When Will the World End: The Physics, Astronomy, and Biology of Earth's Final Chapter

Earth will not end on a single predictable date. The Sun's brightening will boil the oceans in roughly 1 billion years, a plausible large impact could arrive within 100 million to 1.5 billion years, and the Sun's red-giant expansion will likely consume or sterilize the planet in about 7.5 billion years. Each is a probability distribution, not a countdown clock.

Most people asking when the world will end expect one number. Astronomy does not supply one. What it supplies is a set of independent clocks, each ticking at a different rate, each with a different confidence interval. Understanding those clocks, and how they interact, is more useful than any single doomsday date.

The Sun's slow luminosity ramp

Stars like the Sun fuse hydrogen into helium in their cores. That fusion does not run at a constant rate. As helium ash accumulates in the core, the core contracts and heats, and the reaction rate climbs. The Sun has brightened by roughly 30 percent since it formed, and it gains about 1 percent of its current luminosity every 100 million to 110 million years.

For Earth, that slow brightening matters more than any other long-term threat. The planet's surface temperature is set by sunlight, atmospheric greenhouse physics, and a feedback loop between temperature and water vapor. That loop is stabilizing up to a threshold. Past it, warmer air holds more water vapor, which traps more heat, which raises temperature further.

Silicate rock weathering acts as a long-term thermostat because it pulls carbon dioxide out of the atmosphere faster when it is warmer. But weathering has a ceiling. Once incoming solar energy crosses roughly 1.1 to 1.15 times today's value, carbon dioxide drawdown can no longer keep pace. Surface temperatures then climb past the point where liquid water is stable on a global scale.

That threshold arrives in roughly 1 billion years, with a plausible window of 700 million to 1.5 billion. The term for it is the moist greenhouse limit. When it is crossed, stratospheric water vapor builds up, hydrogen escapes to space, and the oceans evaporate from the top down over millions of years.

Here is the detail most summaries miss. Earth will likely stop being habitable for complex multicellular life long before the oceans boil. Complex life depends on a narrower temperature band than microbial life does. Microbes can persist in high-temperature or high-salinity niches long after plants and animals are gone.

Impact risk, quantified

Asteroid and comet impacts follow a different logic. They are stochastic, not scheduled. The probability of a civilization-ending impact in any century is small, but the expected waiting time for the next one at that scale is measurable.

The 2013 Chelyabinsk impactor was about 20 meters across and injured roughly 1,600 people, mostly through shattered glass. It released energy comparable to a few hundred kilotons of TNT. Objects of that size hit Earth every few decades to a century. Nothing about that event threatened the species.

The energy threshold that matters for global consequences is a crater roughly 100 kilometers across. Impacts that large can loft enough dust and sulfate into the stratosphere to suppress photosynthesis worldwide for months to years. The geologic record shows such events are separated by about 100 million years on average. The most recent, the Chicxulub impact 66 million years ago, erased roughly 75 percent of species.

The counterintuitive point is that no known impact can physically destroy Earth. The energy needed to shatter a planet of Earth's mass and gravity exceeds anything the solar system can deliver. Large impacts sterilize and reshape. They do not demolish.

Red giant and after

In about 5 billion years, the Sun exhausts core hydrogen and becomes a red giant. The Sun's radius will expand by a factor of roughly 100, and its luminosity will climb by a factor of more than 1,000 before core helium ignites. Mercury and Venus will be engulfed. Earth's fate depends on how much mass the Sun loses first.

Mass loss matters because it weakens the Sun's gravitational grip, which lets planetary orbits expand. Current models put that orbital expansion at a factor of about 1.5, not enough to save Earth. Most simulations find Earth will be drawn into the outer envelope and destroyed and vaporized, with the debris spiraling inward.

There is a real professional caveat here. The margins are close, and some models allow Earth to survive as a scorched remnant. The honest answer is that the outcome depends on details of solar mass loss and tidal dissipation that remain uncertain at the factor-of-two level.

After that, the Sun sheds its outer layers as a planetary nebula and settles into a white dwarf. Earth, if any remnant survives, is a frozen, airless cinder orbiting a fading ember for trillions of years. The Sun itself fades over roughly 10 trillion years.

Timeline at a glance

Misconceptions worth correcting

The 2012 Mayan calendar prediction is the best-known example of a specific date that failed completely. Nothing happened, because the calendar was a cycle, not a prophecy. Every subsequent date-specific claim about Earth's end has the same structure: a number presented without a mechanism, or a mechanism presented with a wildly miscalibrated rate.

A second misconception is that a nearby supernova or gamma-ray burst is inevitable and imminent. The record shows such events can strip ozone and cause mass extinctions at close range. The nearest known candidate, Betelgeuse, is about 550 light-years away, far beyond the roughly 30-light-year radius where a supernova would cause serious atmospheric damage. It is also not expected to explode for tens of thousands of years at minimum.

A third is the conflation of human catastrophe with planetary destruction. Nuclear war, engineered pandemics, and severe climate change are genuine existential risks to civilization. None of them removes the planet. The distinction matters because the two problems call for completely different responses.

A nuclear winter from a large-scale exchange could suppress global agriculture for years. Engineered pathogens could spread faster than countermeasures can be developed. Runaway climate change could destabilize food systems and governance. Artificial general intelligence misalignment remains an unresolved risk with no agreed probability. None of these risks ends the planet's physical existence.

The professional caveat here is about probability calibration. Existential risk estimates from different research groups differ by orders of magnitude, and no standard methodology exists for assigning them. Treat any single number as a hypothesis, not a measurement.

Practical actions and sensible priorities

Nothing you do will delay the Sun's brightening by so much as a second. What you can do is align your attention with the risks that are actually actionable. The threats with real probability in a human lifetime are almost all terrestrial and almost all mitigable.

Start with impact detection. Surveys funded by NASA and the European Space Agency have catalogued the vast majority of near-Earth objects larger than 1 kilometer. The remaining gap is in the 140-meter to 1-kilometer class, which could cause regional devastation. Continued survey funding and a planetary-defense capability are concrete, tested interventions.

Second, prioritize civilizational resilience over planetary escapism. Food system diversity, backup power, pandemic surveillance, and transparent governance reduce the damage from shocks that are far more likely than any astronomical event. Space settlement is not a substitute for these measures within any realistic time horizon.

Third, calibrate your own anxiety to the timeline. A risk in 1 billion years is not a plan, a deadline, or a call to action. A risk in 30 years is. Sorting threats by timescale is the most useful mental habit you can build around this topic.

Limits of what science can say

Every date in this article is a model output with error bars. The 1-billion-year moist greenhouse threshold assumes a particular climate sensitivity and weathering response. Change either and the number moves by hundreds of millions of years. The 7.5-billion-year red giant figure rests on stellar models that have been tested on old stars but never observed directly in a Sun-like system.

Prediction of stochastic threats is worse. An impact date cannot be known before the object is discovered, and the probability of a civilization-ending impact in the next century is uncertain by at least a factor of a few. The honest scientific position is that we can describe the physics and bound the uncertainty, but we cannot schedule the end.

That limitation is not a failure. It is the correct answer to a question that was never well-posed. The world has no expiry date printed on it, and no clock is counting down toward one.