Bank of America analyst Vivek Arya reports that cloud infrastructure spending jumped to nearly $1 trillion（約160兆円） in 2026—far exceeding earlier expectations of $6–$700 billion（約110兆円）—and is projected to grow another 40–50% to $1.4–$1.5 trillion（約240兆円） in 2027. This acceleration reflects AI moving from experimentation into core enterprise workflows, with NVIDIA's Q1 revenue up 85% year-over-year as a clear indicator of demand. Persistent chip shortages and near-zero idle capacity suggest the infrastructure boom is far from complete.
Summaries like this, in your inbox every morning.
Sign up free →What happened
Bank of America Securities analyst Vivek Arya said cloud capex jumped from a projected $6–$700 billion（約110兆円） at the start of 2026 to nearly $1 trillion（約160兆円） by mid-year, and projects another 40–50% rise to $1.4–$1.5 trillion（約240兆円） in 2027. NVIDIA reported Q1 revenue of $81.615 billion（約13兆円）, up 85.23% year-over-year, with Data Center revenue up 92% and Networking up 199%.
Why it matters
Arya argues AI is shifting from experimental proof-of-concept to embedded enterprise software, driving hyperscalers to commit massive capital. Combined annual recurring revenues at OpenAI and Anthropic already exceed those of some large software firms, justifying continued spending. The Philadelphia Semiconductor Index is up nearly 80% year-to-date through early July 2026.
What to watch
NVIDIA guided Q2 FY27 revenue to $91.0 billion（約15兆円）, with total supply commitments of $119.0 billion（約19兆円）. Arya cited persistent chip shortages and near-zero idle compute capacity as evidence demand remains unsatisfied.
The sharp upward revision in cloud capex reflects a structural shift in how enterprises view AI infrastructure. When the year began, Wall Street expected $6–$700 billion（約110兆円） in spending; by mid-2026, that figure approached $1 trillion（約160兆円）—a dramatic repricing of demand. Arya's core observation is that AI has moved beyond novelty into operational necessity. The evidence he cites is concrete: combined revenues from just two AI companies (OpenAI and Anthropic) now rival established software giants, creating a powerful incentive for hyperscalers to keep building capacity.
The semiconductor sector is capturing most of this upside. NVIDIA's 85% revenue growth and the Philadelphia Semiconductor Index's nearly 80% year-to-date gain through early July 2026 demonstrate how concentrated the benefit is among chip suppliers. Equally telling is Arya's observation about supply constraints—even seven-year-old GPUs remain fully utilized with zero idle compute. This suggests the current capex wave, despite its size, is still insufficient to match demand. The projected 40–50% increase in 2027 cloud spending (to $1.4–$1.5 trillion（約240兆円）) appears to reflect both growth in enterprise demand and persistent recognition that available capacity lags what the market requires.
No comments yet. Be the first to share your thoughts!
Log in to join the discussion
Get curated AI news from 200+ sources delivered daily to your inbox. Free to use.
Get Started FreeFree · takes 30 seconds · unsubscribe anytime
1 minute a day. The AI essentials.
200+ sources · Email / LINE / Slack
