1 OCTOBER 2026 · DEMAND GEOGRAPHY UPDATE
Chinese AI demand does not necessarily become Chinese domestic data-centre demand.
The Financial Times and Reuters report that Tencent agreed a five-year Oracle lease worth about US$7bn for access to roughly 100,000 advanced AI chips across Southeast Asian data centres. The reported rationale is access to chips unavailable in China.
Interpretation: this is a useful boundary condition for the demand thesis. Export controls can increase the value of domestic substitution, but they can also push some Chinese-company workloads offshore. We therefore separate Chinese AI demand from domestic Chinese AIDC demand when testing VNET, GDS and other local infrastructure beneficiaries.
Evidence state: high-quality reported transaction; Tencent and Oracle have not publicly confirmed the commercial details.
30 SEPTEMBER 2026 SWEEP · EVIDENCE ORIGINALLY REPORTED 14 SEPTEMBER
Token activity is beginning to enter the financing layer.
Beijing’s first “亦企Token（词元）贷” programme provided nearly RMB2bn of credit through six banks. The framework can use token production/consumption and compute-service contract value as core credit inputs alongside technical and business factors.
Research implication: this does not prove stronger paid AI demand or make Token a national financial standard. It does show measurable AI activity beginning to function as an economic signal outside the compute stack itself — useful evidence for how a Token-based service economy could connect output, settlement and finance.
26 SEPTEMBER 2026 · DEMAND / ECONOMICS LENS
The industry is starting to distinguish token volume from economically useful output.
A 26 September China Business Journal report, drawing on the newly released China Token Factory Development White Paper 2026 and operator activity, describes a shift from asking whether compute exists to whether infrastructure can produce stable, competitively priced effective tokens. The underlying industry white paper separately tracks service-grade sellable token capacity, token energy efficiency, cost per million Good Tokens and service quality.
Research implication: this is not proof that paid AI demand or margins have improved, and the white paper is not a national standard. It is useful evidence that the commercial measurement problem is moving toward the same endpoint as our thesis: useful service output, utilisation and unit economics rather than raw accelerator, token or MW counts.
Token Factory white paper publisher ↗ China Business Journal · 26 Sep ↗
The role in the system
Demand originators and model ecosystems.
Demand growth creates an opportunity, but activity must become recurring paid consumption.
Companies in this layer
DeepSeek · Alibaba / Qwen · ByteDance / Doubao · Xiaomi
Names identify research roles, not a claim that every company has a partnership with every other company.
What would strengthen the thesis?
Evidence of paid adoption and customer diversity.
What would weaken it?
Technical progress that fails to improve usable output or economic returns; delays, costs or commercial friction that prevent the value test being met.
Follow the evidence
22 SEPTEMBER 2026 · RESEARCH UPDATE
Demand ambition and contracted demand
Alibaba’s new target strengthens the long-term demand case. KC’s reported relationship is a separate test of whether demand reaches independent suppliers. Neither establishes realised returns.