If you own crypto I suspect you’re about to endure a shit-ton of FUD as the labs release cryptography breakthroughs at a record pace over the next few months.
They’ve obviously already made these breakthroughs internally and are first advising the relevant authorities as evidenced by the conspicuous avoidance of the topic in any of the recent math releases.
Why is the overall market so fixated on $CRWV's imminent demise?
At the peak of the Dot Com bubble, $CSCO - the largest company in the world, and the $NVDA equivalent of the era - generated about 10% of its revenue from circular deals through vendor-financed startups.
When those startups lost access to capital, $CSCO's revenue growth went from +66% to -30% within a few quarters, and the stock lost 80% of its value. The same story played out across the dot com supply chain, sending the Nasdaq down 80%.
Today, $CRWV alone makes up roughly 8% of
$NVDA GPU sales. Losing that demand is enough to nuke $NVDA's growth story all by itself.
Oh, and $CRWV isn't an isolated example - every major neocloud is hemmoraghing cash and borrowing costs are going UP across the board.
If you lose the neoclouds, you lose the incremental growth driver for marginal data center capacity, and the AI capex trade goes up in smoke.
That's why we're seeing carnage across the entire AI supply chain, from $CAT and $GEV to $MU, $AVGO and more recently $NVDA.
This is what the beginning of the end of a debt-fueled, circular-financed capex bubble looks like.
these $META shenanigans bc of muse/ personal AI assistants are getting out of hand, so thought helpful to ground this discussion in some first principles thinking
way way too aggressive of a forecast, but let's entertain the following for the sake of the argument
2030, global
-> ecom $5tn x 30% agent share x 3% take rate = $45bn rev pool
-> 4bn users x 10% paying x $20/m = $96bn rev pool
-> total ~$140bn
this will be a market with perfect competition and extremely low switching costs (and massively depend on cooperation from large merchants like $AMZN, but let's also ignore that for a moment). I assume 5+ players
-> $GOOG, $SPCX, at least one of OAI and Anthropic (if they don't go bankrupt)
-> $AAPL offering for free and private, run locally on hardware (which is also the biggest bear case, but let's ignore that as well)
-> tbh only maybe $META bc people don't really trust them today (and I don't see how they'd change that)
-> global/ regional third party providers/ protocols like instinct/ openclaw (at the end of the day if you believe in AI, this is just a compilation of scripts and maybe memory architecture ie there's really no MOAT here)
so let's divide the pool by 5 to be generous = that's ~$30bn rev by 2030 in the blue sky bull case
for reference, consensus 2030 rev for $META is $470bn (agents cannibalize their own ad spend, which is why I haven't added ad rev on top) and consensus 27 CAPEX alone is $200bn
the math ain't mathing
we enter the part of the cycle where AI nukes and tech HFs will be too slow to derisk their book and too full of themselves to pivot into crypto
(and yes, they’ll also be too slow to short AI bc “wen AGI bro”)
the californication of tech public market investing won’t end well