Google CEO Says Meeting New 2027 Compute Requests Would Be Difficult

Sundar Pichai, CEO of Alphabet and Google, said meeting a new request for compute in 2027 would be difficult because capacity is planned years in advance.

Google’s current computing capacity reflects decisions made in 2024, and its infrastructure planning runs two to three years ahead, Pichai said. “If today a customer is coming out and asking us for compute in ’27, that is incredibly difficult to do,” he said. Google leased short-term compute in 2025 to help some customers. Pichai did not quantify the capacity available for such requests or say how many customers could face that constraint.

Google Cloud’s recent growth gives context to the capacity pressure. Revenue for the quarter ended June 30, 2026, climbed 82% from a year earlier to $24.8 billion. Its backlog reached $514 billion, more than $50 billion above the previous quarter. Alphabet expected just over half of that backlog to be recognized as revenue over the following 24 months.

Alphabet spent $44.9 billion on capital expenditures in Q2 and lifted its 2026 outlook from $180 billion-$190 billion to $195 billion-$205 billion as it accelerated capacity delivery to meet demand. Most Q2 capital spending went to technical infrastructure. Alphabet also expects capital expenditures to rise significantly in 2027, but did not give a range.

As of June 30, Alphabet had $85.2 billion in future lease payments, primarily for data center leases that had not yet commenced. The figure adds to the scale of the company’s long-term infrastructure commitments.

Alphabet planned to increase its use of third-party capacity during Q3 2026 while building internal infrastructure. The added capacity was intended as a bridge, and the company expected it to put modest pressure on near-term margins.

Google prioritizes Google DeepMind research and frontier model development when allocating compute, with other business needs planned around that work. Pichai has described Google’s compute trade-off between frontier AI work and near-term Cloud revenue. Directing capacity to research means Google Cloud gives up some short-term revenue.

Pichai said component bottlenecks or economic shocks could change demand and lead the industry to build too much capacity. He expects Google to find uses over time for its planned compute.