The AI infrastructure buildout has reached a scale with no clean historical precedent. Hyperscale technology companies have now committed a record $2.6 trillion in future spending across data-center leases and equipment purchases, including commitments for data centers and power infrastructure that haven't even begun construction yet. Taken together, it's being described as the largest capital expenditure spree in market history.
The scale of the number is what stands out. $2.6 trillion in committed future spending is larger than the annual GDP of all but a handful of countries, and it's being deployed by a small handful of companies racing to secure compute capacity, power access and physical data-center space before competitors lock it up first.
The Bull and Bear Case Are the Same Number
Spending at this scale cuts both ways for how markets read it. Bulls see it as evidence that hyperscalers expect AI demand to keep compounding for years, justifying infrastructure investments that won't fully pay off for a decade or more. Bears, including investors like Michael Burry who have built large short positions against AI-exposed names, see the same number as proof that capital expenditure has decoupled from anything resembling near-term revenue — a debate playing out directly in how Palantir's stock and options market are being priced right now.
Related: GSR: Crypto's Next Bull Run Needs Cooling AI Capex and Fed Cuts
Why Crypto Markets Are Watching Capex, Not Just Rates
The AI capex cycle has become entangled with crypto sentiment in a way that goes beyond simple correlation — the same hyperscaler spending that's soaking up chip supply, power capacity and capital is also seen by some analysts as competing directly with crypto for investor risk appetite. That's the logic behind arguments that crypto's next real bull run may need this exact spending cycle to cool off before capital rotates meaningfully back into digital assets.
