Why Are AI Crypto Tokens Down? Anthropic’s S-1 Filing Exposes an $80.7 Billion Loss

AI crypto tokens are sliding on October 26 for a reason that has nothing to do with their own networks: Anthropic publicly filed its S-1 with the U.S. Securities and Exchange Commission, and the prospectus put a hard number on the cost of the AI trade — an $80.7 billion net loss in the first half of 2026. Bittensor (TAO) is down 1.1% to $250.50, NEAR is down 1.5% to $4.48, and Fetch.ai (FET) is off 0.5% to $0.174, while Bitcoin sits nearly flat near $80,000. The filing changes no AI token’s code, but it sharpens the question traders have been asking all month: if the biggest AI lab loses $80.7 billion in six months, what is an AI token actually worth?

This is the second time in a week that AI-spending worries have shown up in crypto. On October 22 we flagged the same pressure in Bitcoin’s slide below $82,000 on AI-spending fears. Today the fear rotated out of the macro picture and directly onto the coins that trade as a proxy for artificial intelligence.

What the S-1 Actually Says

Anthropic’s own announcement is short and precise. “Today, Anthropic, PBC filed a registration statement on Form S-1 with the U.S. Securities and Exchange Commission relating to its proposed initial public offering of its Class A common stock,” the company wrote, adding that “Anthropic intends to list its Class A common stock on the Nasdaq Global Select Market under the symbol ‘ANT.'” The filing is led by Goldman Sachs and Morgan Stanley at the head of a 22-bank syndicate.

What the announcement does not contain is the price. “The number of shares of Class A common stock to be offered and the price range for the proposed offering have not yet been determined,” Anthropic said — which means every valuation circulating today is a press estimate, not a filing number. Reports have floated a range of roughly $1.5 trillion to $1.75 trillion, down from earlier talk of $2 trillion, against a private-market valuation of about $965 billion in May.

The S-1 itself carries the numbers that matter. Revenue reached $16.29 billion in the first half of 2026, up from $1.16 billion a year earlier — more than a thirteenfold jump. The company booked an $80.7 billion net loss over the same six months, against a $12.9 billion loss in the first half of 2025. Crucially, about $72.4 billion of that loss — roughly 89.6% — came from non-cash fair-value changes on convertible notes, derivative liabilities and SAFEs, the paper re-pricing of instruments tied to Anthropic’s rising private valuation. The operating loss was closer to $693 million.

The context that gets lost in the headline: Anthropic brought in more revenue in the second quarter alone — $11.56 billion — than it did across all of 2023, 2024 and 2025 combined. That is a real growth curve. The cost side is also real, with computing and infrastructure spending approaching $9 billion in the first half.

The AI Token Tape, Coin by Coin

The 24-hour move is a clean read on how the market is pricing the Anthropic news, pulled from Binance at roughly 21:45 UTC on October 26. Bittensor’s TAO led the declines among the majors at -1.11% to $250.50. NEAR followed at -1.52% to $4.48. Virtuals Protocol (VIRTUAL) fell 1.41% to $0.614. Fetch.ai slipped 0.46% to $0.174, and Render (RNDR) was the lone green print at +0.12% to $1.72. Against that, Bitcoin was off just 0.42% near $80,132 and Ether 1.03% near $2,317 — so the AI names are underperforming the two benchmark assets on the day.

The one-day move understates the damage. Bittensor and NEAR are both down roughly 20% over the past two weeks while Bitcoin fell about 4%, a divergence that maps almost exactly onto the cooling AI narrative. TAO sits about 67% below its $757.60 all-time high, meaning the token would need to roughly triple to reclaim its 2024 peak.

The Arithmetic That Should Scare AI-Token Holders

Do the market-cap math and the gap becomes uncomfortable. Bittensor’s market capitalization is around $3 billion; NEAR’s is around $6.3 billion. Combined, the two largest AI tokens are worth a little over $9 billion. Anthropic’s $80.7 billion half-year loss alone is nearly nine times that combined figure — and roughly 27 times Bittensor’s entire market cap. Even if you strip out the $72.4 billion of non-cash fair-value charges and use the ~$693 million operating loss, the point stands: a single AI lab can burn, in real infrastructure spend, an amount that approaches the entire investable AI-token sector.

That is the structural problem the AI-coin trade keeps running into, and it is the same warning Arthur Hayes issued in June when he argued a peak in the AI trade could sink crypto markets. The capital that wants exposure to AI can now buy the equity — a registered, audited business with $16.29 billion in six-month revenue — rather than a token whose value rests on a decentralized-compute thesis that is still being proven.

The Non-Cash Illusion, and Why It Matters

It would be wrong to read the $80.7 billion figure as cash leaving the door. The bulk of it is accounting: as Anthropic’s private valuation climbed, the fair value of the convertible notes and SAFEs on its books re-priced upward, and that re-pricing is recognized as a loss. The company did not write $80.7 billion in checks to chipmakers. The ~$693 million operating loss is the closer proxy for how the core business actually performed in the half.

But the non-cash nature of the loss cuts both ways. If Anthropic’s valuation comes in below the levels embedded in those instruments, the same mark-to-market logic can swing the other direction. For AI tokens, the takeaway is simpler: the AI economy is now public-market-visible, and its headline numbers — even the non-cash ones — will drive sentiment across the entire AI complex, crypto included.

What to Watch Next

Three things resolve the near-term setup for AI tokens. The first is the Anthropic pricing: once the S-1 gains effectiveness and a range is set, the market gets a real anchor instead of press estimates. The second is the token reaction to support — TAO’s $245 level has been flagged in market commentary as the line between a healthy retest and a deeper slide toward its September low near $215, with the 50-day moving average near $273 acting as the ceiling. The third is whether the AI tokens’ underperformance against Bitcoin persists, which would confirm a rotation out of the sector rather than a broad-market dip.

The countercase for AI tokens is equally concrete: a $1,000 position in TAO at $250.50 buys 3.99 TAO, and a return to the $757.60 all-time high would turn that into about $3,023. The decentralized-compute and AI-agent narratives that lifted these tokens through 2025 have not been invalidated — an IPO filing in the equity world does not make on-chain inference less real. What it does is raise the bar: AI tokens now compete for the AI narrative with a public company whose growth numbers are laid out in a federal filing, and until the tokens can point to revenue of their own, they will trade as the higher-beta, higher-risk expression of the same theme.

Anthropic’s own boilerplate says it plainly: “These securities may not be sold, nor may offers to buy be accepted, prior to the time the registration statement becomes effective.” The AI equity trade has not even opened yet. How AI tokens hold up between now and that debut will tell you more about the sector than any single day’s price move.

Frequently Asked Questions

Data: Binance (AI token and BTC/ETH prices, Oct 26, 2026), Anthropic S-1 filing via SEC EDGAR.