inside the tech debate over burry saying neither anthropic nor spacex is worth $1t

michael burry says neither spacex nor anthropic is worth $1t

the hacker news thread runs deeper than the headline – the real argument is over whether the trillion-dollar number can be justified at all, and it breaks into three camps:

agree with burry – $1t is too high:

• no real moat. an ai lab is only as good as its latest model, and that lead is temporary. deepseek v4 already runs at roughly $1/day vs frontier pricing, openrouter serves open-source models cheaply, and chips with models burned directly into them (taalas) are coming. once the model itself is commoditized, the pricing power goes with it
• the fundamentals don't add up. anthropic's actual 2025 revenue is around $4.5b, not the $50b figure people throw around. its first profitable quarter was q2 2026 at ~$500m – on a $1t valuation that's a 0.2% return, meaning earnings would need to grow something like 100x to justify the price
• consumer subscriptions are probably sold at a loss to grab market share. the api is the only price that actually covers cost, so the "everyone pays $50/mo" math is shakier than it looks
• coinbase is the cautionary tale: it ipo'd 5 years ago and has mostly traded below its ipo price since. late public buyers got the worst of it
• the whole market looks detached from earnings anyway – tesla sits at a p/e of 380 and still won't fall. if that can hold, valuations aren't tied to fundamentals right now
• and a reflexive trap: a $1t valuation only makes sense if ai replaces most human labor. but if it actually does that, the entire jobs-and-equities economy breaks down, so the bet undermines itself

disagree with burry – it's worth it, or he's wrong to bet against it:

• the strongest point: insiders are voting with their balance sheets. meta, google, amazon and microsoft are betting their companies on ai – google alone reportedly committed ~$80b to hardware, with over $1t cumulative across the group. these are the people with the real internal numbers, and they clearly don't agree with burry
• the bull math isn't crazy: 650m–1b knowledge workers globally, at ~$50/mo, is roughly $180b/year in potential revenue if adoption keeps climbing
• enterprise lock-in is a moat – integrations and accumulated conversation history make switching painful. (the rebuttal: that stickiness works for a $20–40/mo saas seat, not for customers burning three-to-four figures a day on api calls, who'll happily chase a cheaper equal model)
• currency debasement: if the dollar keeps losing value, $1t could end up "real" in nominal terms. central banks have repeatedly chosen asset inflation over deflation
• on burry himself: his critics say he's usually right but far too early. he posted "no idea how short i am" in nov 2022, then "sell" in jan 2023 – right near the market bottom, before a huge rally. he shorted tesla and nvidia at bad moments and eventually closed his fund. "markets can stay irrational longer than you can stay solvent" (that line is gary shilling's, not keynes)
• and the purely dismissive take: "bro was right once" – meaning the 2008 subprime call and not much since

neither camp – worried about the plumbing, not the valuation:

• this is the part most hn engineers actually care about: forced buyers. when a giant company lists, index funds have to buy it at whatever price it opens at, with no judgment about whether it's overvalued. and once it's inside the index, it can't be punished by the market the normal way, because tanking it would gut everyone's pension and 401k at once
• nasdaq recently changed its rules so a huge ipo can enter the index in about 15 days. the stated reason is that companies now stay private far longer and show up already mega-cap, so the old waiting period looks unrepresentative
• matt levine's spacex "pump" sketch: ipo, sell a small 5% slice to price-insensitive fans at a sky-high valuation, get into the indexes, then unlock more stock the funds are obligated to absorb. important caveat – levine explicitly called this the maximally cynical version and said it is not what spacex is doing and "not actually possible"

two accuracy notes so this doesn't get misquoted: the spacex "pump" is a disclaimed thought experiment, not an accusation. and the claim that the s&p is dropping its profitability requirement is unconfirmed – one commenter flags it as financial-influencer exaggeration.

conclusion

the headline is burry vs the trillion-dollar valuation, but the thread quietly agrees on more than it lets on

the bears doubt the fundamentals, the bulls trust the insiders and the adoption curve – and both sides are arguing about a number

the sharpest people in the thread skipped that fight entirely. their worry isn't whether anthropic or spacex is "really" worth $1t

it's that with index funds as forced buyers and the gates to those indexes getting wider, the price barely has to be right for the money to pour in anyway

that's the part that should make you uncomfortable regardless of which camp you're in: not that the valuation might be wrong, but that the market may have stopped needing it to be right