"Digital superintelligence could arrive as soon as 2025 or 2026."

On June 17, 2025, Elon Musk appeared at Y Combinator’s inaugural AI Startup School in San Francisco, where he sat down with YC CEO Garry Tan for a wide-ranging conversation about artificial intelligence, entrepreneurship, and humanity’s future.

The Intelligence Explosion

Musk opened with a stark prediction: “We are in the extremely early stages of an intelligence explosion. Digital superintelligence could arrive as soon as 2025 or 2026.” He emphasized the magnitude of this transition, comparing it to previous technological revolutions but on an incomparably larger scale.

“If AI development proceeds smoothly without catastrophic outcomes, the economic scale will be more than ten times larger than today’s economy,” Musk stated. The implications, he suggested, dwarf current policy concerns: “Fixing government is like cleaning up a very dirty beach, but the coming AI revolution is a tsunami thousands of feet high. Cleaning up trash on the beach hardly matters in comparison.”

First Principles Thinking: The Foundation of Innovation

When Garry Tan asked about Musk’s approach to seemingly impossible problems, Musk returned to his core philosophy: first principles thinking.

“Most people reason by analogy—they do things because that’s how things have been done,” Musk explained. “First principles thinking means breaking down problems to their most fundamental physical truths and rebuilding solutions from the ground up. This is how you achieve disruptive innovation rather than incremental improvement.”

He traced this methodology through his ventures: questioning why rockets couldn’t be reusable at SpaceX, why electric vehicles had to have limited range at Tesla, and why neural interfaces required invasive surgery at Neuralink.

The Entrepreneurial Journey: From Zip2 to xAI

Musk reflected on his entrepreneurial arc, spanning three decades and fundamentally different industries. Starting with Zip2 in the mid-1990s internet boom, moving through PayPal’s transformation of online payments, then simultaneously scaling SpaceX and Tesla—companies that nearly bankrupted him in 2008.

“People often ask how I manage multiple companies,” Musk said. “The answer is that I don’t really manage them in a traditional sense. I focus on solving the most critical problems in each company, usually related to engineering and product development. Everything else is secondary.”

His newest venture, xAI, represents his direct engagement with artificial intelligence. “With xAI and Grok, we’re building AI systems that are not just capable, but truthful. The most important thing for AI safety isn’t the technology itself—it’s a rigorous adherence to truth.”

Grok 3.5 and the AI Race

Discussing xAI’s progress, Musk provided updates on Grok 3.5, though he was characteristically guarded about specific capabilities. “The pace of improvement is extraordinary,” he noted. “But the real challenge isn’t just making models more capable—it’s ensuring they remain aligned with human values and committed to truth-seeking.”

Addressing concerns about GPU availability, Musk acknowledged the constraints: “Compute is the bottleneck right now. But this is temporary. The real question is whether we’re building AI systems with the right objectives.”

AI Safety Through Truth-Seeking

When pressed on AI safety—a topic where Musk has been both vocal and controversial—he articulated a philosophy centered on truthfulness.

“Many AI safety approaches focus on constraining AI, teaching it what not to say or do,” Musk explained. “I think the better approach is teaching AI to pursue truth rigorously, even when truth is uncomfortable. An AI that values truth above all else is less likely to cause catastrophic harm.”

This philosophy, Musk suggested, contrasts with approaches that embed specific political or social values into AI systems. “Who decides which values to embed? Truth-seeking is the least subjective principle we can optimize for.”

Neuralink: Connecting Mind and Machine

Musk proudly highlighted Neuralink’s 2024 breakthrough: paralyzed patients controlling computers through thought via implanted chips. “This allows people who can’t speak or move to communicate with the outside world. We’re literally giving people back their ability to express themselves.”

Looking forward, he envisions brain-computer interfaces becoming mainstream: “Eventually, this won’t just be for people with disabilities. Neural interfaces will allow humans to keep pace with AI by increasing our bandwidth for information processing.”

The Rise of Humanoid Robots

Musk predicted that humanoid robots will eventually outnumber humans by a factor of 5 to 10. “This will have profound implications for social structure,” he stated. “When there are billions of humanoid robots performing most physical labor, it fundamentally changes economics, employment, and human purpose.”

He framed this not as dystopian but as potentially liberating: “If robots can do any physical task, humans are freed to pursue creative, intellectual, and social activities that machines can’t replicate—at least not yet.”

Becoming a Multiplanetary Species

Returning to SpaceX’s mission, Musk emphasized the existential importance of establishing self-sufficient human presence on Mars. “We should aim to make Mars self-sustaining within 30 years. This isn’t just about having a backup for humanity—it’s about extending the duration of consciousness and intelligence in the universe.”

He acknowledged the scale of the challenge: “Making life multiplanetary is probably the hardest engineering project humanity has ever attempted. But it’s also the most important. Without it, all of civilization is vulnerable to extinction from asteroid impacts, supervolcanoes, nuclear war, or unforeseen disasters.”

Advice for AI Startup Founders

Throughout the conversation, Tan solicited advice for the hundreds of AI startup founders in attendance. Musk’s guidance was characteristically direct:

“Build something people actually want. This sounds obvious but most startups fail because they build something they think people should want, rather than what people actually want. Talk to users constantly. And don’t be afraid to rebuild from scratch if something isn’t working.”

On competition with tech giants, Musk was encouraging: “Large companies have many advantages, but they also have bureaucracy and inertia. Small teams can move faster and take bigger risks. Use that advantage.”

The Coming Transformation

As the conversation concluded, Musk painted a picture of extraordinary transformation ahead. “The 2020s and 2030s will see more change than any period in human history. AI, robotics, brain-computer interfaces, and space exploration are all accelerating simultaneously.”

“The question isn’t whether these changes will happen,” he emphasized. “The question is whether we navigate them successfully. That’s up to all of us—but especially the people building these technologies.”

For the aspiring founders in the audience, Musk’s final words were both encouraging and sobering: “You have the opportunity to shape humanity’s future in ways previous generations never could. Use that power wisely.”