Community discussions
Intel is investing over $100 billion into new fabs — Ohio ($20B now, up to $100B planned), Germany (~€30B / $32B), and Israel ($25B). Much of this is supported by the U.S. CHIPS Act ($8.5B in grants + $11B in loans) and European incentives.
A significant portion of future demand for those fabs comes from the same hyperscalers driving the AI boom — Microsoft, Amazon, and Google — through pre-orders and strategic partnerships.
Microsoft alone has invested over $13B (and up to $58B total commitments) in OpenAI, which spends a large share of that money back on Microsoft’s Azure to train and run its models.
Microsoft then spends tens of billions more buying GPUs from Nvidia — and now Intel’s Gaudi 3 chips — to expand that same AI infrastructure. Intel books AI-related revenue, reinvests part of it into fab expansion, and the cycle repeats.
On paper it looks like explosive AI growth, but in practice, much of it is the same capital moving in circles — financial recycling amplified by subsidies and cross-investment.
It works great while enthusiasm lasts, but when real customer demand slows, or people’s excitement fades, the loop could stop spinning fast.
As I see it, value doesn’t really exist on its own — it’s created by people’s belief in it.
Microsoft, Amazon, Google, Meta → OpenAI, Anthropic, xAI → Nvidia, AMD, Intel → TSMC, GlobalFoundries, Samsung.
The circular AI money loop only works if everyone in it — Microsoft, Nvidia, Intel, OpenAI, Amazon, Google, and the rest — keeps expanding at roughly the same pace. That means more chips, more datacenters, more AI models, and actual customers willing to pay for it all.
The problem is that this system depends on unbroken collective success. If even one or two major players slow down or dont make product or makes bad product, demand stalls and the rest follow — and that’s how bubbles usually end.