The Hype Tax Is About to Come Due

There's a price equity markets eventually charge a story for outrunning its operating numbers. Call it the hype tax. A lot of AI-narrative equities are about to pay it.

The warning got harder to ignore this week. Anthropic's Dario Amodei told an interviewer that some software companies will "completely go bust" if they don't adapt to AI fast enough. Around the same time, Michael Burry, the bubble caller everyone remembers from housing, took a public short position against AI exposure, with the line "sometimes the only winning move is not to play." The two are not friendly to each other's worldview. They're pointing at the same casualty list.

What the forecast actually says

The gap between AI hype and AI delivery is about to start producing visible casualties.

Who pays depends on the lens. Burry's logic implies the casualties are the AI-narrative equities whose valuations got marked up on the story without yet showing the revenue line. Amodei's "go bust" prediction implies the casualties are the legacy software vendors whose moats were workflow friction that AI dissolves.

Those two casualty lists overlap heavily. Companies trading at AI-narrative multiples without AI revenue are often the same companies whose original moat was the workflow friction Amodei is describing. Different framings, same structural pressure underneath.

The discount each warning needs

Burry's record matters, and matters less than the headlines imply. The famous short on housing is what gets him coverage. The record on TIMING bubbles, though, is uneven. He's been early on calls that turned out to be directionally correct. He's also been wrong outright. A famous short position is not the same as a correct call. The reason to take this one seriously isn't his individual record. It's that what he's pointing at lines up with the structural pressure showing up from inside the labs themselves.

Amodei has the opposite credibility problem. His company benefits when incumbent software vendors get disrupted. Take the comment seriously, and discount the framing for the microphone. He's not a neutral observer of which companies should survive the transition. He's an interested party telling investors which companies the transition is coming for. The "go bust" line is, structurally, also a pitch about who ends up buying his lab's products.

The honest read keeps both warnings live and discounts both for incentive. Their evidence is signal; their motives are not pure observation.

Where the hype tax lands first

Strip the abstractions. Which companies actually pay it?

Not the ones that ignored AI. The exposed segment is the ones that bolted a chatbot onto an existing product, declared the strategy complete, and moved on. If you run one of these companies, your investor narrative this year was that you're "AI-enabled." Your earnings transcript next year is going to be asked to show what AI is enabling, in dollars.

The most-exposed sub-segment is mid-market software vendors whose products are essentially structured workflows wrapped in a UI:

• Healthcare-adjacent workflow tools
• Contract-management software
• Mid-market CRM that's mostly process automation
• Category-specific workflow products in finance, legal ops, and supply chain
• Compliance-and-audit tools whose value was the manual checklist

Each of those products' value proposition was, historically, "this turns a multi-step manual process into a guided path." That value proposition shrinks fast when a competitor can ship a single-agent approach that does the multi-step process end-to-end with looser constraints on what the input looks like. The defense isn't shipping an AI feature. It's redesigning the product around what's still differentiated when the workflow itself is automatable.

Where the bull case loses

The bull-case response is that capex stories take three to four years to work through and that any re-rating is overstated for any single earnings cycle. That argument worked when the comparison set was murky. It works less now that the comparison set is sorting, and it works least for the most exposed segment.

Investors carrying AI-narrative companies without AI-revenue lines are about to be asked to defend that position with specifics. CFOs of those companies are about to be asked the same question on earnings calls. Some will defend it with operational data. Others won't, because the strategy was defensible only as long as nobody asked. The asking is what just got harder to dismiss.

If you're a CFO at one of these companies, the time to align your earnings narrative with your operating data was last quarter. The next-best time is right now, before the analyst on the call frames the gap into a quote your stock has to live with for two cycles.

The forecast worth committing to

By Q1 2027, the public-equity AI narrative will have bifurcated into two clearly separable categories. Companies whose AI revenue is large enough to defend their multiple, and companies whose AI line is still aspirational. The second category will have re-rated downward by a meaningful amount, and several won't survive the re-rating intact.

That's a position you can disagree with. The standard counter is that re-rating cycles take years, not quarters. What makes that counter wrong now is that the pressure isn't coming from one camp. A famous short and a frontier-lab CEO are pointing at the same outcome at the same time, with very different reasons to want investors to hear them. That's the textbook tell that the pressure is structural, not cyclical.

The hype tax doesn't honor your roadmap. It honors your numbers.