Big Tech's $700B AI Spending Faces 20% Cost Inflation

When Big Tech companies report earnings starting next week, the headline number won't be revenue growth or profit margins. Investors will be watching capital expenditure forecasts — and trying to decode how much actual AI infrastructure those dollars will buy.

Google, Amazon, Microsoft, and Meta have collectively outlined plans to spend more than $700 billion on AI data centers in 2026. But a significant portion of those increases reflects rising costs rather than expanded capacity, according to new analysis from Morgan Stanley and industry researchers.

The inflation problem

Morgan Stanley estimates that building one gigawatt of AI data center capacity now costs roughly 20% more than it did recently for several leading system configurations. A common Nvidia-based setup has climbed from approximately $29 billion to $35 billion per gigawatt, while a newer architecture has jumped from $41 billion to $49 billion.

The culprits are widespread: memory chip prices have surged, power equipment is scarce, construction materials cost more, skilled workers are harder to find, and electricity grid connections face long delays.

Brad Gastwirth, head of research at Circular Technology, estimates that 20% to 30% of the next wave of AI capex increases will simply cover inflation, with only 70% to 80% representing genuine capacity expansion. Earlier research suggested soaring memory prices alone could explain about 45% of cloud company capex growth this year.

Why it matters

For investors trying to gauge the pace of AI infrastructure buildout, a rising capex number no longer tells the full story. Companies are caught in what amounts to a spending spiral: they order more equipment, which worsens shortages, which drives prices higher, which forces them to raise spending forecasts further — creating still more demand pressure.

Cantor Fitzgerald analysts expect Big Tech's 2026 spending plans to hold steady in the coming earnings reports, but project sharp increases for 2027: $283 billion for Google, $271 billion for Amazon, and $200 billion for Meta.

What to watch

No major tech company is likely to pull back on AI spending while competitors race ahead. But Gastwirth suggests investors should look beyond the headline capex figure and listen for specifics about power capacity, GPU deployments, memory purchases, networking infrastructure, and new data center campus announcements.

"If capex rises alongside those metrics, it points to genuine expansion rather than simply higher costs," Gastwirth said. Without those operational details, a larger spending number may just mean these companies are paying more to maintain their current trajectory.

These details were first reported by Alistair Barr in Business Insider's Tech Memo newsletter.