AI chip wars go circular: Huawei’s Tau rollout meets Tencent’s Oracle leases—who wins the compute race?

Across multiple outlets on 2026-09-30 to 2026-10-01, the cluster highlights how AI compute is being secured through intertwined investment, supply, and customer relationships. NRC questions whether valuations of AI companies are being artificially inflated as major AI firms, chipmakers, and data-center builders increasingly act as each other’s investors, suppliers, and customers in a “circular” financing model. In parallel, SCMP reports Huawei is stepping up the rollout of its Tau-based advanced chips, using its self-developed Tau Scaling Law, and is positioning the processors for adoption in newly released Mate 90 smartphone series, with an ambitious sales trajectory for the Mate XT 2. The Financial Times adds a concrete compute procurement move: Tencent, the WeChat owner, is leasing 100,000 chips from Oracle to accelerate its AI push, aiming to secure access to Oracle’s data-center capacity in Southeast Asia amid competitive pressure from ByteDance and Alibaba. Geopolitically, the compute race is increasingly shaped by export controls, technology substitution, and cross-border infrastructure access. Huawei’s Tau rollout signals continued pressure on Western supply chains by strengthening in-house semiconductor capability, even as the article frames the effort in the context of US export restrictions. Tencent’s Oracle leasing deal shows how US-linked infrastructure can still be monetized by Chinese platforms when capacity is needed quickly, effectively turning data-center access into a strategic lever. Meanwhile, the NRC “circular” financing critique points to a market-structure risk: when the same ecosystem players fund and buy from each other, it can mask underlying demand, distort risk pricing, and amplify volatility if AI capex slows. Net beneficiaries are likely to be firms that can secure both chips and data-center throughput fastest, while losers could include less integrated suppliers facing margin compression or refinancing stress. Market implications span semiconductors, cloud infrastructure, and AI application demand. Huawei’s Tau scaling narrative supports the case for continued investment in China’s domestic chip stack, which can pressure segments reliant on constrained advanced-node supply, even if near-term performance and yield remain debated. Tencent’s 100,000-chip lease from Oracle is a direct signal of sustained AI capex intensity and demand for high-performance accelerators, which can support revenue visibility for Oracle’s infrastructure and for any upstream suppliers tied to those systems. The “circular” valuation concern raises the probability of multiple compression in AI-adjacent equities if financing conditions tighten or if customer commitments prove less durable than implied. For investors, the direction is broadly supportive for compute infrastructure and AI supply-chain throughput, but with elevated risk premia around valuation sustainability and refinancing cycles. What to watch next is whether these procurement and rollout stories translate into measurable throughput, unit economics, and competitive displacement. For Huawei, key triggers include the pace of Tau chip adoption across the Mate 90 series, any disclosed performance benchmarks, and evidence of scaling beyond flagship devices into broader product lines. For Tencent, monitor follow-on capacity leases, utilization rates in Oracle’s Southeast Asia data centers, and whether the deal expands beyond the initial 100,000-chip tranche as competition with ByteDance and Alibaba intensifies. For the broader market, the NRC thesis implies watching credit spreads, AI capex guidance from data-center builders, and signs of demand decoupling from valuations. Escalation risk would rise if export-control enforcement tightens or if financing “circles” show stress; de-escalation would be signaled by transparent earnings quality, stable utilization, and credible evidence that customer demand—not internal cross-holdings—drives growth.

Geopolitical Implications

• 01

  Export controls are pushing China toward deeper vertical integration in chips, while US-linked infrastructure remains a partial bridge via leasing and capacity access.

• 02

  Compute procurement in Southeast Asia becomes a strategic battleground where infrastructure availability can influence AI platform competitiveness.

• 03

  If circular financing distorts valuations, it could trigger market corrections that reshape bargaining power across the AI supply chain.

Key Signals

• —Huawei: disclosed Tau chip performance, yield, and breadth of adoption beyond flagship devices.
• —Tencent/Oracle: expansion of leased capacity, utilization rates, and any renegotiation tied to export-control enforcement.
• —Data-center builders: capex guidance quality and evidence of external customer demand versus internal ecosystem churn.
• —Credit and equity market indicators: spreads and multiple compression in AI-adjacent names if earnings quality deteriorates.