What Does The World Look Like From Here, According to Sam Altman?
In a wide-ranging chat with legendary investor Vinod Khosla, OpenAI’s CEO dropped some mind-bending predictions about where AI is headed—and what it means for businesses, jobs, and your wallet. The conversation revealed a stark contrast: Khosla focused on the disruptive replacement of jobs, while Altman painted a picture of human-AI collaboration leading to a future of radical abundance.
If you want to jump to the interesting part of the conversation, here's our top moments with timecodes. If you want to read more about the key insights, scroll on down below.
Unique Angles, Predictions, and Insights
- Prediction (The 2035-2050 World): Sam Altman predicts that while the fundamental human experience (socializing, biological drives) won't change much, the technological capabilities available to a single person will be unimaginably different, making precise predictions about specific inventions (like Dyson spheres or nanobots) difficult but the scale of change enormous. (1:02)
- Point of View (The Demise of Incumbents): Vinod Khosla frames the future with a stark prediction: "a faster demise of the Fortune 500 in the 2030s than we've ever seen." (2:16) Sam largely agrees but pivots to a more immediate disruption in the software world.
- Insight (The Future of Software): Sam believes we are not far from a world where any software you need can be written "just in time" by an AI. Instead of buying a SaaS product, you'll simply describe what you want to an AI and have it run, fundamentally altering the physics of software companies. (3:03)
- Point of View (The Limit of AI in Jobs): In a fascinating debate, Vinod asks if AI will do 80% of all intellectual jobs. Sam counters that the key isn't what AI can do, but what humans want a human to do. He argues that our "deep biological programming" will always value human connection, suggesting a mediocre human teacher could be more motivating than a perfect AI teacher. (5:26)
- Prediction (The "Vibes" of Progress): Sam predicts the next 18 months will see a massive leap in AI capability (going from "10 to 100"), but it will feel less wild than the initial launch of ChatGPT. The zero-to-one moment was the biggest psychological shock; now, astonishing progress is simply expected. (10:13)
- Insight (The Simple Engine of Progress): Sam boils down the last few years of AI progress to a simple, grinding formula: better algorithms, bigger computers, and more data. He also forecasts a future of "continuous learning" where systems just run forever, constantly getting smarter. (13:04)
- Point of View (Measuring AI Research): Vinod defines the key test for AI-driven research as when "AI is coming up with new hypothesis that it tests by itself." (16:11) Sam offers a more pragmatic take: he's "equally happy" whether the AI develops the hypothesis itself or simply enables a human to have a breakthrough they couldn't have had alone. The only metric that matters is the overall rate of progress. (16:31)
- Actionable Takeaway (For Investors): In what is perhaps his most pointed piece of advice, Sam tells capital allocators to spend 0% of their time trying to fund the next OpenAI and 100% of their time investing in "the thing that comes next"—the novel companies and ideas that are now possible because AGI exists. He compares it to the transistor: the real value was created by companies using the transistor, not the ones making it. (17:45)
- Interesting Story (The Unplanned Success of ChatGPT): Sam reveals that OpenAI initially struggled to build a product for GPT-3. The API's only commercial success was copywriting tools, but the team noticed the internal "playground" was a sleeper hit—some users would just chat with the model all day. This user signal, showing a desire for conversational interaction, was the direct inspiration for building ChatGPT. (23:43)
- Actionable Takeaway (For Founders): Sam shares a critical learning from ChatGPT's early days: "If you have a product that has any retention at all, you're actually in really good shape." He notes that retention for their internal test group was "atrocious," but the small cohort that did stick around used it more and more over time, which was a powerful signal. (25:02)
- Prediction (The True Enterprise Revolution): While AI "virtual co-workers" are powerful, Sam is more excited about what happens when an enterprise can "throw an entire cluster of compute at one really hard problem," like discovering a new material or optimizing a vastly complex supply chain. (29:45)
- Prediction (The Near-Term Disruptor): For the remainder of the year, Sam bets the AI software engineer will be the single most disruptive force in enterprises, creating a significant performance gap between teams that adopt it and those that don't. (31:04)
- Actionable Takeaway (How to Build in Uncertainty): Sam's advice for entrepreneurs is to operate under one core assumption: models will get roughly 10x better every year on every dimension (smarter, cheaper, faster, etc.). The guiding question for product development should be, "Given this relentless improvement, what should I build and when?" (37:48)
- Prediction (The 10-Person, $1B Company): When asked about the timeline for a 10-person company achieving a billion dollars in revenue, Sam states, "I would bet the company has either already started or will start in the next couple of years." (39:40)
- Point of View (AI and Global Equality): Sam pushes back against the narrative that AI will only make the rich richer. He argues that by making powerful tools like medical advice, education, and software creation free or very low-cost for billions of people, AI and capitalism will act as a massive force for global benefit, just as technology has for centuries. (41:29)
- Prediction (A Deflationary World of Status Games): Sam expects AI to create a "hugely deflationary" economy where necessities (healthcare, food, education) become extremely cheap. The excess wealth generated will then flow into "the silliest status games," bidding up the price of things like Da Vinci paintings to a trillion dollars or galaxies to a quadrillion dollars. (45:18)
- Insight (The Future of Competitive Moats): Sam believes some traditional moats like network effects and brands will persist in the age of AI, but many others will disappear. He views the challenge of discovering the new moats as "the fun part of business." (48:03)
- Prediction (The Convergence of AI and Energy): Sam makes a stark economic prediction: "The degree to which 10 years from now I expect the cost of AI to converge to the cost of electricity is hard to overstate." (50:34)
- Insight (The Data Bottleneck in Different Sciences): He speculates that AI might be able to solve physics with existing data alone, but biology is different. To cure diseases, AI will likely need to become part of an "active learning system" that can request new wet lab experiments to generate the specific data it needs. (54:50)
Here's the 10x leap coming in the next 18 months.
Altman expects the progress in AI capability over the next year and a half to be equivalent to going from a level "10 to 100." For context, the entire jump from nothing to today's models was like going from "1 to 10." But don't expect the same societal shock as the first ChatGPT launch. Why? Because now, as Altman puts it, "everybody's already accepted that AGI is going to happen." The progress is now expected to be astonishing.
He also named the first job to be massively disrupted.
When asked about the biggest short-term impact on companies, Altman was crystal clear: the AI software engineer. He predicts this will be "the big story for the rest of the year," as teams that master this new workflow will dramatically outperform those that don't. While other functions like sales and customer support are also being automated, he sees coding as the first major domino to fall.
Here are Altman’s other key takeaways for founders and investors:
- Stop funding AI research labs. His starkest advice was for investors to spend "0% of my time" trying to find the next OpenAI and "100% of my time" funding the applications being built on top of powerful new models (this is Sam talking his own book, of course, but probably good advice??).
- The 10-person, $1B company is coming. He believes a company with a tiny headcount and massive revenue has likely "already started or will start in the next couple of years," citing the example of one person with 50,000 GPUs discovering a new drug.
- Get ready for a "hugely deflationary" world. In the 2030s, Altman expects AI to make essentials like healthcare and education nearly free. The massive wealth created will instead flow to what he calls "the silliest status games," like bidding up the price of fine art to a trillion dollars.
What to do: Altman's message is a clear signal to shift focus from building foundational models to building applications that assume models will be ~10x better every year. For founders, this means designing products for the AI of tomorrow, not today. For professionals, the rise of the AI software engineer is a blueprint for how other white-collar roles will evolve: AI won't just be a tool you use, but a virtual coworker you direct.
The Long View: A World Remade by 2035
Looking towards the 2035-2050 timeframe, Altman predicted a rate of technological change that is "difficult to have a current framework for." While he believes the core "human experience"—our biological drives for connection, status, and community—will remain largely unchanged, the power vested in a single individual will be unrecognizable. "What one person can get done," he stated, "feels like it will be very different."
This shift will bring a reckoning for the corporate world. Khosla predicted a "faster demise of the Fortune 500 in the 2030s than we've ever seen," a sentiment Altman quickly endorsed. "I would bet that most current companies fail to adapt quickly enough," Altman said, pointing to the fundamental change in how software itself will be created. He envisions a world of "just in time" software, where instead of buying a SaaS product, a user can simply ask an AI to build a custom tool on the fly. This alone threatens the foundation of the modern software industry.
The conversation then turned to the future of work, exposing a key philosophical divide. Khosla questioned whether any intellectual professions would survive AI's encroachment. Altman’s response was nuanced. While conceding that AI will eventually be able to perform "maybe almost all of almost all jobs," he argued that human preference, driven by "deep biological programming," will keep people in many roles.
"Maybe you could have a great AI teacher and it will not be as motivating to you as a mediocre human teacher," Altman mused. "I could totally believe there's something very deep in just knowing that it's a real person or not." Khosla pushed back, asserting the objective superiority of AI in fields like medicine and education would win out. Altman’s counter was personal: an AI investor will undoubtedly outperform Khosla on the merits, but his human encouragement would always be more meaningful.
The Near-Term: A 10x Leap in the Next 18 Months
Bringing the timeline closer to home, Altman laid out an explosive prediction for the next 18 months. He described the coming leap in AI capability as moving from a level "10 to 100," a tenfold improvement. For comparison, he characterized the entire journey from pre-GPT models to today's systems as a jump from "1 to 10."
However, he cautioned that this monumental advance won't create the same public shock as the original launch of ChatGPT—what he called the "zero to one moment." The difference, he explained, is expectation. "I think like everybody's already accepted that AGI is going to happen," he observed. The progress is now expected to be astonishing, which paradoxically lessens its psychological impact even as its real-world consequences grow exponentially.
The engine driving this relentless progress, according to Altman, remains a surprisingly simple formula: better algorithms, bigger computers, and more data. This "grind," as he called it, is creating steeper and steeper scaling laws. The next frontier is building "continuous learning" systems that are always on, always improving.
When Khosla asked when AI would take over the task of AI research itself, Altman described not a sudden handoff but a "messy joint acceleration." He painted a picture of a human researcher whose output is amplified 10x by AI tools—a scenario where the AI is doing 90% of the work, but the human still feels they are directing the process. This symbiotic relationship, he argued, will accelerate progress across the entire technology supply chain, from designing data centers to discovering novel chip architectures.
A New Playbook for Business and Investment
For the entrepreneurs and investors in the audience, Altman offered stark, actionable advice that amounted to a complete strategic reset. When asked if industry leaders like OpenAI would simply cement their advantage, he pivoted. "If I were an LP," he began, "I would be spending 0% of my time trying to figure out how to invest in another AI research lab and 100% of my time figuring out how to invest in the thing that comes next."
He compared the emergence of AGI to the invention of the transistor. The big returns didn't go to the handful of companies that manufactured transistors, but to the thousands of innovative companies—from Apple to, ironically, OpenAI—that were enabled by them. "Chase the future," he urged, "not the thing that worked in the past."
This new paradigm will birth a new type of company. Altman confidently stated that a 10-person company generating a billion dollars in revenue has "either already started or will start in the next couple of years." His primary example was scientific discovery: a single person armed with 50,000 GPUs could discover a blockbuster drug, capturing immense value with a tiny team.
In the more immediate enterprise landscape, he identified the "AI software engineer" as the single most disruptive force for "the rest of the year." He noted that companies and teams who master this new way of coding are already beginning to significantly outperform their peers.
The Deflationary End Game
Ultimately, Altman's vision culminates in a radical restructuring of the global economy. He predicts a "hugely deflationary economy in the 2030s," where AI-driven productivity makes essential goods and services like healthcare, education, and energy abundant and nearly free for everyone.
This raises a fascinating question: if everything we need is cheap, where does the immense wealth generated by this new economy go? Altman's answer was both playful and profound. It will flow, he believes, into "the silliest status games to play." He imagines a world where people "go bid up Da Vinci paintings to a trillion dollars or whatever," creating artificial scarcity to satisfy the innate human drive for status and competition, while the floor of human well-being is raised for all.
In this world, technology, and specifically free or low-cost AGI accessible through platforms like ChatGPT, becomes the great equalizer. This is not to say government has no role; Altman sees a critical need for global guardrails, policies to ensure compute remains abundant, and frameworks for data access. But his core belief is unwavering: technology is the most powerful force for spreading global benefit, and the AI revolution will be its greatest chapter yet.
