Overview
- VP Amin Vahdat told an internal meeting the company is targeting a 1,000x increase in serving capability within four to five years by doubling every six months, according to reports.
- A Google spokesperson said surging AI demand is driving new capacity through hardware, software and model efficiencies, highlighting the new Ironwood TPU chips.
- Sundar Pichai warned employees that 2026 will be an intense year and said the Veo video generator’s rollout was constrained by limited compute.
- Executives contend the bigger risk is underinvesting, pointing to Google Cloud’s roughly 34% annual revenue growth to over $15 billion and a $155 billion backlog.
- Industry analysts say serving capacity is emerging as the primary bottleneck as power, cooling, networking and data‑center build times limit how fast providers can scale.