aipod / all-in / 2026-07-03-ai-sovereignty-wars-palantir-…
ALL-IN · Jul 3, 2026

AI Sovereignty Wars, Palantir-Nvidia Deal, SCOTUS Birthright Ruling, Newsom's CA Budget Lie

🎙 Jason Calacanis · Alex Karp · David Sacks · Chamath Palihapitiya · David Friedberg
38 claims10 predictions30 facts6 notable quotes raw JSON ↗

The panel opens on Palantir and NVIDIA's 'sovereign AI' partnership and Alex Karp's CNBC monologue attacking frontier labs, launching a long discussion of 'AI sovereignty' in which Sacks accuses Anthropic of vertically integrating against its own customers (citing the Figma/Claude Design dispute) and Chamath shares 8090 benchmark data showing open-source models wrapped in its own harness can be far cheaper than Claude, while Friedberg predicts enterprises will shift to hub/medium-hub/distributed-spoke AI infrastructure. The group then debates NVIDIA's Nemotron and the Anthropic-OpenAI model-layer duopoly, the brief US export-control action against Anthropic's Fable (Mythos) model, and whether Chinese open-source models like DeepSeek and Kimi should be restricted from import. A sharp Friedberg-Jason argument over AI job loss versus job displacement follows, citing Waymo driver data and a Ramp/Revelio Labs headcount study. The back half covers the Supreme Court's 6-3 ruling striking down Trump's birthright-citizenship executive order — debating the 14th Amendment's original intent versus its text — before pivoting to immigration assimilation and merit-based reform. The episode closes with Friedberg's teardown of Gavin Newsom's 'balanced' California budget as built on accounting tricks amid a wealth exodus and looming pension liabilities, with Chamath and Friedberg forecasting a hard fiscal landing or federal-bailout crisis and Friedberg predicting AOC as the 2028 Democratic front-runner.

SEGMENTS

0:00
Palantir-NVIDIA sovereign AI partnership announcement

Jason announces the Palantir-NVIDIA 'sovereign AI' partnership built on NVIDIA's NemoTron open models, then plays Alex Karp's CNBC monologue attacking frontier labs' enterprise data practices.

2:13
AI sovereignty and the enterprise revolt against frontier labs

Jason coins 'intelligence sovereignty' and Sacks argues Anthropic blindsided Figma with Claude Design and is vertically integrating like Microsoft/Google, urging enterprises to distrust frontier labs with their data.

9:47
Chamath's 8090 open-source experiment and Friedberg's hub-spoke prediction

Chamath cites BCG return-on-capital data and 8090's benchmark showing an open-source model wrapped in its harness is far cheaper than Claude; Friedberg describes life-sciences firms refusing Anthropic's data-sharing offer and predicts a shift to hub/medium-hub/distributed-spoke AI infrastructure.

25:08
NVIDIA's Nemotron, the AI stack, and the model-layer duopoly

Jason asks why NVIDIA is suddenly promoting Nemotron; Sacks lays out the three-layer AI stack (chips/models/apps) and argues the Anthropic-OpenAI model-layer duopoly is bad for the ecosystem and shouldn't be entrenched by regulation.

33:50
Job displacement vs. job loss debate

Friedberg argues there is no data supporting an AI 'job loss' narrative, while Jason insists specific categories (customer support, driving, data entry) will be displaced quickly; they debate Waymo driver data, a Ramp/Revelio Labs study, and a coming premium for 'human in the loop.'

50:21
Fable/Mythos export-control saga

Jason recaps the brief US export-control letter against Anthropic's Fable (Mythos) model and its reversal after Tom Brown replaced Dario Amodei as lead government negotiator; Sacks explains the unique conditions that caused and then resolved it.

54:11
Chinese open-source model import policy

Jason asks why the US doesn't restrict Chinese open-source models like Kimi and DeepSeek; Sacks argues an open-sourced model 'stops being Chinese' once run on US hardware and that an import ban would isolate the US and impose a 'token tax.'

58:17
SCOTUS birthright citizenship ruling

The panel discusses the Supreme Court's ruling striking down Trump's executive order ending birthright citizenship, debating the 14th Amendment's original intent (freed slaves, per Senator Jacob Howard's 1868 debate and the Dred Scott precedent) versus its text today.

1:11:00
Immigration policy: assimilation and merit-based reform

Chamath argues immigrants must assimilate into American culture; Friedberg proposes a 'maker vs. taker' admission framework; Jason floats a points-based, economic-contribution system; Sacks cites Pew polling on deportation attitudes.

1:21:48
California's budget, exodus, and Newsom's 2028 prospects

Friedberg dissects Newsom's 'balanced' $351B budget as built on accounting tricks, tax exodus, and looming pension liabilities; the group forecasts California pension wipeouts, a possible federal-bailout fight, the Billionaire Tax Act's legal fate, and 2028 Democratic prospects.

NOTABLE QUOTES

“Are we really going to outsource the battlefield of this country to the consensus view in Silicon Valley? That is f-ing insane.”
Alex Karp · 1:24 — The episode's cold-open thesis — Karp's CNBC monologue framing AI sovereignty as a national-security and enterprise-safety issue, not just a business dispute.
“There is no free pizza. There's no free beer. When somebody like Sam Waltman comes to you and says, here's some free tokens, your alarm should go up.”
Jason Calacanis · 19:03 — Crystallizes the episode's warning to founders and enterprises about the hidden cost of free frontier-lab access.
“why don't I just take GLM, control it entirely soup to nuts on my own hardware inside of the United States with only US citizens that can touch it. It just seems like the brain dead obvious thing to do.”
Chamath Palihapitiya · 20:30 — Chamath's concrete, first-person conclusion from the 8090 benchmark data — the practical endpoint of the AI-sovereignty argument.
“There is no job loss with AI. It is an absolute scam to tell the world that AI is taking away jobs and destroying jobs and the world is shifting.”
David Friedberg · 33:50 — The sharpest statement of the episode's job-loss-vs-displacement debate, directly challenged by Jason afterward.
“calls it balanced and walks away while the house burns.”
David Friedberg · 1:21:53 — Friedberg's framing line for the entire California-budget teardown that follows.
“There needs to be a hard landing.”
Chamath Palihapitiya · 1:34:03 — Chamath's blunt prediction for how California's fiscal spiral ultimately resolves.

PREDICTIONS ON THE RECORD

Dated and falsifiable — full record at /api/episode/<id>.

David Friedberg · the next couple of months · 15:41
Enterprise AI infrastructure will shift from a large-hub/large-spoke cloud model to a large-hub/medium-hub/distributed-spoke model, with enterprises building their own proprietary models and running their own inference.
“I think the model is shifting where we're going from large hubs, large spokes, to large hubs, medium hubs and then a distributed spoke model, where there will still be neo-clouds and hyperscalers that are being used for inference. But people will also have their own inference instances that they're going to run for their own enterprise setting. And everyone I think is walking this path and they're going to walk this path over the next couple of months.”
Jason Calacanis · the next 3 years · 30:01
Token costs will fall roughly 90% per year for the next three years due to AI model competition.
“The cost of tokens Friedberg is going to go down 90 percent a year for the next three years.”
Jason Calacanis · 10 years · 47:04
Humanoid robots (e.g., Optimus) will fully replace human package sorting and delivery at Amazon within 10 years.
“every single package sorting, every single package delivery will not be done by humans in 10 years.”
David Friedberg · the next year and a half · 45:20
Within the next year and a half, 'human in the loop' will be recognized as a premium good people pay extra for, even amid growing AI automation.
“Human in the loop will be the premium and we will all realize in the next year and a half, the importance of human in the loop, even in the context of AI and automation working. Human in the loop is going to be more valuable and then you're going to pay a premium for human in the loop.”
David Friedberg · 2028 presidential election · 1:28:44
AOC will become the Democratic presidential front-runner and eventually president, driven by cost-of-living anxiety and wealth-disparity backlash.
“I do predict, yeah, so I predict AOC will be president. I mean, obviously, there's a lot of dispersion of what could happen here. But I would say AOC would be my front-runner based on the movement for the extraordinary cost of living in the United States right now.”
David Friedberg · the years ahead · 1:28:44
If a federal bailout of California's liabilities occurs, red states will question remaining in the union, precipitating a genuine crisis of the union in the years ahead.
“And that is where I think you face a crisis of the union in the years ahead.”
Chamath Palihapitiya · roughly the next decade · 1:29:59
California will undergo a 'hard landing': a negotiated settlement wiping out state pensions, followed by a wholesale replacement of the California Constitution and a red-wave political overhaul.
“I think what will happen is that you will wipe out the California pensions and you'll wipe out the pension obligations and you'll do some sort of negotiated settlement. And I think that's where you're going to have a wholesale replacement of the California Constitution.”
Chamath Palihapitiya · a decade (by 2036) · 1:32:57
If California's Billionaire Tax Act passes, it will spark 200-300 lawsuits that take a decade to resolve, ultimately losing at the Supreme Court by 2036 after billions in legal fees.
“you're going to have 200 or 300 individual lawsuits that will take a decade to meander through the courts. And ultimately, I think a lot of people have standing here. And I just wonder what will they have proven at that point, because it will be a decade later, it will be 2036 and the California government will have had to spend billions of dollars on legal fees to fight all these billionaires, all to end up losing at the Supreme Court anyways.”
David Sacks · unspecified · 1:34:15
California will not swing back toward Republican governance ('red wave'); instead it will intensify its existing left-leaning political trajectory.
“this idea that somehow things get bad enough where you get a red wave in super blue California and everything gets magically fixed, not going to happen. What's going to happen is a blue state is going to get bluer.”
Jason Calacanis · unspecified · 1:35:43
If Friedberg and Chamath stay in California, they will end up paying 60-65% in combined taxes.
“I guarantee you, Friedberg and Chamath, you will be paying 60 to 65 percent tax if you stay in California.”

WHO CLAIMED WHAT

Every position taken in the episode, attributed to its holder, with conviction (0–1) and the verbatim receipt.

David Sacks — 14 claims

opinionconviction 0.854:08
Anthropic is following a Microsoft/Google-style playbook of using its dominant model-layer position to vertically integrate into and capture the lucrative verticals built on top of its models.
“They are going to use their dominant position in the model to then grab more and more territory in any interesting and lucrative vertical.”
opinionconviction 0.84:08
Dario Amodei's push to restrict open-source AI models on safety grounds serves Anthropic's business interest in limiting enterprise choice at the model layer, not genuine safety.
“It's dangerous to his business model because his business model requires that customers don't have a lot of choice at the model layer.”
opinionconviction 0.826:48
The AI model layer has become a duopoly (Anthropic and OpenAI) that Anthropic is trying to entrench via a safety-focused regulatory-capture agenda.
“We have Anthropic pushing for a regulatory capture agenda that would probably enshrine that duopoly situation at a regulatory level because they're pushing for a safety agenda where Dario explicitly says that these other models are not safe.”
opinionconviction 0.8526:48
Government should refrain from entrenching the Anthropic/OpenAI model-layer duopoly and instead keep the model layer competitive, since competition benefits chip makers, developers, enterprises, and consumers alike.
“they should do everything they can to keep the model layer competitive because competition is what brings out the best and it's good for the ecosystem.”
opinionconviction 0.7546:15
Entry-level customer-support work is not a great career, but most US enterprises have already outsourced that tier abroad, making the higher-value escalation tiers much harder for AI to replace.
“I would not want to be a level one customer support rep because I do think that if all you're doing is answering forgotten password requests, that's not a great place to be. However, those jobs are basically outsourced to the Philippines and places like that a long time ago.”
opinionconviction 0.852:00
The Anthropic export-control episode was a one-off caused by a unique confluence of factors and shouldn't be read as a signal of broader US tech-export policy toward allies.
“I think if you change any one of those three facts, this would have ended up not causing the government to basically send that letter to Anthropic.”
opinionconviction 0.7554:56
Once an open-source model is forked and run entirely on US hardware with no data flowing back to China, it effectively stops being a 'Chinese' model from a security standpoint.
“once a model is open-source, it stops being Chinese in a way, right? Because you can now take that model, you can fork it, you can create your own version. You run it in an American data center on your own hardware. There's no packets going back to China.”
opinionconviction 0.855:54
Banning Chinese open-source AI model imports would isolate the US, since the rest of the world will keep using open models, subjecting American enterprises to a costlier 'token tax.'
“if you were to do something like ban open-source in the United States, you'll put the United States on an island. The rest of the world is not going to follow suit.”
opinionconviction 0.6556:53
If US open-source AI models prove better than Chinese ones, the market should decide adoption without an import ban; the US should also weigh restricting Chinese robots given backdoor-security risk.
“if American models, open models are better than Chinese open models, they will win and people will want to adopt them. So if what you're saying is true, that our open models are better or getting better, are going to surpass the Chinese ones, leave that to the market to decide.”
opinionconviction 0.81:01:42
The original purpose of the 14th Amendment was narrowly to secure citizenship for freed slaves' children, not to grant automatic citizenship to children of illegal immigrants.
“the original purpose and understanding of the 14th Amendment was to make sure that the children of freed slaves would have citizenship rights. That was the purpose of it. That was obvious.”
opinionconviction 0.751:08:08
By ruling on birthright citizenship, the Supreme Court removed Congress's space to legislate on immigration edge cases, which is a mistake regardless of the underlying merits.
“what the Supreme Court has done here is remove that space for Congress to legislate.”
opinionconviction 0.71:16:33
The DSA faction's push for open borders and abolishing ICE is an extreme, unpopular position even within the broader spectrum of deportation attitudes.
“the only people who really aren't are these DSA types who want open borders, they want to abolish ICE, they want mass AMC, they are lunatics.”
assertionconviction 0.751:20:12
The DSA's growing electoral wins, even unseating a 15-term Colorado congresswoman, show socialist-aligned candidates are broadly popular within the Democratic Party now, not just a New York phenomenon.
“Colorado, a 15-term congresswoman, 30 years, lists loss to a DSA. So it's not just New York.”
opinionconviction 0.751:37:28
Mega-donors' large-scale philanthropy (e.g., MacKenzie Scott's $26B in gifts to left-leaning causes) hasn't solved the problems it was supposed to solve, undercutting the moral case that billionaires' wealth alone can fix societal problems.
“So why isn't the problem solved now that she just donated $26 billion? That's what they said Elon needed to do to solve world hunger. They always treat your money as if it belongs to them or belongs to the government.”

Jason Calacanis — 8 claims

opinionconviction 0.7525:08
NVIDIA's open-source Nemotron model is now indistinguishable from Claude for the vast majority of everyday searches.
“You will not be able to tell the difference between Jensen Wong's open source, LLM Sacks and Claude for 95% of your searches. I guarantee you.”
opinionconviction 0.7536:19
Job displacement, not net job loss, is the correct frame — some job categories will be retired quickly even as AI creates new ones.
“Certain jobs will be retired and they're going to be retired at a faster rate than you can imagine. But other jobs will happen.”
opinionconviction 0.837:24
Customer support, data entry, business-process-outsourcing, and driving jobs are the obvious low-hanging-fruit categories that will be displaced by AI.
“Customer service jobs are going to go away because consumers prefer to talk to the AI and it's perfect at that. Entry level jobs around data entry, around business process outsourcing, those are going to go away so well, driving cabs and those kinds of jobs.”
assertionconviction 0.839:17
Waymo driver-hiring data in cities where it has hit critical mass is present evidence, not a future prediction, that self-driving displaces cab drivers.
“If you talk to Uber and Waymo in the cities where Waymo is present and has gotten past a couple of hundred cars, they've stopped recruiting drivers, drivers are either static or going down in those markets. So that is absolute evidence that this is happening and the rollout is going to be fast and furious.”
opinionconviction 0.751:07:02
The US has a moral obligation to grant a path to citizenship for long-term (non-criminal) illegal immigrants and their children, since the country effectively waved them in to work.
“We have a moral and ethical obligation to them and to their children. We should give them a path to citizenship unless they're criminals and their children should be Americans.”
opinionconviction 0.751:14:53
A points-based immigration system rewarding English proficiency, cultural knowledge, and economic contribution — as used by Canada, Australia, and New Zealand — would improve US immigration outcomes.
“Most of the great functioning democracies in the world have point-based systems.”
opinionconviction 0.81:21:17
Chasing a short-term socialist-messaging boost will ultimately implode the Democratic Party by losing moderates and swing voters.
“They're going to implode the Democratic Party.”
opinionconviction 0.81:34:55
California and other high-tax-and-spend states deliver far less value per dollar spent on citizen services than lower-cost, more self-reliant states like Texas and Florida.
“you simply do not get two or two and a half times or two point X times the value living in those states in terms of services.”

David Friedberg — 7 claims

assertionconviction 0.815:41
Life-sciences companies are refusing to share proprietary data with Anthropic's new life-sciences model because doing so would commoditize the core asset they've spent billions of dollars building.
“I think nearly everyone I've spoken with has woken up to the fact that they are basically trying to commoditize everyone's business.”
assertionconviction 0.933:50
There is no data-supported 'slippery slope to job loss' from AI; the media won't abandon the job-loss narrative because doing so would destroy their credibility.
“There is no job loss with AI. It is an absolute scam to tell the world that AI is taking away jobs and destroying jobs and the world is shifting.”
opinionconviction 0.743:33
As automation increases, there will be a growing premium for genuine human interaction — bartenders, personal drivers, and masseuses will command higher prices, not be replaced.
“I think that there's a lot of value in human interaction that's going to get a premium in the future.”
opinionconviction 0.71:03:50
Under the intentionalist reading, the 14th Amendment's framers never contemplated mass immigration, but the textualist reading of its plain language is what the Supreme Court's ruling followed.
“If you think about the intention at the time, it was really built around free slaves. There wasn't a consideration at that time of having this massive movement of immigration that was taking place or that has taken place since, that wasn't even in the realm of consciousness or consideration when this amendment was being passed. So the intentionalism argument would be, hey, wait a second, this wasn't really what they intended, but the textual representation is what it is, and that's what the ruling kind of led to in this particular case.”
opinionconviction 0.81:05:49
Birthright citizenship should apply to children of legal residents of the United States, not to children of temporary visitors or people here illegally.
“I do think birthright citizenship should be endowed to the children of legal residents of the United States, whether they're a citizen or not.”
opinionconviction 0.851:13:02
Immigration policy should admit people based on whether they come to work and produce ('makers') versus draw social services ('takers'), regardless of cultural background.
“If the primary motivation for the individual to come to the United States is to work to progress themselves, to become a maker, to create things, to work in such a way that they can generate income, save capital, buy things, advance their family's position”
opinionconviction 0.851:21:53
Newsom's claimed 'balanced' budget is achieved through accounting tricks — borrowing to cover the gap between revenue and expenses — not genuine fiscal balance.
“there's revenue in their spending. When they call it a balanced budget, that's not the way we all talk about it in the business world, where you have like a P&L, you have revenue and expenses. They have revenue and expenses and the expenses exceed the revenue.”

Chamath Palihapitiya — 6 claims

opinionconviction 0.8514:29
Given the risk of frontier labs absorbing and later competing on enterprise proprietary knowledge, it is now derelict and irresponsible for a company not to pursue an independent path to AI.
“there is a very legitimate question, which is, if you are a reasonable company, why are you not finding an independent way to access this intelligence in a way that doesn't leak your edge away? To do so at this point now is becoming derelict and irresponsible.”
opinionconviction 0.824:13
Renting intelligence from the same frontier lab that serves your competitors converges everyone's output — you can't rent judgment from a shared source and expect differentiation.
“You can't rent intelligence from the same place that rents it to your competitor.”
opinionconviction 0.824:42
Renting a shared AI model creates a lowest-common-denominator problem where a company and its competitors converge to the same output, eliminating any reason to pay for it.
“It becomes the lowest common denominator problem where you and your competitors now look exactly the same. Why would you do that?”
opinionconviction 0.7520:30
Given what post-training with harness telemetry could do, self-hosting an open-source Chinese model like GLM entirely on US soil is the obviously correct and much cheaper choice versus trusting a frontier lab.
“why don't I just take GLM, control it entirely soup to nuts on my own hardware inside of the United States with only US citizens that can touch it. It just seems like the brain dead obvious thing to do.”
opinionconviction 0.851:11:00
Western countries are losing their unique cultural identity through unassimilated mass immigration, and immigrants have a responsibility to embrace their host country's values rather than recreate their home country.
“I think if you seed your immigration, you're seeding your culture, and I think that people have a responsibility to keep the values of what made that country unique in the first place.”
opinionconviction 0.751:34:03
Pension holders in California should be angry at Democratic state politics, since a hard fiscal landing that wipes out pension guarantees is coming.
“There needs to be a hard landing. I really hope it doesn't result in the wipeout of these pensions. I think the pensioners should be extremely upset with California democratic politics.”

Alex Karp — 3 claims

assertionconviction 0.851:24
Enterprises get little value from frontier-lab AI while surrendering their proprietary IP and data to those labs.
“the basic view among enterprises in this country is, I'm going to chill wax and waste my time with tokens, I'm going to get no value and they're going to get my IP.”
opinionconviction 0.91:24
The US should not outsource control of AI model weights for government/defense applications to the Silicon Valley consensus.
“Are we really going to outsource the battlefield of this country to the consensus view in Silicon Valley? That is f-ing insane.”
assertionconviction 0.84:08voiced by David Sacks
True AI safety for an enterprise means retaining control over its own compute, models, data stack, and proprietary knowledge rather than ceding it to a model provider.
“what technical customers want is control over their compute, their models, their data stack, and their alpha, meaning their proprietary knowledge. They want to know they own the means of production, he said, and it's not being transferred to someone else.”

QUERY THIS EPISODE

agent access
# full structured record
curl 'https://aipod-two.vercel.app/api/episode/all-in-2026-07-03-ai-sovereignty-wars-palantir-nvidia-deal-scotus-birthright-r'

# or over MCP
aipod_episode { "id": "all-in-2026-07-03-ai-sovereignty-wars-palantir-nvidia-deal-scotus-birthright-r" }