Alphabet Sells TPU Chips to Outside Customers
Alphabet has begun selling its Tensor Processing Unit (TPU) chips directly to selected outside customers, moving beyond its previous model of offering access only through Google Cloud. The first systems were delivered to customer data centres in the second quarter of 2026, generating initial direct-sales revenue. Alphabet expects most revenue from current agreements to arrive in 2027, with quarterly sales likely to fluctuate because of large equipment orders.
Meta is running TPUs for production workloads in its own data centres. Oracle plans to offer the chips through Oracle Cloud Infrastructure in 2027, making it the first hyperscaler outside Google to do so. Analysts estimate external TPU revenue could reach about $8 billion in 2026 and between $51 billion and more than $75 billion in 2027, though these are forecasts, not company guidance. The commitments are included in Google Cloud’s $514 billion backlog at the end of Q2. Google Cloud revenue rose 82% year on year to $24.8 billion in Q2, before the impact of TPU sales.
The move gives Alphabet a new route to compete with Nvidia in AI accelerators and broadens the market for its custom chips. Adoption will depend in part on improving compatibility with developers’ software tools. For traders, the key signals are future TPU revenue disclosures, delivery progress in 2027 and whether additional cloud providers adopt the hardware.
Neutral
The news is primarily about AI hardware and cloud infrastructure, so its direct effect on cryptocurrency prices and market stability is limited. It does not report a change in crypto regulation, token demand, blockchain activity or digital-asset liquidity. Any immediate trading reaction in crypto is therefore more likely to come through broader technology-sector sentiment and risk appetite than through a direct fundamental link.
In the short term, Alphabet’s expansion could reinforce investor interest in AI infrastructure and chip competition. That may support risk-on sentiment across technology assets, including crypto, but the effect is likely to be modest and sensitive to wider market conditions. The article also highlights execution risks: TPU revenue is expected to be uneven, most sales are forecast for 2027, and software compatibility remains a challenge. These uncertainties could temper enthusiasm if deliveries or customer adoption fall short.
Over the longer term, successful TPU adoption could diversify AI chip supply and intensify competition with Nvidia. Historically, major AI hardware announcements have often influenced crypto indirectly through movements in technology shares and changing expectations for AI-related investment, rather than consistently driving broad crypto rallies. Traders should watch subsequent revenue disclosures, delivery milestones, cloud-provider adoption, and wider indicators such as Bitcoin’s correlation with tech equities and overall liquidity. On the information provided, the most appropriate market classification is neutral.