Sam Altman: "Never a Better Time to Do a Startup"

Predictions and Expectations:

• Sam Altman expects this will be "the best time in the world to do a startup" due to the vision of AGI.
• Sam Altman believes startups "will continue to be important forever" and that "startups will be much more important" if AI causes a big change.
• Sam Altman is not sure if the advice to move to the Bay Area will be true "for the next 10 years" but suspects it will.
• Sam Altman expects that "startups created today" will be "much more valuable" and "much more impactful than startups of the past," betting "there's some future trillionaires" in the audience.
• Sam Altman believes the best startups tend to cluster "when the technology landscape is moving very quickly," costs are coming down, and cycle times are short, predicting the current moment feels "very big for those things."
• Sam Altman believes talent in startups will increasingly "cut against many years of experience in favor of people who have a lot of fluency with the tools" like AI agents.
• Sam Altman believes that "if you can't find anybody else that shares your belief" you should "pay attention to that" regarding co-founder matching.
• Sam Altman believes "the premium on doing YC is bigger now than it's like ever been before" and the distance between YC and second place has expanded.
• Sam Altman believes "anybody who's not taking [AI safety] seriously" is "not taking this seriously enough," expecting that "loss of control accidents are not entirely theoretical things."
• Sam Altman believes "power will be too concentrated" if not distributed via startups, fearing a "long-term disaster" where "one company or person or model, having more power than everybody or everything else."
• Sam Altman expects that worldwide demand for inference will grow "10x a year for the next many years," noting internal YC accounts suggest it "might be 90,000x."
• Sam Altman believes "we will sort of never be out of the compute shortage" as demand for high-quality intelligence is "effectively uncapped."
• Sam Altman expects that in another six and a half years, the average person might use "500 billion tokens a month" and the token leader might use "a quadrillion or quadrillions of tokens a month."

Timelines and Milestones:

• Sam Altman predicts "the next six months" will feel "like maybe equivalent to the last two years of model progress."
• Sam Altman expects this "to happen again" regarding token usage scaling within "another six and a half years."
• Sam Altman expects these scaling trends to come to pass "10 years from now" when intelligence is fully on tap.
• Sam Altman remembers a milestone where it took "three months to build" a startup in "2020 years ago," contrasted with the ability now to do similar work in "like seven minutes by a coding agent" or "17 seconds" with agents.

Technology and Product Direction:

• Sam Altman believes "we will see a golden age of startups" where founders do things "completely impossible" a year ago in "sort of like the YC timeframe."
• Sam Altman expects "much more of that" where people grow up using AI to "automate an entire startup of agents" with small teams and compute.
• Sam Altman believes "the world does not understand how to intuit exponentials" and expects "new exponentials forming right now" that the world "will eventually" understand.
• Sam Altman believes the "best time to do a startup" will be "right now" as the field goes through a "very deep period" of model progress.

Market and Industry Outlook:

• Sam Altman expects "hard tech" startups at YC to grow from "five or ten percent" for many years to "15, 20, 25 percent" on the back of easier agent use and falling costs.
• Sam Altman expects the average "tokens per month" per capita to rise from a current "100,000" to "500 billion" in six and a half years, assuming the current leader scales from 100,000 to hundreds of billions.
• Sam Altman fears a dystopia 10 years from now where we "overreact to AI safety," resulting in "material abundance" but "no freedom" or "agency" in a "perfect surveillance state."
• Sam Altman believes the economy is best off when "power is very widely distributed" and fears "concentration of power to a degree we have never seen."

Company Plans:

• Sam Altman plans to "fight for why startups are important" and "help put like relatively more power in the hands of founders" to "encourage more people to start them."
• Sam Altman plans to "push for more startups" and encourage founders to take "more ambition and bigger swings."
• Sam Altman plans to "enable as many startups, as many companies, as many people as possible" to ensure AI power is widely distributed.
• Sam Altman plans to "ensure enough of a safety bar" to prevent incidents like the Hugging Face leak while avoiding imposing a single worldview.

Financial Guidance:

• Sam Altman expects demand for inference to grow "10x a year for the next many years."
• Sam Altman notes internal YC accounts suggest demand growth "might be 90,000x."

Risks and Caveats:

• Sam Altman fears that if power concentrates in "a small number of models or companies," it will lead to a "long-term disaster" where society is locked into "one AI's or one person's or one company's moral worldview."
• Sam Altman fears a future where the society "overreacts to AI safety," resulting in a world with "no privacy" and "nothing left in the world for you to really do."
• Sam Altman cautions that if a startup "starts the same startup as everybody else," they get hype but "much less frequently the big outcomes."
• Sam Altman warns that making "sarcastic comments" on Twitter to score "internet points" is a "morally bankrupt thing to do" that will "poison your soul."

Confidence and Disagreement:

• Sam Altman is sure that "Paul Graham is probably the most important force in startups of the last few decades."
• Sam Altman is confident that "startups are the coolest thing in all of business" and "really like the main thing that keeps the economy from becoming stagnant."
• Sam Altman is not sure what the "new exponentials" are but believes "there's got to be new exponentials forming right now."
• Sam Altman expresses doubt that the "live-action role play" meme regarding startups is helpful, calling it "deeply offensive" and asserting "earnestness" is what YC loves.