Hyperscaler capex arms race $680B 2026/$7T DC w/power crunch
Key Questions
What is the projected total capex for the Magnificent 7 hyperscalers in 2026?
The Magnificent 7 companies are projected to spend $680 billion in capex during 2026. This figure reflects the ongoing AI infrastructure arms race across major tech firms.
How much is Meta planning to invest in AI and data centers?
Meta has outlined plans to spend between $125 billion and $145 billion on AI-related infrastructure. The company is also cutting 8,000 jobs to reshape costs around AI priorities.
What are the monthly spending rates for Anthropic and xAI?
Anthropic and xAI together are spending approximately $1.25 billion per month on AI development and infrastructure. This pace underscores the rapid capital deployment in frontier model training.
What is the Google and Blackstone TPU joint venture?
Google and Blackstone plan to launch a new U.S.-based AI cloud company focused on TPUs, backed by $5 billion in equity from Blackstone as majority owner. The venture aims to expand TPU cloud capacity.
What were Nvidia's Q1 revenue results and key drivers?
Nvidia reported Q1 revenue of $81.6 billion, up 85% year-over-year, with data center revenue reaching $75.2 billion. The company also announced an $80 billion share buyback amid strong AI demand.
What supply and infrastructure challenges persist in AI data centers?
Power constraints and HBM memory shortages continue to limit AI infrastructure scaling. These bottlenecks are flagged across hyperscaler buildouts and Nvidia's supply chain.
How is oil and gas sector responding to the AI capex boom?
Oil and gas companies are increasing capital expenditure after years of restraint to support rising energy demand from AI data centers. Crude prices hovering near $100 reflect this dynamic.
What other major AI infrastructure investments are underway?
HIVE announced a $2.5 billion AI gigafactory in Toronto, while Alphabet issued a record yen bond to fund AI infrastructure. Nebius is pursuing a $25 billion capex plan for its AI platform.
Mag7 $680B 2026; Meta $125-145B; Anthropic-xAI $1.25B/mo; Google/Blackstone $5B TPU JV; Nvidia Q1 $81.6B rev (+85%), $75.2B data center, +$80B buyback. Power flags and HBM shortages persist.