Google Cloud TPU Revenue Recognition and $513.9 Billion Backlog, October 2026

BNP Paribas raised Alphabet's price target to $450 on TPU business outlook, as Google Cloud's committed backlog stands at $513.9 billion and the company begins recognising TPU system revenue for the first time.

BNP Paribas raised its Alphabet price target to $450 this month, citing the company's TPU business outlook — the latest in a sequence of sell-side upgrades anchored to Google's custom-silicon programme. Piper Sandler moved to a $400 target in September, projecting that TPU chip revenue could reach $104 billion by 2028. The escalation in analyst targets reflects a structural shift: Alphabet has begun recognising revenue from TPU systems for the first time, and Google Cloud's committed revenue backlog has reached $513.9 billion, a figure that some analysts now treat as a more significant valuation input than near-term quarterly cloud growth rates alone.

The operational context behind both milestones is material. Google Cloud's single-quarter revenue grew 82% year-over-year, according to September 2026 reporting, while TPU demand has reportedly outpaced TSMC's advanced CoWoS packaging capacity — a supply constraint that frames the backlog as a deployment challenge as much as a commercial one. Alphabet is simultaneously funding both Nvidia's NVLink Fusion and AMD's UALink interconnect programmes, a hedge against vendor lock-in that signals the breadth and complexity of managing a custom-chip strategy at this scale. Berkshire Hathaway's Greg Abel described Google as a significant player in AI in September, publicly anchoring the firm's $10 billion stake in Alphabet to that positioning.

The capital commitment underlying this strategy has moved decisively into multi-hundred-billion-dollar territory. Slicast's own compilation of SEC XBRL filings shows Alphabet's capital expenditure reached $91.45 billion in FY2025 — up 74.1% from $52.53 billion in FY2024 — representing 23% of $402.84 billion in that year's revenue, placing Alphabet third in absolute capex among the six hyperscalers tracked. The pace has not slowed: the period ending March 31, 2026 produced the largest single quarter on record at $35.67 billion, and management has since guided 2026 full-year capex to $195–205 billion, roughly $15 billion above prior guidance, with reports noting a further significant increase has been signalled for 2027. To fund this trajectory, Alphabet closed a $25 billion senior notes offering on August 10, 2026, as confirmed by the company's SEC 8-K filing of that date, with tranches maturing from 2028 onward. Goldman Sachs has projected that five major Big Tech companies will collectively spend $1.2 trillion on AI infrastructure through 2027, framing the current cycle as the largest capital deployment since nineteenth-century railroads.

The risks are equally concrete. Alphabet's stock was reported to have declined on investor questions about the ROI and capital intensity of the TPU programme, and according to July 2026 coverage, the Q2 earnings release — which beat revenue estimates — still saw the share price fall in after-hours trading as the capex guidance was revised upward. At $340.59, the stock sits well below both sell-side targets. A separate report notes that Google Cloud may face a situation by 2027 in which OpenAI and Anthropic together account for approximately 48% of its revenue — a concentration that would represent an unusual dependency for a platform-scale cloud service and creates binary risk if either relationship changes. Alphabet is reportedly executing a strategy to offload $250 billion in capex obligations and has backed leases on 2.4 gigawatts of capacity across ten projects, an approach that limits balance-sheet exposure but introduces counterparty and execution risk. Google DeepMind also saw a leadership transition, with Demis Hassabis moving to the role of Alphabet chief scientist and Jeff Dean departing after 27 years to found a new venture; the implications for the research continuity underpinning the TPU roadmap remain to be seen.

Three variables will define how the TPU thesis plays out. First, TSMC CoWoS packaging capacity: meaningful easing of supply constraints would directly accelerate Google Cloud's ability to convert its $513.9 billion backlog into recognised revenue. Second, the capex deployment mix: whether Alphabet deploys the $195–205 billion 2026 guidance through owned infrastructure or increasingly relies on third-party compute — as recent reporting suggests is already occurring — affects both earnings quality and the credibility of the offload strategy. Third, customer-base diversification: whether Google Cloud broadens its enterprise revenue base meaningfully, or whether the reported concentration around two AI-native clients deepens through 2027, may prove the most consequential structural question for the platform's long-term revenue quality.