AI infrastructure — $700B this year
Analysts estimate hyperscalers will spend nearly $700 billion on AI infrastructure in 2026, and broader S&P 500 corporate AI investments could top $1 trillion—fueling which stocks look attractive for long-term exposure to the theme reported. Market writeups singled out names like Oracle, AMD and Micron as potential undervalued AI plays worth considering in growth allocations.
Alphabet, Microsoft, Meta and Amazon each raised 2026 capital-expenditure guidance during early-2026 earnings season, signaling an industry-wide shift toward sustained infrastructure buildouts (cnbc.com). Goldman Sachs’ research shows consensus Wall Street capex estimates for the hyperscaler group climbed to about $527 billion for 2026, up from roughly $465 billion earlier in the cycle (goldmansachs.com). Gartner projects global AI spending at $2.52 trillion in 2026, with a $401 billion component labeled "AI infrastructure" driving much of that expansion (gartner.com). OpenAI’s Stargate initiative and Oracle’s role include a reported multi-hundred‑billion commitment and roughly 4.5 gigawatts of additional data center capacity in recent announcements and updates (openai.com). Oracle boosted fiscal‑2026 capex guidance into the tens of billions as it scales Oracle Cloud Infrastructure to service large AI contracts, according to company commentary and coverage of its Q2 FY2026 call (channeldive.com). NVIDIA’s fiscal‑2026 results showed record data‑center revenue and full‑year revenue of $215.9 billion, underscoring its dominant supplier position to large AI customers (nvidianews.nvidia.com). Memory suppliers tightened supply lines as Micron indicated 2026 HBM capacity is effectively booked and signaled multi‑billion capex increases, while AMD expanded Instinct GPU availability and guided data‑center revenue into 2026 growth ranges (trendforce.com) Moody’s and industry observers flagged grid and power‑supply stresses tied to large data‑center projects and warned of rating and operational considerations as hyperscale buildouts accelerate (msn.com)
