Sat, 29 Jun 2024 04:05:00 +0000
Within 30–60 days of the earlier episode being recorded (the clip they replayed), the Democratic Party will replace Joe Biden with a different presidential candidate ('a switcheroo').
“I believe that's what's going to happen in the next 30 to 60 days. So I'm predicting... There's going to be a switcheroo... 100%.” View on YouTube
Joe Biden withdrew as the Democratic nominee on July 21, 2024, within the predicted 30-60 day window.
In the then-upcoming presidential debate referenced in the replayed clip, Donald Trump will decisively outperform Joe Biden, Biden’s support in the polls will fall to around 30%, and subsequently Democratic Party leadership will orchestrate a ‘graceful’ withdrawal by Biden and replace him with another candidate.
“I think Trump's going to demolish him in a debate. I think he'll sink to 30% in the polls. And then the Democrats will find a way to give him a graceful out, and then they'll field somebody else.” View on YouTube
Trump was widely seen as decisively outperforming Biden in the June 27, 2024 debate, Biden's standing collapsed sharply afterward, and Democratic leadership did ultimately orchestrate his withdrawal and replacement with Kamala Harris.
From June 29, 2024, within 30 days (by around July 29, 2024), Joe Biden will no longer be an active candidate in the 2024 presidential race (i.e., he will drop out or be replaced).
“the hot swaps coming. I just I'm telling you right now, he is not going to be in this race in the next 30 days.” View on YouTube
Joe Biden withdrew from the 2024 race on July 21, 2024, within 30 days of this June 29 prediction.
Following the June 27, 2024 debate, Joe Biden will leave office before his term ends, Kamala Harris will become President for roughly four months and will be celebrated as the first female president, then she will announce she is not running for election (citing family or similar reasons), and the Democratic Party will instead put forward two new, more moderate candidates to contest the 2024 presidential election, with that new general‑election race effectively beginning about 15 days after Biden exits.
“what's going to happen right now, I guarantee you, is he's out. We're going to have President Kamala. She's going to get her flowers for four months as she gets to be the first female president of the United States. Then she steps out of the way. She decides she's not going to run because she's got things to do with her family. And there will be two new people who will be moderates. And then the real election starts in about 15 days.” View on YouTube
Biden did not leave office before his term ended, remaining president through January 20, 2025; Harris did not become president early, but instead remained vice president and became the Democratic nominee, running against and losing to Trump rather than declining to run.
As a consequence of the Biden debate fallout and internal party dynamics, the Democratic Party will undergo a significant internal realignment or ‘reset’ (e.g., leadership, platform, or coalition changes) over the next election cycle or two.
“I think the Democratic Party is It's probably at risk of a pretty meaningful reset.” View on YouTube
The Democratic Party underwent significant internal soul-searching and leadership debate following its 2024 election loss, continuing into 2025-2026.
Within 72 hours of June 29, 2024, Barack Obama will personally speak with Joe Biden and, as a result of that intervention, Biden will agree to step down as the 2024 Democratic presidential candidate (the ‘hot swap’ will occur).
“Obama's going to give him a call. I guarantee you Obama is going to talk to him in the next 72 hours, guarantee the hot swap happens.” View on YouTube
Biden's withdrawal did eventually happen (the 'hot swap'), but it took about 24 days rather than 72 hours, and unfolded through a broader, more gradual pressure campaign involving donors and congressional leaders rather than a single decisive Obama phone call.
Despite post‑debate pressure, Joe Biden will not be replaced as the Democratic nominee before the 2024 general election; there will be no ‘hot swap’ of the nominee.
“I actually think that on balance, the hot swap is not going to happen.” View on YouTube
The 'hot swap' did happen; Biden withdrew and was replaced by Kamala Harris as the Democratic nominee in July 2024.
In the 2024 U.S. presidential election, Jason will not cast his vote for Joe Biden.
“I am not voting for Biden. I have eliminated Biden as a possibility. I was waiting to see what happened last night.” View on YouTube
Since Biden withdrew and was not on the general-election ballot, this prediction about not voting for Biden was trivially borne out.
As OpenAI deepens its ties with U.S. national security and large institutional investors (e.g., BlackRock, T. Rowe, major mutual funds), future U.S. Senate scrutiny of OpenAI will increasingly occur in confidential, national‑security–framed settings, and the company will face effectively no serious legal or regulatory harassment such as IRS audits (i.e., probability of an IRS audit becomes near zero over the coming years).
“So now what happens is, when you have Senate hearings about this stuff, it's more likely that it's confidential behind closed doors. It's under the purview of national security. All these things are beneficial to OpenAI. And then secondly, they were able to get Elon to drop his lawsuit conveniently... The only thing left is to get shares into the hands of the blackrocks, the tea rose, the all the big mutual fund apparatuses of the world that then syndicate to all the individual investors of the world. And you have everything. You have government connections. You have no real legal overhang. Then the likelihood that an IRS agent all of a sudden decides to audit OpenAI is basically zero.” View on YouTube
There is no clear public confirmation of OpenAI Senate scrutiny consistently shifting to confidential national-security-framed settings or of IRS audit probability specifically becoming near zero.
The startup Safe Superintelligence Inc. (SSI), which emphasizes safety in developing superintelligence, will not become a top‑tier winner in the AGI race; its safety‑first approach will cause it to lose competitively to faster‑moving rivals like OpenAI and Google over the next several years.
“I'm not bullish about that pitch because I think it it makes the company a little bit schizophrenic... So you've got to move very fast. On the other hand, you're saying you're going to basically make this very safe... safety concerns are a brake pedal... And so I don't but I don't think that's the recipe for winning because you're the guys who want to move slower.” View on YouTube
Safe Superintelligence Inc. has continued operating and raised substantial funding at a multi-billion-dollar valuation through 2025, but it has not yet shipped a leading frontier product, so a clear competitive-loss verdict is premature.
In the emerging race to AGI over the coming years, AI companies that prioritize safety more than their competitors will systematically lose market share and technological leadership, leading to a Darwinian dynamic that favors less safety‑constrained firms.
“I think this is maybe the tragic situation is we're going to have this competition by all these different companies to advance AI. And the companies that care about safety more than others are going to lose. And so you have this Darwinian effect going on where there's going to be a race to AGI.” View on YouTube
Anthropic, a safety-focused AI lab, has grown into a major competitive player through 2024-2025, complicating the thesis that safety-prioritizing firms systematically lose in the AGI race.
Over the next few years, frontier foundational AI models will converge in capabilities such that they become near‑interchangeable commodities from the user’s perspective (a “consumer surplus”), with only marginal performance differences between major providers.
“Foundational models are quickly becoming a consumer surplus. Every model is roughly the same. They keep getting better and better, but they're also approaching these asymptotic returns.” View on YouTube
Some convergence in capability among top frontier models has occurred, but meaningful differentiation between leading labs (OpenAI, Google, Anthropic, xAI) has persisted rather than full commoditization.
As training data pipelines become more curated and refined, the cost of training competitive frontier models will rise to levels that are untenable for most companies, leaving only a few large tech firms able to afford state‑of‑the‑art model training over the next several years.
“And it looks like one of those variables that people are looking at is how you basically take the internet not as raw data, but then you actually kind of refine it and refine it some more and then use that as the basis of learning. And what that does is it drives up model costs to a degree that are probably untenable for most companies except, but for a few.” View on YouTube
Frontier AI training costs have continued to escalate dramatically, concentrating the ability to train state-of-the-art models among a small number of well-funded firms.
According to Dario Amodei (as cited by Chamath), by around 2027 the cost to train a top‑tier, “good functional” frontier AI model could reach approximately $100 billion, up from mere billions today.
“So I think it was Dario Amodei, the CEO of anthropic, who said the cost of a good functional model today is in the billions, but by 2027 it could easily approach $100 billion.” View on YouTube
This is a prediction about model training costs reaching roughly $100 billion by 2027, which has not yet arrived as of this validation.
If frontier model training costs rise toward ~$100B by ~2027, only mega‑cap tech firms such as Google, Microsoft, Meta, and Amazon will be able to fund such efforts, and smaller AI startups (including Ilya Sutskever’s SSI) will be unable to raise comparable capital to compete at the very top end.
“The problem that that represents for Elia's company, and I wish him the best of luck. But the reality is there isn't a $100 billion for him to have. Google will find it. Microsoft will find it. Facebook will find it... Amazon will find it. But I suspect that these other startups, there just isn't that much money going into AI because the returns don't justify it.” View on YouTube
Mega-cap tech firms have indeed dominated the largest AI compute investments, but well-funded AI-native startups such as OpenAI, Anthropic, and xAI have also raised tens of billions of dollars, complicating the claim that only the largest incumbents can compete.
Over the medium term, the foundational model market structure will resemble ride‑sharing: roughly one dominant commercial winner in closed‑source models, alongside numerous open‑source alternatives that are asymptotically similar in capability, with competition focused primarily on cost and compute efficiency.
“I think that you could make a claim that the AI foundational model market will look similar to that one startup can probably win, but there will be a bunch of open source alternatives. They're all asymptotically similar. And so it's an arms race on cost and compute.” View on YouTube
Rather than a single dominant closed-source winner, several strong competitors (OpenAI, Google, Anthropic, xAI) have coexisted, alongside a real tier of competitive open-source models like Llama and DeepSeek.
Venture capital firms will not fund multiple AI companies at the ‘hundreds of billions of dollars’ scale required for frontier model training; instead, that level of capital deployment will remain limited to a small number of mega‑cap incumbents over the coming years.
“And I just don't see VCs having the temperament and the wherewithal to fund hundreds of billions of dollars into multiple companies to do that.” View on YouTube
In aggregate, investors have funded multiple AI companies at extraordinary scale, with OpenAI, xAI, and Anthropic each raising tens of billions of dollars, even if no single venture fund alone provided that scale of capital.
If regulators do not require a la carte pricing that neutralizes bundle advantages, Microsoft will continue adding new products to its Office/365 bundle each year and will, over time, systematically dominate broad swaths of enterprise software markets through this bundling strategy.
“If we don't do that, I do think that Microsoft will use the power of the bundle to systematically dominate enterprise software, and they won't take on everybody at once. But like I said, they'll every year they'll add a new product to the bundle.” View on YouTube
Microsoft has continued adding AI and Copilot features to its Office/365 bundle and has maintained and strengthened its dominance across enterprise software through 2024-2025.