Fri, 11 Jul 2025 21:28:00 +0000
CloudKitchens/Lab37's automated bowl-making machines will begin deployment at customer sites within the current calendar quarter (the same quarter in which this episode, released July 11, 2025, was recorded).
“the point is, is, well, where we are right now is we've got customers. And so those customers are starting to deploy this quarter.”
Within roughly 20 years of this discussion (by about 2045), machines will be widely available that can produce very high-quality, highly personalized meals for people, effectively automating meal preparation at scale.
“If you go to like in 20 years, I think everybody agrees you, you will have machines making very high quality, very personalized meals for everybody.”
Elon Musk/XAI’s GPU cluster used for training Grok will be scaled up to approximately 1,000,000 GPUs over time, from the previously announced 100,000 and 250,000 GPU configurations.
“he made this huge bet on this 100,000 GPU cluster. People thought, wow, that's a lot. Is it going to bear fruit? Then he said, no, actually, I'm scaling it up to 250,000. Then he said, it's going to scale up to a million.”
xAI's Colossus cluster scaled well beyond 250,000 GPUs by late 2025, reaching over 550,000 GPUs with a stated goal of 1 million GPUs.
Tesla will release its ‘Hardware 5’ Autopilot/FSD computer in the next calendar year (relative to this episode) and that new hardware will significantly improve the capabilities of Tesla Full Self-Driving compared with prior hardware generations.
“hardware five is coming out on Tesla probably next year, which is going to make a big difference in what FSD can do.”
Future versions of Grok (after Grok 4) will not be trained on conventional public or proprietary datasets scraped from the existing internet or other ‘in-the-wild’ human-created corpora.
“The other crazy thing that he said subsequent versions of grok are not going to be trained on any traditional data set that exists in the wild.”
Later Grok versions increasingly incorporated synthetic and agent-generated data alongside traditional web-scraped data, but training has not become entirely free of conventional datasets.
XAI will shift Grok’s future training regime to rely primarily on synthetic data generated by AI agents themselves, using that agent-produced synthetic data as the main driver of model training instead of human-generated datasets.
“He said that he's going to have agents creating synthetic data from scratch that then drive all the training, which I just think is it's crazy.”
Synthetic data generation by AI agents became an increasingly important part of frontier model training pipelines through 2025-2026, though it has not become the sole or primary driver replacing human data entirely.
Within approximately 1–3 years of this conversation, the use of human-labeled data for AI training will largely disappear across most applications; after that window, very few AI systems will rely on human-labeled datasets for training.
“there may be a year, two years, three years max when anybody uses human labeled data for maybe anything.”
Elon Musk’s attempt to create a new third political party (the American Party) will fail to become a viable or successful third party in U.S. politics.
“Elon is probably a replacement level politician… the third party stuff is not going to work.”
Musk's America Party remained largely aspirational through 2025-2026, without formal ballot registration success or major electoral wins, consistent with the prediction that the third-party effort would not become viable.
The U.S. Senate filibuster rule will eventually be abolished; it will not persist indefinitely in its current form.
“I think that the filibuster, it's just a matter of time. I think it's on borrowed time.”
The Senate filibuster remained intact as of mid-2026 despite recurring predictions of its demise; whether it is eventually eliminated remains to be seen.
At some future point, a Senate majority leader will eliminate the filibuster (or effectively neutralize it), allowing the majority to pass measures with a simple majority vote instead of 60 votes.
“I do think the filibuster is going to go away… I think somebody will eventually get impatient and just steamroll this thing.”
No Senate majority leader had eliminated the legislative filibuster as of mid-2026; it remains a live but unresolved possibility.
Approximately half of Jason Calacanis’s top 10–20 close friends in his ‘friend list’ will choose to join or publicly support Elon Musk’s American Party if it is formed.
“Out of those, 50% will join Elon's party.”
No verifiable data on Jason Calacanis's personal friend group's political affiliations with the America Party was found.
Candidates backed by Elon Musk’s new American Party will be able to win a few U.S. House of Representatives races but will not win any U.S. Senate races in the foreseeable initial cycles of the party (i.e., in its first election cycles after formation).
“I think you can win a few House races. I think that's doable. I don't think you can win a Senate race.”
The America Party had not run or won any federal House or Senate races as of mid-2026, so this prediction has not yet been tested.
If and when a specific large-scale federal workforce reduction plan under Trump’s EO is litigated up to the Supreme Court, the Court will still rule in favor of allowing its implementation, but by a narrower margin than 8–1.
“This one's a little bit easier. It'll get more complicated. Again, like this EO is only approved to allow for the planning. I think the vote might be closer. I think there's still a majority on the Supreme Court for the actual implementation, but it may not be eight one when there's a specific plan that has to navigate its way through the courts again.”
No specific Supreme Court ruling on a large-scale federal workforce reduction plan under this particular executive order had been confirmed as of mid-2026 with the vote margin predicted.