A Trump Science Advisor Says Elon Musk And Sam Altman Won’t End Up With AI’s Money: ‘There Will Be A Billionaire That Will Emerge That Has Zero Net Worth Today’

A Trump Science Advisor Says Elon Musk And Sam Altman Won’t End Up With AI’s Money: ‘There Will Be A Billionaire That Will Emerge That Has Zero Net Worth Today’

Tue, September 15, 2026 at 2:45 PM GMT+3 5 min read

A man President Trump appointed to his science and technology council in March says the three people everyone assumes will own the AI economy will not, and that the person who does is currently worth nothing. "I don't buy into the narrative that Dario, Elon, and Sam end up having all the value in AI," David Friedberg said in a clip published on September 2 from a Diary of a CEO interview. "Because in the next 10 years, there will be a billionaire that will emerge that has zero net worth today, downloaded open-source AI, and built a company that made them a billionaire. And there're going to be someone that's going to come from nowhere."

He used first names. The three he meant are Dario Amodei of Anthropic, Elon Musk, and Sam Altman of OpenAI. Friedberg co-hosts the All-In podcast, runs the venture firm The Production Board, and was named to the President's Council of Advisors on Science and Technology on March 25.

More News from Barchart

Frontier proprietary models are expensive per unit of output; hosted open-weight models are not. Anthropic's top tier and OpenAI's flagship both list $50 per million output tokens on their published pricing pages, with input at $10. Running one of the cheap open models through a commodity host currently costs between about $0.18 and $0.60 per million output tokens. So there is a real opportunity right now for those who can understand open-weight models and monetize and market the right opportunities in that gap.

Sponsored Content: This Startup Is Targeting a Massive Untapped Energy Market

That comparison sets a small open model against the single most expensive closed one. Compare like with like and it collapses. DeepSeek's largest open-weight model, released under an MIT license, runs about $2.60 per million output tokens on one host, roughly 19 times cheaper than the top proprietary tier, rather than a hundred. Moonshot's Kimi K3 is around $15, which is about three times cheaper. And Google's current Pro-tier model, which is proprietary, lists at $12. The honest version of Friedberg's claim is not that the frontier is free, but rather that capable models are now cheap enough that capital is not the barrier it was.

The scale he is betting against is large, and every figure attached to it is an estimate. Forbes put Musk's net worth at about $938 billion this week while Bloomberg estimated about $904 billion, a $34 billion disagreement about the same man. Forbes estimates each Anthropic cofounder is worth $15.5 billion, assuming a stake of just over 1.6%; Bloomberg last said they were worth about $8 billion in May, assuming under 1%. Anthropic declined to comment on the stakes when Forbes asked.

The Altman number quietly supports Friedberg. Forbes puts his worth at roughly $3.3 billion, outside the world's thousand richest people, and states he holds no equity in OpenAI at all, attributing his fortune to outside investments. By that estimate, the chief executive of a company valued in the hundreds of billions has captured almost none of it personally, which is stranger than anything in the clip.

Friedberg also made a flat historical claim that does not survive unqualified. "From data, from evidence, over all periods of technological revolution like this, we've never seen jobs decline," he said. On total payroll headcount over long spans, he is right: US nonfarm employment hit an all-time high of about 159.1 million in August, and no 30-year window since 1939 is negative. Narrow the frame, though, and it stops holding. Twenty-two 10-year windows in that series are negative, all ending between 2009 and 2011. Manufacturing employment peaked at 19.6 million in June 1979 and is now 12.6 million, down about 35%. The share of prime-age men in work has fallen roughly 10 points since the early 1950s, and the aggregate held up mainly because women entered the workforce.

Economists disagree about the mechanism, not the headcount. Daron Acemoglu and Pascual Restrepo found in a 2020 study that one additional robot per thousand workers cut the employment-to-population ratio by 0.2 percentage points and wages by 0.42%. David Autor has argued the opposite emphasis, that automation raises output in ways that increase labor demand. On AI specifically, Stanford's Digital Economy Lab reported last month that employment of 22- to 25-year-olds in AI-exposed occupations sits about 19% below where it would be had it tracked less-exposed peers, while saying in the same paper that it finds no evidence of widespread displacement.

One more number needs handling. In a passage that also made a broader political argument this piece does not take up, Friedberg said that "63% of Americans live paycheck to paycheck. People can't afford to pay their bills, and it needs to be fixed." That figure traces to a LendingClub survey from November 2022 in which people described themselves that way. The same series now reads 67%. Bank of America, measuring actual customer transactions instead, puts it closer to 24%, and Bankrate's survey of workers at 34%. The number moves that much because three different things share one phrase.

What would make Friedberg right is observable rather than rhetorical: a company built on downloadable weights reaching a billion-dollar valuation without a frontier lab underneath it. He has made large valuation calls before, including that one piece of SpaceX could be worth a trillion dollars inside 18 months.

On the date of publication, Caleb Naysmith did not have (either directly or indirectly) positions in any of the securities mentioned in this article. All information and data in this article is solely for informational purposes. This article was originally published on Barchart.com