Anthropic's IPO is the honest price test AI needed. Don't chase the pop

What do you do with a company that grew revenue about fourteenfold, lost $80.7 billion in six months and wants to be worth around $2 trillion? You read the prospectus before you read the headlines. The market's first instinct was to do the opposite, and we think that's a mistake.

What the market cheered, and what it skipped

Anthropic filed its S-1 to list on the Nasdaq under the ticker "ANT." As Yahoo Finance reported, the filing shows about $16.3 billion in first-half 2026 revenue. US stocks rose on the morning of the filing, and chip names such as Nvidia and Micron were higher in premarket trading.

Reading that as a verdict on Anthropic would be a stretch. A morning of gains is not a valuation. In our view, the filing gave AI optimism an excuse. It did not give it proof.

The $80.7 billion loss is scarier on paper than in cash, but the bill is real

The headline loss needs context, and in fairness the context helps the company. FinanceFeeds explains that roughly $72.4 billion came from fair-value changes on convertible instruments, essentially a side effect of the valuation rising. The operating loss was $693 million, down from $2.135 billion a year earlier. That is a company moving in the right direction.

Still, the accounting doesn't make the economics easy. Anthropic spent about $8.95 billion on computing and infrastructure in the half, versus $2.33 billion a year earlier. WION's reading of the filing puts cloud commitments at $111.1 billion with Google Cloud, $110.0 billion with Amazon Web Services and $44.9 billion with Microsoft Azure, and estimates the combined obligation across all its deals at $450 billion to $660 billion over a decade. Those are commitments, not revenue, and they have to be paid whether or not the growth continues.

Our view: the question isn't whether Anthropic is a good business. Growth and enterprise traction (more than 1,900 customers generating at least $1 million in trailing 12-month revenue) say it probably is. The question is whether a business that must keep buying compute at this scale deserves the price being floated. Reports mention a $1.8–$2 trillion target, against $965 billion in its last private round. Those targets are reports, not offering terms; the filing sets no price range or share count.

Why this IPO is good for honest pricing

Until now, valuations like this were set in private rounds by investors with strong reasons to believe. A public listing adds skeptics, short sellers and quarterly reports. We think that is healthy, and the timing is brutal in a useful way: CNBC reports that CoreWeave and Nebius lost over 30% in October, Firmus pulled its IPO, and SK Hynix scrapped plans to take its Solidigm unit public. The market looks like it is already separating the AI stories it believes from those it doesn't.

SpaceX is the cautionary example. It raised $85.7 billion in the biggest IPO on record, and its shares finished last week slightly below where they opened in June. Size of offering, in other words, doesn't guarantee a lasting pop.

Governance deserves more attention than it will get. A single Class F share held by a founders' LLC carries 50.1% of general voting power, and the Long-Term Benefit Trust is structured to elect a majority of the board. The company itself warns this may deprive stockholders of a chance to sell at a premium over market prices. Buyers would be paying for growth while accepting almost no say. That can be a fair trade, but it should be priced in.

What we'd do: wait for terms, then wait for a quarter

This is not personal investment advice, and nobody can promise how ANT will trade. But the sequence seems clear to us. First, the price range and share count are still unknown. Second, a pre-IPO contract on Hyperliquid that moved between roughly $1,857 and $1,980 tells you about speculative appetite, not about an exchange-listed price. Third, a reported listing in the week of November 16 is a target, not a certainty.

Retail investors will probably feel pressure to get in on day one. We'd resist it. The first trading session tends to reward allocation and speed; the following quarters reward those who read the compute commitments, the margin trend and the governance fine print. If Anthropic is as strong as its revenue suggests, it will still be strong after the pop fades. If it isn't, the pop is exactly when you don't want to be buying.