The AI Model Labs Have Been De-Facto Nationalized
Today’s Letter is brought to you by Arch Public!
Unlock unparalleled returns with Arch Public’s algorithmic trading tools. Our Bitcoin Algorithm Arbitrage Strategy has delivered an astounding 247% annual return over the past three years.
The entries, and exits speak for themselves; precision that drives success. Trusted by more than 15,000 customers and industry leaders, we’ve partnered with Gemini, Kraken, Coinbase and Robinhood to bring you cutting-edge solutions.
Whether you’re a seasoned investor or just starting, our proven strategies maximize your potential. Join the ranks of those who trust Arch Public to navigate the markets with confidence.
Talk to us today and discover why our expertise sets us apart.
To investors,
All hell is about to break loose in the American AI model labs. They have previously enjoyed a dominate position in the market that allowed them to rent superhuman intelligence to billions of people and millions of businesses.
This is why Anthropic is rumored to be doing more than $50 billion in annual revenue with 15-20% free cash flow margin. You don’t accidentally build a company of that scale without having an oligopolistic position in the market. They were able to achieve this growth while OpenAI was growing to tens of billions in revenue, along with the stiff competition from Google’s Gemini, xAI’s Grok, and a few others.
But now there is a much, much bigger threat on the playing field: Chinese open-source models.
To understand what is happening, we first have to realize the problem that most companies have been facing over the last few months. Here is what I tweeted on May 22nd this year:
“There is a regime shift happening in how companies use AI models right now. Over the last year, founders and executives heavily encouraged their teams to adopt the technology. They created token leaderboards and pushed employees to get AI-pilled.
But then the expenses started to hit company financials.
Now executives are working to increase token efficiency. They still want their teams using AI models, but they are finding ways to cut overall token usage. The goal is no longer “use as many tokens as possible,” but rather “be as productive per token as possible.”
It may seem like a subtle change. It is not. Teams are re-writing their internal and external software to cut token usage, while increasing usage & productivity.
I see it inside our companies on a daily basis. For example, we have cut overall token usage at CFO Silvia but we have continued to grow the user base, assets, and total output.
This will be a major theme over the next 12 months.”
A month after that tweet, it seems that I drastically underestimated how fast and severe the pivot would be. Not only are companies looking to reduce token costs by changing their workflows or rewriting how their software uses American AI models, but now some companies are abandoning the closed-source models for a large portion of their work.
Rohan Paul writes:
“UBS says 60% of companies now watching AI budgets are moving to cheaper models and open-source Chinese models
The pressure is coming from extreme bills, including users spending up to $35K/month, teams exceeding quotas by 200%, and companies cutting internal AI tools from 5 to 2.
Companies are not abandoning AI, they are using model routing, which sends easy tasks to cheaper models and saves premium models for hard reasoning, code, and long-context work.
Chinese open-source models such as Qwen, DeepSeek, MiniMax, GLM, and Kimi now fit the enterprise cost curve because they can be run locally or used through cloud catalogs.”
So what do you do if you are an American AI model lab that is watching your customers start adopting Chinese open-source models? Well, you try to stop the exodus. One path is to out compete these open-source models, but you have already been working on that with your internal technology teams. The other path is to look for regulatory protection that stops or slows the opposition.
Bill Gurley points out that Anthropic appears to have engaged Option #2 over the last few weeks. He posted the following in a series of tweets:
“These overly aggressive grabs at regulatory capture come at the exact time as price rationalization & optimization are pushing partners and customers towards other solutions (as the example shows). Hope the government knows they are being manipulated…This is what’s causing Anthropic to aggressively beg for government protection (Rohan’s tweet above). Customers are finding cheaper alternatives. Keeping employees requires continuing ultra-rich secondaries ($$$) that are dependent on revenue growth. When you can’t win on the field go to DC…Anthropic could have sued in court (as others have). But they want something far more valuable than simple restitution. They want the US government’s “protection against competition” for years and years. A court can’t provide that.”
These are strong words coming from Gurley. There are always two sides to a story though. The counter-argument comes from Brad Gerstner when he wrote:
“Fable is currently export controlled & rumors are that 5.6 will also be subject to an approval framework. Whatever jiu jitsu the Chinese are using to get us slow down our own frontier models while letting their models run free appears to be working. Who is capturing who?”
As I said, all hell is about to break loose inside these large language models. They have essentially agreed to the US government demand to allow the latest models to get government approval. That is de-facto national control even if the large AI labs don’t realize it yet.
So the question becomes what are the ramifications of this development?
The most thorough analysis I have heard comes from Box CEO Aaron Levie. He writes:
“We now have de facto AI regulation. It’s not obvious why from here on out models that have certain levels of capability or are trained on certain compute sizes won’t have to be reviewed by the government before release.
Realistically, as AI models became more and more powerful this was going to be inevitable (I think it’s too early, but here we are). So now it’s mostly just interesting to think about the implications and scenarios from here. A few would be:
* America gets to control who gets access to frontier intelligence and when. This generally works as long as we remain at the frontier at all times and don’t have a risk of being surpassed. At the moment we have a clear lead in frontier intelligence so this is a good bet, but lots of motivated parties would love to change that.
* This likely creates backlog of AI releases which means that we will see less rapid fire back and forth jumps in model progress. Bull/fine case is that we just get bigger step functions per release at a slower rate and we end up at the same point we would have. Bear case is those incremental smaller jumps were necessary for the continued flywheel of innovation.
* Other countries likely have even more incentive to at least hedge their bets with sovereign AI strategies so aren’t dependent on access to US AI all times. Previously this was relatively moot because the alternative wasn’t good enough, but that could change out of necessity and what we’re seeing in China.
* Open weights obviously a big winner here as it becomes what likely sovereign AI gets built out on, and what (for now) can still be released to the market without the same controls. One interesting question would be how regulation eventually extends to open models, which would have its own set of long term consequences.
Anyway some big updates to everyone’s mental models of AI regulation as a result of the capabilities we’re now seeing in AI. Wild times.”
Wild times indeed. The AI industry is only going to grow in national importance, which means politicians and regulators are going to have a field day fighting for more control over time. That changes the playing field significantly and it gives the open-source models a big boost in their fight for adoption.
People and businesses want cheaper access to superhuman intelligence. The US government wants more control over the technology. And investors are looking for higher and higher rates of growth.
You simply can’t get all three of those at the same time, so at least one group is going to be disappointed. I will let you decide for yourself who will be the winners and losers.
Hope everyone has a great end to their week. I will talk to you on Monday.
- Anthony J. Pompliano
Founder & CEO, ProCap Financial (Nasdaq: BRR)
🚨 BloFin: Built for Traders Who Want an Edge
BloFin is a fast-growing cryptocurrency exchange built for serious crypto and TradFi traders.
Trade 500+ spot pairs and hundreds of futures and perpetual pairs across crypto and TradFi, plus stock indices and commodities like oil and gold — all with deep liquidity, professional-grade tools, and zero internal market makers, so you're never up against the house.
To celebrate their partnership with us, BloFin is giving away $100,000 in Deposit & Trade Rewards based on your activity during the campaign.1
The K Shaped Economy Continues To Widen
Darius Dale is the founder and CEO of 42 Macro, a leading investment research firm.
In this conversation, we break down what Kevin Warsh's Fed chairmanship means for monetary policy, the K-shaped economy creating historic affordability pain for millions of Americans, and why every investor must participate in asset markets to survive financial repression.
Podcast Sponsors
Figure – True DeFi Democratized Prime to earn ~9% APY! They also have the lowest industry interest rates at 8.91% with 12 month terms! Take out a Bitcoin Backed Loan today and buy more Bitcoin. Check out Figure! Figure Lending LLC dba Figure. Equal Opportunity Lender. NMLS 1717824. Terms and conditions apply.
Arch Public - Arch Public’s cutting-edge algorithmic tools ignite profits, harnessing razor-sharp data analytics to nail perfect entries, exits, and risk management. Turn volatility into opportunity and do it hands free with Arch Public. (Oh, and yes, try us out for FREE too!)
mogul — Invest in tokenized residential real estate with targeted yields, monthly rent payouts, and no landlord headaches. Learn more at mogul.club/pomp.
BitcoinIRA - Save up to 37% in capital gains taxes on your retirement investments. Signup today and win up to $4,000 in rewards.
Plaud - Plaud builds AI-powered wearable devices designed to help people capture, organize, and recall important information from real-world conversations and moments.
Simple Mining offers a premium white-glove Bitcoin mining service. Want to grow your Bitcoin stack? Visit https://www.simplemining.io/pomp
Uphold - Uphold is the all-in-one platform to trade, earn, stake, and swap across 300+ assets with real-time proof-of-reserves and any-to-any conversions. Manage your entire crypto portfolio in one place at www.uphold.com
Bitget - Bitget is the world’s largest Universal Exchange (UEX). Trade crypto, gold, forex, stocks and more on one platform. 24/7 access, deep liquidity, low fees, up to 500x leverage. Join Bitget today.
Award-winning Fountain Life - Energy supercharged. Memory sharper. Life extended. Ready for the best investment you’ll ever make? Schedule a life-changing call at www.FountainLife.com
BloFin - BloFin is a fast-growing cryptocurrency exchange focused on providing professional-grade trading tools, deep liquidity, and a secure trading environment for crypto traders worldwide.
You are receiving The Pomp Letter because you either signed up or you attended one of the events that I spoke at. Feel free to unsubscribe if you aren’t finding this valuable. Nothing in this email is intended to serve as financial advice. Do your own research.
Terms and conditions apply.*
Great read Pomp. Couldn’t this move the needle a lot where many big corps do the Lilly model and set up their own private “Data center” where they buy the racks from Nvidia and Dell and use their own proprietary data and bypass the government completely. This would open up another cap ex channel for all the “whole rack” sellers and solve the government determining models, advancements in tech etc.
Anyone running "free" models from the PRC does so at extremely high risk.