I keep pressing the buy button on Amazon (NASDAQ:AMZN | AMZN Price Prediction), and the reason is embarrassingly simple: the crowd is looking at the capex line and flinching, while I am looking at what that capex actually builds.
That is the whole thesis in one sentence. Amazon is spending in 2026 to own the compute layer of the AI economy in 2027 and beyond. The market treats the near-term free cash flow squeeze as a wound. I read it as a receipt.
The Three Numbers That Keep Me Buying
Start with AWS. In Q1 2026 it grew 28% year over year, fastest in 15 quarters, on a $150 billion run rate. Jassy said “It is very unusual for a business to grow this fast on a base this large.” Segment operating margin 37.7%, operating income $14.2 billion. That is the engine.
Next, chips. Amazon’s silicon business (Graviton, Trainium, Nitro) topped a $20 billion revenue run rate growing triple digits year over year. Trainium2 is largely sold out. OpenAI committed to roughly 2 GW of Trainium capacity beginning in 2027. Anthropic secured up to 5 GW. AWS backlog sits at $364 billion, and that excludes an Anthropic commitment for over $100 billion.
Third, the whole business prints. Q1 EPS came in at $2.78 versus a $1.653 estimate, a 68.18% beat, the fifth straight beat. Operating income $23.85 billion, up 29.6% YoY. Operating cash flow $26.03 billion, up 52.99%. ROE 22.29%, net debt to EBITDA 0.45, interest coverage 35.17x.
The stock closed at $247.55, up 7.25% YTD, while AMZN slipped 4.7% since the April filing. Analyst target is $312.87 with 62 buys and zero sells.
Why Not the Obvious Alternatives
Microsoft (NASDAQ:MSFT) and Alphabet (NASDAQ:GOOGL) are the obvious cloud picks. Amazon leads both on custom AI training silicon at Trainium’s scale. Its chip business is over $20 billion run rate with more than $225 billion in revenue commitments and OpenAI plus Anthropic locked in. Jassy: “Trainium will save us tens of billions of dollars of CapEx each year and provide several hundred basis points of operating margin advantage.”
Walmart (NYSE:WMT) is the retail comp. Amazon’s Online Stores grew 12%, third-party sellers 14%, and unit growth hit 15%, the highest since COVID. Walmart does not carry a $150 billion cloud business alongside that.
Meta (NASDAQ:META) is the ad alternative. Amazon’s ad segment ran $17.24 billion in Q1, up 24% YoY, with over $70 billion TTM. One AMZN share buys three growth engines.
The Real Risk
TTM free cash flow fell to $1.2 billion, down 95%. Q1 capex was $44.2 billion; the 2026 plan is roughly $200 billion. Long-term debt climbed to $119.1 billion from $65.6 billion. If AI monetization slips, that overhang stings.
Jassy addressed it directly: data centers carry 30-plus year lives, chips and servers five to six years. I am fine with a compressed FCF year sitting behind $364 billion of backlog.
What Keeps the Buy Button Active
Polymarket assigns 98.5% probability to 2026 capex exceeding $170 billion. Consensus has accepted the number without repricing the outcome. That gap is my window. The five-year base case models $523.92, a 111.64% total return. I keep buying because 2027 will not send an invitation.
Contact [email protected] for any questions or corrections.
Top Gaining Stocks
Top Losing Stocks