Walliskroneckerwallis.com
- Exploring the Science and History of Thermodynamics - NYASnyas.org
- Physics: Thermodynamicsencyclopedia.com
- A History of Thermodynamics: The Missing Manual - PMCpmc.ncbi.nlm.nih.gov
- W. J. M. Rankineen.wikipedia.org
- Carnot Cycle - Chemistry LibreTextschem.libretexts.org
- Landauer’s Principle: Past, Present and Futurencbi.nlm.nih.gov
- Generalization of the Landauer Principle for Computing Devices Based on Many-Valued Logicncbi.nlm.nih.gov
- information and energypermacomputing.net
- Landauer's principleen.wikipedia.org
- Landauer%27s principleen.wikipedia.org
- Landauer's erasure principle in a squeezed thermal memoryarxiv.org
- AI Data Center Power Consumption: Managing the GPU Energy Crisisintegrar-iot.com
- Power-Bound, Not GPU-Bound: AI Data Center Power Constraints Are the Real 2026 Bottleneckspheron.network
- The Race to Rethink Data Centers for AI’s Power Surgebloomberg.com
- Nuclear fusion for AI: A pathway to power data centers sustainablyarxiv.org
- The Environmental Impact of AI Servers and Sustainable Solutionsarxiv.org
- AI has high data center energy costs — but there are solutionsmitsloan.mit.edu
- Energy Considerations for Large Pretrained Neural Networksarxiv.org
- AI's Environmental Impact: Making an Informed Choicemarmelab.com
- Measuring and Improving the Energy Efficiency of Large ...oa.upm.es
- Holistically Evaluating the Environmental Impact of Creating Language Modelsarxiv.org
- Quantum Thermodynamicsarxiv.org
- The Landauer Principle: Re-Formulation of the Second Thermodynamics Law or a Step to Great Unification?ncbi.nlm.nih.gov
- A systematic review of electricity demand for large language models: evaluations, challenges, and solutions - ScienceDirectsciencedirect.com
- Expert Assessment: The Systemic Environmental Risks of Artficial Intelligencearxiv.org
- Large Language Models Carry Enormous Energy Consumption and Costthegrizzlynews.org
- The Energy Footprint of Humans and Large Language Modelscacm.acm.org