Dwindling cash and soaring memory costs: Tech's AI buildout has ballooning price tag
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Goldman Sachs projects AI spending among megacaps will reach $765 billion in 2025 and climb to nearly $1.2 trillion in 2027.
- Amazon raised its 2025 capex forecast to $220 billion (highest among hyperscalers) while posting $7.6 billion in negative trailing-12-month free cash flow; Meta reported a 91% year-over-year cash generation drop and Alphabet posted negative free cash flow for the first time on record.
- Apple is particularly exposed to the memory crunch, having already raised prices on Macs and iPads, with CEO Tim Cook warning memory costs will keep rising beyond September and iPhone price hikes potentially following.
- Tesla CEO Elon Musk called memory pricing "insane" on the earnings call and publicly thanked Micron for giving Tesla "a very significant allocation on reasonable terms" — a rare CEO shout-out to a component supplier.
- Microsoft had its best trading day since 2008 after better-than-expected results and raised capex guidance, while Wedbush called Amazon's report the "cleanest beat" among hyperscalers.
- Chinese open-weight AI labs are releasing models that close the performance gap with OpenAI and Anthropic (each near $1 trillion on the private market) at much lower prices, per the article.
- JPMorgan's Dana Harlap wrote the market is now "more critical — and more discriminating" about hyperscaler AI spending, scrutinizing returns even when companies beat revenue estimates like Google's 82% cloud growth.
Why it matters: For the first time on record, Alphabet posted negative free cash flow while Meta's cash generation plunged 91% — yet Microsoft rallied on its best day since 2008, showing Wall Street now judges hyperscalers individually on AI returns rather than rewarding them as a bloc. With Apple consumers already facing price hikes and Chinese labs squeezing the pricing power of OpenAI and Anthropic, the cost side of the AI trade is squeezing both buyers and sellers.
