© Drew Angerer / Getty Images News via Getty Images
I keep hitting the buy button on Alphabet (NASDAQ:GOOGL | GOOGL Price Prediction) for one reason: the company is turning its in-house AI silicon into an outside business, with first dollars landing this year.
For a decade, Google’s Tensor Processing Units were an internal cost lever. That posture is changing by the end of 2026, as AI turns Alphabet into an external chip and compute seller. On the Q1 2026 call, Sundar Pichai told investors Google will “begin to deliver TPUs to a select group of customers in their own data centers in the hardware configuration to expand our addressable market opportunity.” CFO Anat Ashkenazi added that TPU hardware agreements are already in the Cloud backlog, with “a small percent of them to come through as revenue later this year and then the majority to be realized as revenue in 2027.” That is the silicon shift driving my purchases.
The Receipts
Demand is real. Google Cloud revenue grew 63% year over year to $20.03 billion in Q1 2026, and backlog nearly doubled quarter on quarter to over $460 billion. Cloud operating margin climbed from 17.8% to 32.9%, and revenue from products built on Google’s GenAI models grew nearly 800% year over year.
The money to deliver is committed. Alphabet raised 2026 CapEx guidance to $180 billion to $190 billion, up from $175 billion to $185 billion, after the Intersect acquisition, and Ashkenazi said 2027 CapEx will “significantly increase compared to 2026.” Q1 CapEx alone was $35.67 billion, up 107.44% year over year. Buildouts at this scale only make sense if capacity gets sold externally.
The base business funds it. Q1 EPS came in at $5.11 against a $2.63 estimate, the fourth consecutive quarter beating expectations. Operating income rose 30% to $39.7 billion, with operating margin at 36.1%. Search revenue grew 19% to $60.4 billion. The quarterly dividend was raised 5% to $0.22 per share. At a P/E of 26 on a $4.295 trillion market cap, with an 83.14% one-year return, I am paying a reasonable multiple for the compounder underneath the AI story.
Why Alphabet Over Alternatives
Hyperscaler investors typically reach for Microsoft, Amazon, or Meta. I pass on them. Pichai’s framing is what I trust: “The fact that we own frontier models and own the silicon really helps us stay ahead of the curve… I think we are the only provider in the market that offers all of these vertical stack elements.” Microsoft relies on a partner for its frontier model. Amazon builds accelerators but does not ship a consumer frontier model at Gemini’s scale. Meta buys most of its compute. Alphabet is the only US-listed name pairing a proprietary frontier model (Gemini processing 16 billion tokens per minute via direct API, up 60% from the prior quarter), its own accelerator generation (TPU 8t with 3x the processing power of Ironwood), a hyperscale cloud, and global ad distribution under one roof. That vertical stack keeps my dollars here.
The Risk
CapEx is eating free cash flow. FCF fell 46.63% year over year in Q1 to $10.12 billion. If external TPU revenue slips deep into 2027 instead of trickling in late 2026, the FCF gap widens before it closes. I own that risk. A $462 billion Cloud backlog and 57 Buy ratings against zero Sell ratings tell me demand is booked and timing is a detail, not a thesis break.
I will keep buying Alphabet until the silicon shift stops being a footnote in the transcript and starts being a line item on the income statement.
Contact [email protected] for any questions or corrections.
