Post

#GoogleBacksAIInfra #GoogleBacksAIInfra Alphabet has kept escalating its AI infrastructure bet through 2026. In June, it raised its 2026 capex guidance to $195-205B (up from $180-190B) and announced $80B in equity offerings — including a $10B private placement from Berkshire Hathaway — specifically to fund the buildout, notable because Alphabet generates well over $100B/year in free cash flow yet still chose to raise equity, signaling capex has outgrown even Google's own cash engine. The payoff shows up in Cloud: revenue jumped 82% to $24.8B in Q2, with backlog reaching $460B — a real leading indicator, since multi-year compute commitments at that scale aren't speculative pilots. CFO Anat Ashkenazi confirmed Q2 capex hit $44.9B, split roughly 60% servers/40% data centers and networking, with a large chunk funding Google's seventh-gen TPU (Ironwood), which went generally available in April. The tension: this spending pace is now standard across all hyperscalers — Microsoft, Alphabet, Amazon, and Meta combined are committing $660-690B in 2026 alone, nearly double 2025 levels — and Wall Street keeps punishing capex increases even when cloud results beat expectations, worried returns aren't keeping pace with the scale of investment. Alphabet is betting the opposite: that owning compute at this scale is the actual moat, not a cost to be minimized.