Futures
Access hundreds of perpetual contracts
CFD
Gold
One platform for global traditional assets
Options
Hot
Trade European-style vanilla options
Unified Account
Maximize your capital efficiency
Demo Trading
Introduction to Futures Trading
Learn the basics of futures trading
Futures Events
Join events to earn rewards
Demo Trading
Use virtual funds to practice risk-free trading
CFD
U.S. stock CFD derivatives
US Stocks
Access real US stocks and ETFs
HK Stocks
Trade quality Hong Kong-listed stocks
Korean Stocks
SK Hynix
Real Korean stocks and top assets
Stock Futures
High leverage, 24/7 trading
Tokenized Stocks
Backed by real stock assets
IPO Access
Unlock full access to global stock IPOs
GUSD
Mint GUSD for Treasury RWA yields
Stocks Activities
Trade Popular Stocks and Unlock Generous Airdrops
Launch
CandyDrop
Collect candies to earn airdrops
Launchpool
Quick staking, earn potential new tokens
HODLer Airdrop
Hold GT and get massive airdrops for free
IPO Access
Unlock full access to global stock IPOs
Alpha Points
Trade on-chain assets and earn airdrops
Futures Points
Earn futures points and claim airdrop rewards
Promotions
AI
Gate AI
Your all-in-one conversational AI partner
Gate AI Bot
Use Gate AI directly in your social App
GateClaw
Gate Blue Lobster, ready to go
Gate for AI Agent
AI infrastructure, Gate MCP, Skills, and CLI
Gate Skills Hub
10K+ Skills
From office tasks to trading, the all-in-one skill hub makes AI even more useful.
Palantir CEO Karp slams "Token Economy" as deeply flawed, but he's still selling patriotic gold shovels.
Palantir CEO Alex Karp took aim at CNBC on July 1, saying the token billing model of OpenAI and Anthropic is “completely wrong,” relaying complaints from corporate clients that paying for tokens does not buy them any value. But Palantir also sells something else that’s hard to verify—by leaning on national security contracts and pie-in-the-sky valuations.
(Background: Michael Burry of The Big Short said Palantir is just a low-margin SaaS outsourcer! Anthropic is eating away at it.)
(Additional context: Even Meta has started telling its 6,000 employees to use AI sparingly; usage volume does not equal effective output.)
Table of Contents
Toggle
Key Takeaways
On July 1, Alex Karp (Alex Karp), CEO of U.S. national security and data analytics software company Palantir, appeared on CNBC (the U.S. business TV network)’s flagship morning program Squawk Box, firing at OpenAI and Anthropic—two of the highest-valued AI labs in the world right now.
He didn’t name names or directly scold anyone, but that line—“something has already completely gone wrong”—was hard for Silicon Valley as a whole to pretend it didn’t hear. I watched the clip twice, and the more I watched, the more it felt like this was not just background chatter between AI companies, but a direct jab at how the AI industry defines “what it has produced” by “how much it burned,” and at who has the standing to step forward and expose it.
What Is Completely Wrong
That day on camera, Karp said he’d been hearing corporate clients complain, and he relayed the exact words as follows.
The sharpest accusation here is not Karp himself running up huge token bills for Palantir. It’s the fact that he is relaying the clients’ feelings (and this framing matters a lot—Karp is smart enough to put the most cutting words into someone else’s mouth, keeping himself at a safe distance).
The second layer is his own judgment. He says what enterprises truly want is “control over their own compute, models, data stack, and alpha,” and that “they want to own the means of production.” He even helped spell out the questions buyers should ask model vendors.
Karp’s point is that a token-based pricing business model is, in essence, using “how much you burn” to replace “what you’ve done.” Token is the smallest unit that AI service providers charge for based on usage. Every time the model processes a piece of text, it gets billed once. In theory, the more you use it, the more you rely on this tool—and reliance should equal value. But the problem is that burning more doesn’t necessarily mean you get more in return. Inside Silicon Valley, there’s even a term for this mindset: tokenmaxxing—treating the act of driving up token usage itself as the goal, rather than treating it as a means to achieve an actual objective.
The Illusion of Consumption
This logic trap won’t be unfamiliar to readers of “dong chain” (the crypto scene). In the last bull market, the whole industry also used consumption to stand in for value—on-chain transaction volume, how much Gas was burned, how high Total Value Locked (TVL) stacked up. Those numbers were all used as proof of whether a project “has value.”
The numbers look beautiful, and the token price looks even better. But burning more doesn’t mean you actually built something. TVL can be stacked by swapping value between left and right hands. Transaction volume can be boosted with wash trading. If Gas fees are burned aggressively, it doesn’t necessarily mean the underlying technology actually solved any problem—or that the project is truly useful.
Now the AI industry is facing the same issue. Token usage keeps climbing. Cloud service providers’ financial reports look great. But corporate clients are starting to pull out their bills and ask the most basic question: What did I actually get in exchange?
The Keys to the Token
Karp’s proposed solution is that enterprises should control their own compute, models, and data stack—not outsource everything to model vendors and pay them based on usage.
This is not just talk. By the end of June, Palantir had just announced an expanded partnership with Nvidia to deploy open-weight Nemotron models for the U.S. government in a “sovereign environment.” In plain terms, it means moving the model weights onto its own ground and running them there—without handing the models off to others. In the business world, companies really are re-evaluating the data autonomy that comes with building their own environments.
Palantir Is Also Selling Shovels
But once the story gets to this point, Karp—standing in front of the TV camera—has exposed the illusion others build by stacking consumption volume. If you think about it carefully, however, what Palantir sells is also based on a narrative that’s difficult to verify. That narrative is called sovereignty, it’s called national security, and it’s called pie-in-the-sky valuation.
Back in April this year, Karp floated his famous “22-point manifesto,” arguing that Silicon Valley should not only build apps, but should also invest in making weapons—and even calling for the U.S. to reinstate conscription.
Palantir’s quarterly results this season are indeed strong. For Q1 2026, revenue was $1.63 billion, up by about 85% year over year, and U.S. commercial revenue surged by 133% year over year.
Michael Burry (Michael Burry), the famous “Big Short” investor, poured cold water on the company in April this year as well. He said Palantir, in essence, is a low-margin SaaS outsourcing company, and Anthropic is eating up its business. Put Burry’s doubts about Karp alongside Karp’s doubts about OpenAI and Anthropic, and the vibe becomes clear.
Everyone feels like they’re selling the real thing, while everyone else is selling illusions.
Frequently Asked Questions
When Karp says “tokens create no value,” who is he accusing?
This statement is Karp relaying corporate client complaints, not a direct accusation of Palantir’s own invoices. The criticism is aimed at token-billing model service providers like OpenAI and Anthropic.
Will open-weight models replace the token billing model?
When enterprises deploy open-weight models on their own, they can save on volume-based costs and retain data control. However, they still need to build their own compute and operational capabilities. So this is a matter of diverging paths, not a full replacement right now.