They still don't get it. It is not the science and discovery of new drugs that are the road block. Its the 5 years it takes to run clincal trials to prove it works and its safe for humans. It will always take 12 months just to get FDA approval. I know of at least 5 to 10 drugs developed by tech bio companies over the last 5 years that all failed in clinic.
@TangoJazz99@Raman_Macro@chontang They are hiding costs in contracts with google and amazon, read Ant s1 carefully. They have about the same revenue as openai in H1, and openai lost 21.6 billion.
To be fair, this is what us accountants call "the oldest trick in the book".
The great thing about learning how to do accounting properly, instead of just asking Claude what it thinks, is that you've known for months/years that AI folk have been lying to us all...
@iLibertarian@dantrappx@Yuchenj_UW Not really, plenty of small biofirms have less than 5 people. For instance Actithera is also a one person biotech established 4 years ago. And even some public traded biotech firms have less than 10, KALA bio, for instance.
Well there it is. Anthropic justifying the 3-4x rise in Opex after $GOOG and $AMZN subsidy ends.
*finger quotes* We expect higher compute needs and have switched to short-term compute that will impact our Operating Margins going forward in the near term.
It costs about 4-5x as much to do short term compute than long. Assuming even if just 20% added short term compute (200 MW) is needed a quarter without new compute online, it would immediately increase Anthropic OpEx from $4-$5 billion / GW to $15-$20 billion / GW in a single quarter. This is roughly in line with the true cost of compute.
$16 billion x 4 = $64 billion
200 MW = $13 billion
1 GW ($4-$5 billion subsidized) + $13 billion 200 MW short term = $18 billion, roughly in line with $16 billion / GW true cost
Anthropic could be using “short term compute” or just entering a new contracted price with $GOOG and $AMZN and you wouldn’t be able to tell the difference.
It gets worse. Expect Anthropic margins to “meaningfully compress” even as our business improves.
“We expect compute and infrastructure expenses to increase in absolute dollars and remain our largest operating expense for the foreseeable future”
“At our current stage, this cycle is an inherent aspect of our business and will cause operating margins to fluctuate from period to period, at times meaningfully, even if the underlying economics of the business continue to improve.”
@TMTLongShort Yeah! I agree with you 100%! You really speak the quiet part loud Mr Pod. You should organize an initiative to the White House calling for a ban on all countries other than the United States using OpenAI and Anthropic models immediately!👍🏻💪🏻🇺🇸
@TMTLongShort@d_savickij@LukeGromen LOL, you didnt know that openai hired a lot of blender artists to label data for model training this May just to enable astra to make 3d models? You think AI can just learn things by themselves without real data?😂
One of the more under-appreciated details in Anthropic's S-1:
Nearly half of Anthropic's revenue now flows through cloud marketplaces - 48% in the first half of 2026, up from 11% in 2023. That's roughly $7.8B of revenue in six months sold through Bedrock, Vertex and Azure Marketplace instead of direct.
Which means the hyperscalers get paid twice! They sell Anthropic the compute - Anthropic has committed hundreds of billions of dollars to AWS, Google and Microsoft - and then they take a fee on roughly half of the revenue distributed through their marketplaces.
Also on gross margin...those marketplace fees don't sit in cost of revenue. They're in sales, marketing & partnerships, so they never touch the headline 85% gross margin everyone is quoting today. Fees paid to channel partners alone rose ~$2.7B year over year in H1 2026.
Hyperscalers feel like they're kind of the winner here all around though reading through this.
The reason frontier LLMs will never be profitable is because the minute they stop spending, the free models catch up.
You're in a constant loop of increased spending to stay ahead, while everyone behind you catches up at a fraction of the cost by using your model.
LLMs are just commodity with no moat.
Todo ano é a mesma história, o Lula ganha a eleição, a Bolsa cai, o dólar sobe e aparece uma multidão anunciando o apocalipse econômico.
Três meses depois, a Bolsa bate recordes, o dólar recua e… silêncio absoluto.