Alphabet Inc. (GOOGL)

News Summary

• Alphabet guided 2026 CapEx at $175–185B (nearly double 2025), with ~60% directed at servers and ~40% at data centers and networking; management explicitly stated Google expects to remain supply-constrained on AI compute throughout 2026, signaling that demand for AI accelerators is outpacing even the most aggressive infrastructure buildouts in the industry.
• Google Cloud grew 48% YoY to $17.7B with operating margin expanding to 30.1% and backlog surging 55% sequentially to $240B, driven by enterprise AI adoption — with management framing the business as having shifted from AI investment mode into AI monetization mode, a dynamic that validates the commercial pull for high-performance inference infrastructure.
• Gemini serving unit costs fell 78% over 2025 through model optimization and utilization gains, while first-party API token throughput hit 10 billion tokens per minute (up from 7B last quarter) — concrete benchmarks illustrating the inference efficiency curve that cloud-scale AI operators are achieving and that competing chip vendors must match or undercut.
• Google's seventh-generation Ironwood TPU was cited as a key Cloud AI differentiator, with Pichai framing custom silicon as part of a vertically integrated stack and a competitive moat — reinforcing that hyperscalers are actively developing in-house alternatives to merchant AI chips, intensifying the competitive landscape Cerebras operates in.
• The Gemini app reached 750 million monthly active users and agentic AI products were highlighted as a primary growth driver for Cloud enterprise revenues, underscoring accelerating end-user and enterprise adoption of AI that is translating directly into infrastructure demand at scale.

Financial Highlights

Earnings Call Highlights

Alphabet's Q4 2025 call was defined by an unmistakable tone of confidence bordering on triumphalism, with management leaning hard into AI infrastructure as the central growth engine across every business segment. The $175–185B CapEx guide for 2026 — nearly double the prior year — was the dominant financial signal, and management framed it not as a risk but as a deliberate, demand-driven necessity, explicitly acknowledging they expect to remain supply-constrained throughout 2026. Analysts probed the sustainability of this spend and the monetization pathway for AI products, but received largely bullish responses grounded in accelerating Cloud revenue (48% growth, $240B backlog up 55% sequentially) and Search re-acceleration. The overall posture was that Google has moved from AI investment mode into AI monetization mode, with the infrastructure buildout now visibly translating into revenue.

• Google announced 2026 CapEx guidance of $175–185B, with roughly 60% directed at servers and 40% at data centers and networking; just over half of ML compute is expected to support the Cloud business, signaling massive continued demand for AI accelerators and infrastructure at scale.
• Cloud revenue grew 48% to $17.7B with operating margin expanding to 30.1%, driven by 'accelerating growth in enterprise AI products generating billions in quarterly revenues' — specifically citing TPU and GPU infrastructure deployment alongside Gemini model demand as the primary drivers, with backlog more than doubling year-over-year to $240B.
• Sundar Pichai confirmed Google remains supply-constrained on AI compute and expects that condition to persist through 2026, stating 'I do expect to go through the year in a supply constrained way' — a direct signal that demand for AI chips and infrastructure is outpacing even Google's aggressive build-out.
• Gemini serving unit costs dropped 78% over 2025 through model optimizations and utilization improvements, and first-party API token throughput reached 10 billion tokens per minute, up from 7 billion last quarter — concrete efficiency and scale metrics relevant to inference infrastructure economics.
• Google's seventh-generation Ironwood TPU was cited as a key differentiator for Cloud AI infrastructure, with Pichai framing TPUs not as a standalone product but as part of a vertically integrated stack — suggesting Google views its custom silicon as a competitive moat rather than a merchant chip opportunity.