What Is the Kardashev Scale — And Could Humanity Ever Reach Type III?

The Kardashev Scale ranks a civilization by how much power it controls. Nikolai Kardashev, a Soviet astronomer, proposed it in 1964 as a way to guess what kind of radio signals an extraterrestrial society might send. Type I commands the energy of a planet, Type II the output of a star, and Type III the output of a galaxy. By the standard formula, humanity currently scores about 0.73.

That small decimal hides a great deal of arithmetic. The gaps between the types are not steps but cliffs, each one roughly ten billion to one hundred billion times taller than the last. This article walks through what each type means, how the 0.73 is calculated, what the physics says about the engineering, and why no one has found a Type III civilization.

Key takeaways

• Type I uses the energy available to a planet. Kardashev's own 1964 figure was about 4 × 10^12 watts, a level close to what humanity already used then. Carl Sagan's later convention puts it at 10^16 watts.
• Type II uses a star's full output. For the Sun that is about 3.8 × 10^26 watts, roughly 20 trillion times current human power use.
• Type III uses a galaxy's energy, about 4 × 10^37 watts in Kardashev's definition.
• Humanity scores about 0.73 on Sagan's formula, based on roughly 19 to 20 terawatts of primary energy use in 2023.
• A 2015 survey of about 100,000 galaxies found none that looked heavily populated by a civilization using most of its galaxy's starlight.

Where the scale came from

Kardashev, who was born in Moscow in 1932 and died there in 2019, presented a paper titled "Transmission of Information by Extraterrestrial Civilizations" at a 1964 conference in Soviet Armenia, according to Wikipedia's biography of Kardashev. The paper was about radio astronomy, not futurism. If a civilization wanted to be heard across interstellar distances, Kardashev asked, how much power could it afford to transmit, and how big a signal would we need to be able to detect?

To answer that, he needed a way to sort possible civilizations by their energy budgets. He chose three round benchmarks. As summarized in the Wikipedia article on the scale, the original definitions were:

• Type I: a technological level close to what Earth had reached at the time, about 4 × 10^19 erg/s (4 × 10^12 W).
• Type II: a civilization that harnesses the energy of its star, about 4 × 10^33 erg/s (4 × 10^26 W).
• Type III: a civilization that harnesses the energy of its galaxy, about 4 × 10^44 erg/s (4 × 10^37 W).

An erg is a tiny unit of energy from the older centimeter-gram-second system, and one erg per second equals 10^-7 watts. The scale is logarithmic, which means each type is not a bit more than the last but orders of magnitude more. A useful comparison is the Richter scale for earthquakes: a magnitude 7 is not slightly bigger than a 6, it releases about 32 times as much energy.

Kardashev's original Type I was pegged to the power humanity already used in the early 1960s. By his own definition, we passed Type I decades ago. The popular claim that we are "not yet Type I" depends entirely on a later redefinition.

How Sagan turned three boxes into a ruler

Kardashev gave three discrete categories. Carl Sagan, writing in 1973, supplied an interpolation that lets any civilization be scored with a decimal. The Kardashev scale entry on Wikipedia gives the formula as K = (log₁₀ W − 6) / 10, where W is power in watts.

The formula makes K = 1 equal to 10^16 W, K = 2 equal to 10^26 W, and K = 3 equal to 10^36 W. Notice that these anchors are not identical to Kardashev's. Sagan's Type I is a factor of 2,500 above Kardashev's 4 × 10^12 W, and his Type III is a factor of about 40 below Kardashev's 4 × 10^37 W. The ten-fold spacing between types in the formula is a convenience, not a statement about nature.

That explains an inconsistency you will find in popular accounts. Earth intercepts roughly 1.7 × 10^17 W of sunlight (a calculation covered below), yet Sagan's Type I sits at 10^16 W. The Sagan benchmark is not "all the sunlight hitting Earth." It is a round number, about 6% of the incoming sunlight.

The 0.73 calculation, step by step

Here is how the headline number comes out. The Energy Institute's Statistical Review of World Energy reported that world primary energy consumption reached a record 620 exajoules in 2023, according to DieselNet's summary of the 2024 edition. An exajoule is 10^18 joules.

1. Convert the annual total to joules: 620 × 10^18 J.
2. Divide by the number of seconds in a year (about 3.156 × 10^7) to get an average power: roughly 1.96 × 10^13 W, or about 19.6 terawatts.
3. Take the base-10 logarithm: log₁₀(1.96 × 10^13) is about 13.29.
4. Apply Sagan's formula: (13.29 − 6) / 10 = 0.729.

The Wikipedia entry reports the same result, 0.73, using a 2021 figure of 595 exajoules, which works out to about 18.9 terawatts. The score barely moves from year to year because the formula takes a logarithm. Even a 10% jump in global energy use changes K by only about 0.004.

The caveat is that "primary energy" counts the heat content of fuel burned, not the useful work delivered. A typical combustion engine turns much of its fuel energy into waste heat, so the usable power humanity actually applies is smaller. For ranking purposes the convention is to use primary energy anyway.

Type I: mastering a planet

How much energy is "all of Earth's"? The simplest estimate uses sunlight. The International Astronomical Union's 2015 nominal value (Resolution B3) for solar irradiance at Earth's distance is 1,361 watts per square meter. Earth presents a disk of about 1.27 × 10^14 square meters to the Sun. Multiply, and you get about 1.7 × 10^17 W, or 170 petawatts (a petawatt is 10^15 W).

Human power use of 19.6 TW is about 0.01% of that. In round numbers, the sunlight striking Earth carries nearly 9,000 times more power than the entire human economy burns. Type I, in this framing, would mean capturing and using a large share of that stream, along with geothermal, tidal, and nuclear sources.

The arithmetic of growth gives a rough timeline. Reaching Sagan's 10^16 W from today's 1.96 × 10^13 W requires a factor of about 510. If energy use compounded at 2% per year, which is close to the 2% growth the Energy Institute reported for 2023, the factor of 510 takes about 315 years. At 3% per year it takes about 210 years. This is a back-of-envelope extrapolation, not a forecast. Michio Kaku and others have popularized estimates of 100 to 200 years, and those depend on assuming faster growth than recent decades have delivered.

Three obstacles matter more than the growth rate:

• Waste heat. Every watt used eventually becomes heat. A civilization drawing a tenth of the sunlight Earth receives would be adding a heat load that is a visible fraction of the planet's natural energy balance, which makes climate a hard constraint on energy growth. The scale quietly assumes that heat can be managed.
• Capture and storage. Sunlight arrives unevenly across day, night, season, and weather. Using it at scale means grids, storage, and transmission that do not exist yet. Our articles on how solar panels work and why grid storage is hard describe where the engineering stands.
• Coordination. Energy infrastructure at planetary scale requires agreements among many governments. The technical problems are real, but the political ones may be harder.

Claims that a Type I civilization would "control the weather" or "suppress earthquakes" are speculation layered on the scale. Kardashev defined Type I by power, not by capabilities, and no one has shown that a given power budget confers control over hurricanes or faults.

Type II: mastering a star

The Sun's nominal luminosity, defined by the International Astronomical Union in 2015, is 3.828 × 10^26 W. Dividing by today's human power use gives about 2 × 10^13. In plain terms, the Sun puts out roughly 20 trillion times as much power as all human activity consumes. Earth intercepts only about 4.5 × 10^-10 of the Sun's output, or less than one part in two billion. Almost all of the Sun's light travels into empty space. A Type II civilization is one that stops letting it.

The Dyson sphere, and why Dyson did not mean a solid shell

The classic Type II megastructure is named for the physicist Freeman Dyson, who published "Search for Artificial Stellar Sources of Infrared Radiation" in Science in 1960, according to Wikipedia's article on the Dyson sphere. Dyson's idea was about detection: a civilization that captured most of its star's light would have to reradiate the energy as waste heat, mostly in the infrared, and astronomers could look for stars glowing in the wrong color.

Dyson was also explicit that a rigid shell would not work. He wrote that "a solid shell or ring surrounding a star is mechanically impossible," and described instead a swarm of objects on independent orbits. Science fiction popularized the solid sphere. Engineering studies since then have focused on the swarm.

A swarm is like a flock of starlings rather than a hollow ball. Each collector is a thin, light mirror or solar panel in its own orbit. No single piece has to hold up the others, so the structural problem that kills the solid shell disappears. The new problems are scale and control: how to build, launch, and keep track of an enormous number of independent objects, and how to prevent collisions. Our explainers on Dyson spheres and on what would happen if we built one around the Sun cover the options in more depth.

Where would the material come from?

The usual answer is that a civilization would take apart rocky bodies in its system and refashion them into collectors, a process often illustrated with Mercury because it is the closest large rocky body to the Sun. The details vary widely depending on how thin and light the collectors are. There is no standard, agreed-upon figure for the mass needed, and published estimates depend on assumptions about collector efficiency and how much starlight the swarm must capture. For that reason this article does not quote a single number.

Other Type II designs

The literature includes several variants beyond the swarm:

• Dyson bubble: very thin reflective sails held in place by radiation pressure rather than orbital motion.
• Statite arrays: satellites that hover by balancing gravity against light pressure.
• Stellar engines: structures that use a star's own output to move it, which some authors propose as a way to relocate a star system.

All remain paper studies. No one has built anything remotely comparable.

Type III: mastering a galaxy

Kardashev defined Type III as a civilization that harnesses its galaxy's energy output, about 4 × 10^37 W. The Wikipedia entry notes that Lemarchand defines this as the luminosity of the entire Milky Way. Wikipedia's Milky Way article gives an estimated total stellar mass of roughly 4.6 to 6.4 × 10^10 times the mass of the Sun and a star count of 100 to 400 billion, which is on the order of 10^11.

The jump from Type II to Type III is 4 × 10^37 divided by 4 × 10^26, or a factor of 10^11. That is one hundred billion Type II civilizations' worth of power. A Type III civilization would not be a single society that grew larger. It would be a network of star systems that have somehow stayed connected across tens of thousands of light-years.

Expansion time

The main constraint is travel time, not energy. The Milky Way's disk is roughly 90,000 to 100,000 light-years across (Wikipedia gives about 87,400 light-years for the optical disk, with the true extent poorly known). Even at a tenth of the speed of light, crossing it takes close to a million years, and spreading to most stars takes longer because probes must stop, build copies of themselves, and launch again.

Stuart Armstrong and Anders Sandberg's 2013 paper in Acta Astronautica, "Eternity in six hours", argues that spreading across even intergalactic distances may require only modest energy and resources for a civilization that already commands a star, using automated replicating probes. Their paper is an analysis of feasibility, not an engineering plan, and it explicitly serves to sharpen the Fermi paradox rather than to predict what anyone will build.

What a galaxy-scale civilization could draw on

Beyond starlight, a Type III civilization could tap two other major sources:

• Fusion. Hydrogen fusion converts about 0.7% of the fuel's mass into energy. That is why stars last billions of years on a finite supply.
• Black holes. Wikipedia's article on accretion disks says accretion can convert about 10% to over 40% of an object's mass into energy, compared with around 0.7% for nuclear fusion, with the higher values for rapidly spinning holes. That makes accretion roughly fifteen to sixty times more efficient than fusion.

The second item explains why astronomers consider quasars, which are supermassive black holes accreting gas, to be the most luminous persistent objects in the universe. A Type III civilization would plausibly exploit black holes directly, though how is entirely speculative. The Penrose process, in which energy is extracted from the rotation of a spinning black hole, is a real piece of general relativity but has never been demonstrated, and applying it at scale is a thought experiment. For the physics, see our articles on black holes and quasars.

Looking for Type II and Type III civilizations

Because a Dyson swarm would convert starlight into infrared heat, astronomers can search for it. The same logic applies at galaxy scale: a galaxy whose starlight was mostly captured would look dim in visible light and bright in the mid-infrared.

Galaxy-scale searches

The Glimpsing Heat from Alien Technologies (G-HAT) team searched NASA WISE infrared data for about 100,000 nearby galaxies. A Penn State news release from April 15, 2015 reports that no galaxy appeared to be largely colonized by an advanced civilization. About 50 galaxies had unusually high mid-infrared emission, which the team said required follow-up to determine whether natural processes or technology produced it. Lead researcher Jason Wright was quoted: "Either they don't exist, or they don't yet use enough energy for us to recognize them."

The result is narrow. It excludes galaxies mostly converted into heat-radiating technology in a given infrared window. It says nothing about a Type III civilization that radiates its waste heat differently or that uses a small fraction of its galaxy's light.

Star-scale searches

A 2024 paper, Project Hephaistos II, searched about five million objects from the Gaia, 2MASS, and WISE catalogs for stars with infrared excess that could indicate a partial Dyson sphere. The authors identified seven candidates, all M-dwarf stars. They stopped short of claiming discoveries: the abstract describes the candidates as deserving further analysis and says their infrared excess is hard to explain with known astrophysical phenomena, a weaker statement than ruling out natural explanations. Dust-obscured background galaxies are one of the mundane possibilities.

Tabby's Star

The star KIC 8462852 became famous after Planet Hunters volunteers flagged its irregular dimming, and some commentators raised the possibility of a Dyson structure. A 2018 analysis by Boyajian and colleagues, published in The Astrophysical Journal Letters, found that the dips dimmed the star by different amounts at different wavelengths. That pattern rules out opaque material and favors dust, mostly grains smaller than a micron. The authors did not explain the star's long-term fading, and they noted that changes in the star itself remain possible. The fair summary is that dust is the leading explanation and the star remains unusual.

Search	Target	Year	Result
G-HAT (WISE)	About 100,000 galaxies	2015	No galaxy largely colonized; about 50 with unusual mid-infrared emission
Project Hephaistos II	About 5 million stars	2024	7 M-dwarf candidates, unconfirmed
Tabby's Star follow-up	KIC 8462852	2018	Dust favored; long-term dimming unexplained

What the silence means

The Milky Way holds stars many billions of years old, which gives any civilization plenty of time to have grown, even at slow expansion speeds. If Type III societies were common, the galaxy should show it. So far it does not. This is the connection to the Fermi paradox, which asks why a galaxy this old is not visibly crowded. Our articles on the Fermi paradox, the Drake equation, and the Great Filter cover the standard answers.

Within the Kardashev framework, the possible explanations include:

1. Life or intelligence is rare. If technological civilizations almost never arise, there is nothing to see.
2. A filter lies ahead or behind. Some step, perhaps the emergence of life, perhaps surviving industrial technology, may be extremely hard.
3. Expansion is not inevitable. Civilizations may choose not to grow, or may stay small for reasons we cannot guess.
4. They are hard to see. Waste heat could be handled in ways we do not search for, or the civilizations could be in the wrong band.
5. The scale is the wrong lens. If advanced societies shift toward computation and efficiency rather than raw energy, they might be invisible by this measure.

The G-HAT quote above captures the logic: the silence supports "they don't exist" and "they aren't doing the thing we are looking for" equally well.

Why the scale is a thought tool, not a measurement

Scientists use the Kardashev Scale to frame questions, not to measure anything. Its weaknesses are well known.

• Energy is not capability. A civilization with far better computation per watt might accomplish more while using less power. Several authors have argued for measures based on information or computation instead.
• It assumes growth. The scale presumes societies expand their energy use. A society that deliberately holds its footprint constant would stay near Type I forever and not be any less advanced.
• It is human-shaped. The scale describes a technological civilization with industry, much like ours. As Wikipedia notes, critics such as Galántai have argued for survival-based rather than energy-based scales, and John Barrow proposed an "inward" scale that measures mastery of ever smaller structures instead.
• Extensions are speculative. Types IV and higher, such as Kaku's proposal of a civilization drawing on dark energy, have no observational footing.

An earlier episode shows the risk of over-reading the scale. In 1963, Caltech astronomers cataloged a radio source named CTA-102, and Kardashev cited it as having characteristics close to an artificial source. It is now known to be the nucleus of a galaxy, not a Type III civilization, as the Wikipedia entry summarizes. Jill Tarter's view, relayed there, was that the episode reflected insufficient knowledge at the time.

What this means for us

The practical value of the scale is in what it clarifies. Humanity uses a tiny fraction of the sunlight that strikes it, and the barrier to using more is mostly engineering and economics, not physics. At 2% annual growth, a tenfold increase in power takes about 115 years. Getting to Sagan's Type I takes roughly three centuries of steady 2% growth, and every doubling brings the waste-heat problem closer.

Beyond that, the ratios become so large that growth itself is the wrong frame. Going from Type I to Type II is a factor of 10^10. At a sustained 1% annual increase, that takes about 2,300 years, and a 1% annual increase for 2,300 years would be a historical anomaly by any measure. Going to Type III adds another factor of 10^11 plus the speed-of-light barrier. The scale's real lesson might be less "how do we get there" than "why would growth on this scale continue at all," which is the question the empty sky invites.

Frequently asked questions

Q: Is the Kardashev Scale used by professional scientists? A: It is used as a conceptual framework in SETI and in discussions of the Fermi paradox, not as a precise measurement. Searches such as the G-HAT galaxy survey and Project Hephaistos test specific predictions that follow from its logic, such as infrared waste heat from large energy users.

Q: What number is humanity on the scale? A: About 0.73. Sagan's formula is K = (log₁₀ W − 6) / 10, with W in watts. World primary energy use of 620 exajoules in 2023 averages about 19.6 terawatts, or 1.96 × 10^13 W, which gives K of about 0.73. By Kardashev's original 1964 definition, which pegged Type I near the power then in use, humanity is already past Type I.

Q: Could renewable energy get us to Type I? A: Sunlight reaching Earth is about 1.7 × 10^17 W, thousands of times more than humanity uses, so the resource is not the limit. The limits are building enough collectors, storing and moving the energy, and dealing with waste heat. Type I on Sagan's definition would require roughly 500 times today's power use.

Q: Are Dyson spheres detectable? A: In principle, yes, because a structure capturing starlight would reradiate heat in the infrared. Project Hephaistos found seven M-dwarf candidates in 2024, but they are unconfirmed and have plausible natural explanations. Tabby's Star, once discussed as a candidate, now has dust as its leading explanation.

Q: Could a civilization use black holes for energy? A: Theory says yes. Accretion onto a black hole can convert roughly 10% to over 40% of the infalling mass into energy depending on spin, compared with 0.7% for hydrogen fusion. The Penrose process could in principle extract rotational energy. Both remain theoretical for any engineered use.

Q: Why has no one found a Type III civilization? A: The 2015 G-HAT survey of about 100,000 galaxies found none largely colonized. Explanations range from rarity of intelligence to different waste-heat signatures to civilizations that choose not to expand. The data cannot distinguish among them.

The takeaway

The Kardashev Scale looks like a ladder, but it behaves more like a map of how large the universe is compared with us. Humanity's 19.6 terawatts is tiny against the 170 petawatts of sunlight on Earth, vanishingly small against the Sun's output, and unimaginably small against a galaxy's. The scale's most durable contribution is the question it forced onto astronomers: if energy-hungry civilizations are common, where is their heat? So far, the infrared sky has no answer.

Sources

• Kardashev scale, Wikipedia, with Kardashev (1964) and Sagan (1973) as primary sources
• Nikolai Kardashev, Wikipedia
• Dyson sphere, Wikipedia, citing Dyson (1960), Science 131, 1667
• Penn State, "Search for advanced civilizations beyond Earth finds nothing obvious" (April 15, 2015)
• Suazo et al., Project Hephaistos II (2024), arXiv:2405.02927
• Boyajian et al., Tabby's Star (2018), arXiv:1801.00732
• Armstrong and Sandberg, "Eternity in six hours" (2013), bibliographic record
• DieselNet, Energy Institute 2024 Statistical Review summary

Related Reading

• What Are Dyson Spheres and Could We Actually Build One?
• What Is the Great Attractor? The Mysterious Force Pulling Our Galaxy
• How Old Is the Universe — And How Do We Actually Know?
• What If Humanity Became an Interstellar Civilization?