The underlying risk picture is broadly unchanged. Project Jupiter’s power-related delay continues to test data-centre financing and contract structures, while investors report greater selectivity toward AI-linked debt. Strong accelerator demand remains a counterweight, but capex still materially exceeds the available cloud-revenue and free-cash-flow evidence. No material China-driven change was reported since the prior briefing; recent DeepSeek-Huawei and Alibaba developments continue to strengthen China’s domestic stack and potential price-performance competition.
DAILY BRIEFING
AI financing and power constraints remain focal risks; China’s software progress raises competition but no verified cycle break.
Elevated cycle risk is unchanged: hyperscaler investment and data-centre construction remain far ahead of measured cash-return and cloud-revenue proxies, while credit markets are becoming more selective. China caused no material new change since the prior briefing, but DeepSeek-Huawei software portability and Chinese full-stack expansion keep downward pressure on model economics plausible while potentially broadening global AI adoption and compute demand.
- Aggregate hyperscaler capex growth of 72.29% coincides with aggregate free-cash-flow growth of -16.11%, leaving returns sensitive to sustained demand and utilisation.
- AWS and Google Cloud revenue growth of 24.30% remains well below the 64.73% capex growth comparator.
- Oracle’s Project Jupiter force-majeure notice followed power delays; Reuters reported the project is delayed by about a year and that the episode is affecting financing discussions elsewhere.
- Reuters reported that credit investors are demanding greater concessions and reassessing concentration limits for AI-linked issuance as expected hyperscaler borrowing rises sharply next year.<|DELIM_zaimDYt|>
- NVIDIA Data Center revenue grew 117% year over year to $89.00 billion in the latest reported quarter, supporting the view that accelerator demand remains strong.
- Reuters reported that Blue Owl said Project Jupiter’s force-majeure notice did not alter the parties’ financial commitments, so the project-specific delay is not evidence of a cancelled tenant commitment.
- China’s domestic software and chip-stack progress may lower inference costs and accelerate deployment, which could expand token demand even if it compresses some model-provider pricing.
- Huawei reported more than 1,000 deployed industry projects in its AI Foundry ecosystem, although these company-reported figures do not establish their profitability or compute intensity.
- Whether Project Jupiter’s power delay changes funding terms, tenant obligations, construction milestones or lender protections.
- Whether AI-linked debt concessions widen relative to comparable investment-grade issuance as supply increases.
- Whether grid connection delays and equipment lead times translate into broader data-centre schedule deferrals.
- Whether cloud backlogs convert into recognised revenue and free cash flow rather than further contracted capacity alone.<|DELIM_zaimDYt|>
Competitive position: China’s position is strengthening across open models, software portability and domestic infrastructure. DeepSeek-Huawei’s Ascend tooling directly targets CUDA dependence, while Alibaba and Huawei are pursuing vertically integrated chips, clusters, models and enterprise platforms. Export restrictions and continuing reliance on US chips for some leading-edge training remain meaningful constraints.
Effect on the cycle: Verified evidence supports a more credible Chinese alternative stack, not proof of parity or broad commercial displacement. Inference: cheaper capable Chinese and open-weight models could compress model pricing and shift rents from frontier model owners toward cloud, chips and applications. The same lower costs could accelerate adoption and total compute demand, making the net investment-cycle effect ambiguous.
SOURCE LINKS AT PUBLICATION
Relevant reporting
These links point to third-party publishers. Availability, access and paywalls remain under each publisher’s control.
Oracle, Blue Owl project delay sends ripples through AI financing, sources say
Oracle’s power-related force-majeure notice for Project Jupiter reportedly delays completion by about a year, testing project-finance risk allocation while Blue Owl says commitments remain intact.
CREDIT · CAPEX · MACRO ↗Corporate bond buyers get picky with flood of AI debt
Investors reported greater selectivity and demand for concessions on AI-linked debt; this tests financing capacity rather than indicating imminent default risk.
CREDIT · CAPEX ↗AI profits still look far off, new analysis says
A cited academic analysis estimates a large cumulative gap between AI spending and revenue, reinforcing uncertainty over return timing; it is an external estimate, not company-reported financial guidance.
CAPEX · MONETISATION · VALUATION ↗Nvidia-backed Neocloud Nscale faces years of delay because UK's 'largest AI supercomputer' can't get enough power from the grid to feed its data centers
The reported UK power-connection delay is further evidence that grid availability can prevent announced AI capacity from becoming operating capacity.
MACRO · CAPEX ↗DeepSeek and Huawei release open-source Ascend AI programming tools to reduce reliance on Nvidia CUDA ecosystem
Open-source compute and communication libraries plus Ascend support for TileLang address a key software-portability constraint on Chinese accelerators; adoption and sustained utilisation remain unproven.
OPEN-MODELS · SEMIS ↗Alibaba Unveils Roadmap on Full-Stack AI Strategy from Chips, Cloud Infrastructure, Models to Agents
Alibaba announced new chips, Qwen model plans and a target above 20GW of operated global data-centre capacity by 2032; these are corporate plans, not realised capacity or returns.
OPEN-MODELS · SEMIS · CAPEX · MACRO ↗Huawei Cloud Rolls Out Enterprise AI Products Across the Board, Building an Open Agentic Cloud
Huawei reports more than 1,000 deployed projects in its Industry AI Foundry, offering directional evidence of domestic enterprise adoption but not independently audited monetisation.
MONETISATION · OPEN-MODELS · SEMIS ↗Alibaba Cloud’s 2026 Apsara Conference: Full-Stack AI Roadmap Along with Global Market Expansion Plan
Alibaba set out model, chip and cloud expansion plans, including a target above 20GW of global operated data-centre capacity by 2032; the target is aspirational rather than installed capacity.
OPEN-MODELS · SEMIS · CAPEX · MACRO ↗