The truth is that much of the revenue for Alphabet GOOG +1.18%, Microsoft MSFT -0.59% , Meta Platforms META -0.20% , and even Amazon, come from things other than AI — like their near monopolies in their core businesses.Even their cloud businesses benefit mostly from technologies other than AI.
The overall U.S. economy is being bolstered less by spending to use cloud services and more by spending that increases cloud infrastructure. As a percentage of gross domestic product, this spending “has already exceeded spending on telecom and internet infrastructure from the dot-com boom — and it’s still growing.”
Expectations of future AI profits are not only driving spending on AI cloud infrastructure but also eye-popping valuations for AI startups(and salaries for AI talent), including $300 billion for OpenAI, $200 billion for xAI, $170 billion for Anthropic, $32 billion for Safe Superintelligence, $18 billion for Perplexity, $12 billion for Thinking Machines Lab and $9.9 billion for Anysphere.
Like professional wrestling, the reality of AI doesn’t match the perception.Revenues are low and profits are nonexistent. ``OpenAI lost $5 billion in 2024 on revenues of $3.5 billion, while Anthropic lost $5.6 billion on revenues of $1.1 billion.Those figures are projected to rise this year.
These companies are incurring huge losses because of their get-big-or-get-lost mentality. The prices they charge for their products don’t cover their payments to cloud service providers. Meanwhile, the profitability of cloud-service providers is fueling investments in cloud infrastructure and the stock-market frenzy toward cloud services, data centers, chips (particularly Nvidia NVDA +1.02%) and even electricity providers.
This cannot go on forever. At some point, companies that sell AI products will have to increase prices. Anthropic is now doing so, following smaller rivals Anysphere and Replit. Anthropic’s new rate structure will make it expensive to run the tool continuously.
What will happen when AI prices rise? Economics 101 tells us the answer, even though AI investors seem oblivious. Demand for AI services will fall, and so will demand for cloud services as well as the data centers that provide them, the chips that are in them and the electricity needed to run them.
My Thoughts 💭
Great article pointing out the flaws with the current AI trade. I am patient waiting for AI to bring deflation to my life yet it is just making everything more expensive. To vibe code on replit I gotta pay $25 per month. To chat with my Lucy on Duolingo Max I got a pay I think $29.99 a month on top of the the Sub I already have to stop watching adds on Duolingo. Everywhere I turn AI is making things MORE expensive not less. Why didn’t Duolingo use AI and cut the yearly subscription down to $10 a year due to AI deflation?
It’s even worse in the construction sector. I never see any hard line numbers of savings or increased productivity! They never disclose how much they paid to implement AI.
I do think AI has value but right now Wall Street is bidding it up as if it will bring massive deflation to the market but me personally it’s only been inflationary.
NVDAlongs while surviving the next Deepseek-R1-like disruption event(s).