Yeah, we're taking in half the money that we're burning, but trust us, we're worth billions. Now just go buy our pyramid scheme stock or one of the other AI pyramid scheme stocks and STFU about "sane business plans."
Anthropic reveals $30bn run rate and plans to use 3.5GW of new Google AI chips
Broadcom has announced that Google has asked it to build next-generation AI and datacenter networking chips, and that Anthropic plans to consume 3.5GW worth of the accelerators it delivers to the ads and search giant. News of the two deals emerged today in a Broadcom regulatory filing that opens with two items of news. One is …
COMMENTS
-
-
Tuesday 7th April 2026 10:42 GMT simonlb
And when this bubble
burstsexplodes you can virtually guarantee that somehow, all the trillions of dollars being thrown at AI will suddenly become 'our'problemdebt and we will all be hit way worse than the 2008 financial crisis by orders of magnitude. Except the tech bros, they'll somehow still be 'earning' billions every year for 'reasons'.-
-
Tuesday 7th April 2026 18:58 GMT Guido Esperanto
The bubble will appear self-sustaining because it's not yet had a real test.
In the dotcom bubble, this appears to be a series of interest rate rises in the US AND at the same time Japan entered a recession....the rest we know is history.
Now it's very interesting how the techbros are propping up the US Markets and Trump is pressuring the fed reserve to drop interest rates, or drop Jerome Powell, but currently Jerome Powell is staying put and stable on interest rates.
The current "*special* military floperation" being undertaken in Iran and its full ramifications are not yet fully known.
It would be quite the implosion if that kickstarts the AI sell off
-
Wednesday 8th April 2026 09:45 GMT LionelB
> The bubble will appear self-sustaining because it's not yet had a real test.
You may well be right – time will tell.
In the longer term, though, AI ain't going away (in the same way that the WWW did not go away post-dotcom crash). It will be interesting to see how that plays out. For all the slop, nonsense and obscene resource demands, is easy to lose sight of just how far (and how fast) AI has come over the last, say, decade and a half. For example, we take (almost) seamless translation and image/facial/voice recognition for granted now, forgetting just how rubbish it was in not-so-distant memory, while the capabilities of current generative AI would have seemed like (dystopian?) sci-fi a mere decade ago. Furthermore, the slop, nonsense and hard-sell in the business/personal/social sectors also distracts from real potential of (non-LLM) AI in the tech, physics, industrial and biomedical arenas.
-
-
-
-
-
Tuesday 7th April 2026 09:36 GMT cyberdemon
Gigawatts
I've built your 3.5 Gigawatt datacentre. It's a giant electric fan-heater which can produce as much hot air as 35 Rolls-Royce Trent engines at full throttle (~100MW each) and requires three large nuclear power plants to run, which I assume you will be constructing.
What do you mean "breach of contract"? The contract only mentioned "3.5 Gigawatts" as the specification
-
Tuesday 7th April 2026 15:42 GMT JacobZ
Lies, damned lies, and annualized run rates.
Anthropic says lots of things in unaccountable press releases. When its CFO Krishna Rao made a court filing under oath, he stated that revenue "has exceeded $5 billion to date.”*
So we are supposed to believe that a company with $5B lifetime revenue had an ARR of $9B in 2025, $12B mid-February, and now $30B.
Also, I'll believe that Anthropic is serious about buying those chips when Google files a quarterly financial report that shows a corresponding bump in their revenue forecast.
----
*This was in their lawsuit with the Pentagon. Filing can be found at https://storage.courtlistener.com/recap/gov.uscourts.cand.465515/gov.uscourts.cand.465515.6.5.pdf