Microsoft, Alphabet, Amazon, and Meta boosted capex by 50% to a total of $106B in H1 2024, as they build AI infrastructure and pledge more investment hikes

Technology stocks have been volatile as Microsoft, Meta, Amazon and Google report huge increases in their investments in artificial intelligence

Financial Times 2024-08-04

Context & Ripple Effects

This marks an acceleration from the companies' earlier more than $32B in combined Q1 data-center and capital spending: Microsoft, Alphabet, Amazon and Meta are committing capital to the infrastructure needed to develop and run AI services. The reported share-price volatility shows investors were already weighing the scale and timing of those outlays against prospective returns.

The subsequent coverage makes this a clear early stage of an escalating spending cycle: the group was later projected to surpass $200B of 2024 capital expenditure, before reporting $246B for the full year.

First-order effects

Microsoft, Alphabet, Amazon and Meta redirect substantially more cash toward AI infrastructure, raising near-term capital intensity and making investment pace a more prominent factor in how their results are assessed.
The four companies signal that current build-outs are not one-off projects: suppliers and internal infrastructure teams gain clearer demand visibility from pledges of further spending increases.

Second-order effects

The simultaneous commitments increase pressure on each company to keep pace with rivals' infrastructure capacity, making a unilateral slowdown harder while AI capabilities are treated as competitive differentiators.
Larger infrastructure budgets concentrate more of the AI race in assets that require large upfront financing, sharpening investor scrutiny of whether product and cloud demand can support the spending.

Third-order effects

If these parallel increases persist, AI competition may increasingly favor companies with the cash flow and balance-sheet capacity to fund recurring infrastructure build-outs, rather than only those with strong models or applications.
The pattern points to an AI infrastructure capital cycle in which market valuations are more sensitive to the gap between investment commitments and evidence of monetization; the durability of that cycle remains contingent on demand.

The trend: AI is becoming a capital-intensive platform race, with the largest technology companies treating infrastructure scale as a core competitive asset.

Discussion

@carnage4life Dare Obasanjo on x
“During a gold rush, the best business to be in is selling picks and shovels” - Jensen Huang, CEO of Nvidia He hasn't said this out loud but I'm sure he's thought it. 🙃 [image]
@pradeepviswav @pradeepviswav on x
This is not entirely positive for Nvidia because of four of them have their own AI accelerators. Their reliance on Nvidia will come down significantly.
@radnorcapital @radnorcapital on x
I think its too early to call for slowing AI infrastructure spend when those who are spending are signaling the opposite. Companies don't make long term capital investments expecting near term returns (although $META clearly seeing benefits to their ad business). Eventually
@modestproposal1 @modestproposal1 on x
Interesting how META CFO talks about the fungibility of GenAI capex, meaning that infra built out for training can be repurposed for inferencing and core AI use cases. “We can flex capacity where we think it will be put to best use”. [image]
@rachel_grfn Rachel Griffin on x
interesting counter to narrative that ads/surveillance are becoming less central to political economy of big tech & now it's all about control of AI/computing infrastructure - only big co actually making money from AI is Meta, using it to sell more ads https://www.ft.com/...
@james_muldoon_ James Muldoon on x
Interesting to see that the one company making AI pay the bills is Meta - and it is doing so by integrating it into its advertising model. https://www.ft.com/...

Chronicles