Elon Musk at Davos: Space AI, 100GW Solar, and a 2031 Timeline for Superhuman Intelligence

The most interesting thing about this Davos conversation isn’t the vision. Musk has talked about Mars, robots, and solar before. What’s different here is the density of falsifiable commitments: 100GW solar manufacturing in 3 years, space AI deployment cheaper than ground in 2-3 years, robot public sales by end of 2027, AI surpassing collective humanity by 2031. Larry Fink, running the world’s largest asset manager, kept pushing on a single question: will any of this actually reach ordinary people?

A Quick Overview

BlackRock CEO Larry Fink hosted Elon Musk for a 36-minute fireside chat at the 2026 World Economic Forum Annual Meeting in Davos. Fink opened by noting Tesla’s 43% compound return since IPO (vs. BlackRock’s 21%), then steered the conversation through AI and robotics, energy bottlenecks, space economics, self-driving timelines, and Musk’s personal philosophy. The tone was friendly but substantive, with Fink consistently pressing Musk on how technological abundance translates to broad economic benefit rather than narrow concentration.

Robots Will Outnumber People

Musk’s company mission has quietly expanded from “sustainable energy” to “sustainable abundance,” and the mechanism is straightforward: economic output equals average productivity per robot times the number of robots. When robots outnumber humans, goods and services become so abundant that “you won’t even be able to think of something to ask the robot for.”

The use cases he emphasized weren’t industrial. They were personal: watching children, caring for pets, looking after elderly parents. He pointed out a demographic reality that makes this urgent: there simply aren’t enough young people to care for the old.

“My prediction is there’ll be more robots than people.”

Fink pressed on whether this abundance would be broad or narrow. Musk’s answer relied on market dynamics rather than policy: AI costs are already very low and dropping on a month-to-month basis, open-source models lag closed ones by only about a year, and AI companies naturally chase the largest possible customer base. The conclusion: abundance distributes itself because companies want maximum reach.

On human purpose in a post-scarcity world, Musk was unusually candid. You can’t have both mandatory work and universal abundance. They’re inherently incompatible. “Nothing’s perfect,” he said, and moved on.

Tesla Optimus robots are currently doing simple factory tasks, with complex tasks expected by end of 2026 and public sales by end of 2027.

The Real Bottleneck Is Electricity, Not Chips

AI chip production is growing exponentially, but electricity supply grows only 3-4% per year. Musk’s striking prediction: by late 2026, the industry will be producing more chips than it can power. The exception is China.

“We’re very soon, maybe even later this year, we’ll be producing more chips than we can turn on.”

China’s solar numbers are staggering. Manufacturing capacity: 1,500GW per year. Actual deployment: over 1,000GW per year. Paired with batteries for steady-state power, that translates to roughly 250GW, equivalent to half of total US electricity consumption (500GW average). Just from solar. Just in one year.

For the US, Musk offered a calculation that makes the scale tangible: a 100-mile by 100-mile area (160km x 160km) of solar panels could power the entire country. A small corner of Utah, Nevada, or New Mexico. The same math works for Europe with parts of Spain and Sicily.

So why isn’t it happening faster? In the US, extremely high tariffs on Chinese solar panels artificially inflate costs. Musk’s solution is to build domestically: SpaceX and Tesla are each constructing 100GW of annual US-based solar manufacturing capacity, expected to take about three years. For other countries, his advice was blunt: buy China’s incredibly low-cost solar panels directly.

Then he offered his cosmic perspective on energy. The sun is 99.8% of the solar system’s mass. Jupiter is about 0.1%. Even if you burned Jupiter in a thermonuclear reactor, the sun’s energy share would still round up to 100%. Teleport three more Jupiters into the solar system, burn them all, and the sun’s share still rounds to 100%.

“It’s really all about the sun.”

The Lowest-Cost AI Is in Space

This was the conversation’s most provocative claim. Within 2-3 years, Musk argued, space will be the cheapest place to deploy AI.

The physics: solar panels in space produce 5x more energy than on the ground. No day-night cycle, no seasons, no weather, no atmospheric attenuation. Cooling is essentially free: point the radiator away from the sun, and you’re looking at 3 degrees Kelvin. Solar on one side, cooling on the other.

The prerequisite is full rocket reusability. SpaceX has landed boosters over 500 times with Falcon 9, but the upper stage still burns up on reentry, costing the equivalent of a small jet per launch. Starship, the largest flying machine ever built, aims to achieve full reusability in 2026. If it works, access to space drops by a factor of 100, to below air freight rates (under $100 per pound).

“The lowest cost place to put AI will be space, and that’ll be true within two years, maybe three at the latest.”

The long-term scaling potential: hundreds of terawatts. No land use on Earth required.

Self-Driving Is Solved

Musk declared full self-driving “essentially a solved problem,” noting weekly software updates and a particularly telling market signal: insurance companies are now offering half-price policies for FSD users. When the people whose entire business is pricing risk give you a 50% discount, that’s a stronger endorsement than any benchmark.

Robotaxi service is live in several US cities, with widespread rollout expected by end of 2026. Supervised FSD approval in Europe is expected as early as next month (February 2026), with a similar timeline for China.

A Tiny Candle in a Vast Darkness

Musk’s philosophical thread was about consciousness as an endangered phenomenon. We don’t know of life anywhere else. With 9,000 Starlink satellites in orbit, SpaceX has never had to maneuver around an alien spacecraft.

“The image in my mind is of a tiny candle in a vast darkness, a tiny candle of consciousness that could easily go out.”

SpaceX exists to ensure that if disaster strikes Earth, consciousness continues elsewhere. The Mars trip is 6 months each way, with launch windows every 2 years. Asked if he wants to die on Mars: “Yes, but just not on impact.”

On aging, he offered a genuinely interesting observation: every cell in your body ages at the same rate. Nobody has an old left arm and a young right arm. This implies a synchronizing clock across 35 trillion cells, and once identified, aging becomes “incredibly obvious” and “very solvable.” But he also noted death’s upside: without it, society risks ossification, losing the vibrancy that comes from generational turnover.

His closing advice was characteristically binary:

“It is actually better to err on the side of being an optimist and wrong rather than a pessimist and right.”

Some Thoughts

What makes this conversation worth tracking as a historical document isn’t the vision itself, most of which Musk has articulated before. It’s the specificity of the commitments, each with a clear timeline that will be validated or falsified:

• AI surpasses any individual human: end of 2026
• AI surpasses all of humanity collectively: 2030-2031
• Space becomes lowest-cost AI deployment: 2-3 years
• Starship full reusability: 2026
• Optimus public sales: end of 2027
• 100GW US solar manufacturing: ~3 years

Larry Fink’s persistent “but will it be broad?” questioning reflects the capital world’s central anxiety about AI. Musk’s market-dynamics answer (near-zero costs + companies chasing max users = natural distribution) is logically clean but assumes AI becomes a universal utility like the internet rather than a scarce resource controlled by a few. That assumption deserves scrutiny.

One notable signal: Musk publicly recommended other countries buy cheap Chinese solar panels while criticizing US tariff policy for hindering deployment. Given his role as DOGE head and his relationship with the US government, this is a position worth watching.