Alphabet captures 2026 AI capex gains through custom TPU silicon ramp, while Amazon begins 2027 surge with Trainium/Graviton diversification.

Alphabet and Amazon both closed Q1 FY2026 with blockbuster AI infrastructure disclosures. Alphabet is already converting compute into ad and cloud dollars, while Amazon is still pouring concrete for the 2027 wave of Trainium capacity tied to OpenAI and Anthropic. Comparing them now clarifies who earns the capex back first.

**Alphabet's Momentum**

Alphabet posted revenue of $109.90 billion, up 21.8% year-over-year, with Search & Other at $60.4 billion and Google Cloud up 63% to $20.03 billion. EPS came in at $5.11 versus a $2.63 consensus — a decisive beat driven by the ad engine.

CFO Anat Ashkenazi raised Alphabet's 2026 capex guidance to $180–190 billion and flagged that "2027 CapEx to significantly increase compared to 2026." Meanwhile, Sundar Pichai noted that Gemini has already cut core AI response costs by more than 30%, and enterprise GenAI revenue grew nearly 800% year-over-year. The Google Cloud backlog sits at $462 billion, with roughly 50% converting to revenue in the next 24 months.

Alphabet's free cash flow fell to $10.12 billion, down 46.63%, a metric to watch as the company tries to outrun depreciation with ad monetization.

**Amazon's Bet on 2027**

Amazon delivered $181.52 billion in revenue, up 16.61%, with AWS at $37.587 billion, growing 28% — its fastest growth in 15 quarters. The custom chips business now runs at a $20 billion revenue run rate, growing triple digits. However, the biggest committed workloads — OpenAI's 2 GW of Trainium — do not start until 2027.

Andy Jassy is playing a longer game. He guided to approximately $200 billion in 2026 capex across AI, chips, robotics, and Kuiper satellites. Trailing twelve-month free cash flow fell to $1.2 billion, a 95% drop, while long-term debt jumped to $119.1 billion from $65.6 billion. These are big bets, but the cash math looks ugly today.

**The Outlook**

For Amazon, Q2 guidance of $194–199 billion in revenue and Trainium3 supply commitments matter more than the retail earnings report. Polymarket assigns a 97.7% probability that Alphabet beats Q2 earnings and a 93.5% probability that Amazon beats.

The thesis favoring Google in 2026 is compelling: the ad stack turns TPU spend into cash inside quarters, not years. Alphabet shares are up 89.43% over the past year, while Amazon's 7.11% year-to-date return reflects a patience penalty. If OpenAI actually lights up 2 GW of Trainium in early 2027, the narrative rotates. Until then, Alphabet's capital velocity looks tough to beat.