When Tencent's free cash flow turned negative for the first time in Q2 2026 and Alibaba launched its first placement in seven years since listing, a key question has landed on bond investors' desks: when will the AI financing wave truly flow into China's onshore bond market?

On August 26, the fixed income team at Huatai Securities (including Zhang Jiqiang, Qiu Wenzhu, and Wen Chenxin) outlined in their report "AI Financing: From Cash Machine to Capital Machine" that overseas AI giants have shifted from being "cash machines" to "capital machines." While China's onshore bond market has no problem absorbing supply, the issue lies in insufficient financing demand. When supply does pick up, it represents more of an opportunity for investors — but the key variables that need to loosen will determine when this opportunity materializes. The report's calculations show:

China's AI industry chain currently has a relatively low share of debt financing, though financing demand from major domestic tech firms has increased since Q2 2026. Based on assumptions of 2%, 10%, and 20% compound annual growth rates for capex from 2026 to 2030, total domestic AI financing demand across the three scenarios would range from RMB 1.27 trillion to RMB 3.5 trillion, corresponding to new onshore bond supply of RMB 91.3 billion to RMB 260.8 billion, or annual supply of RMB 18.3 billion to RMB 52.2 billion. The bond market can easily absorb this, but the key is resolving bottlenecks around usage, currency, and tenor to unlock AI-related bond supply. In the near term, AI financing is more likely to bring structural changes to bond supply, affecting products such as tech-innovation bonds, panda bonds, REITs, and offshore bonds.

Turning-Point Signal: Financing Demand from Major Domestic Tech Firms Begins to Rise

In Q2 2026, Tencent Holdings Ltd (HKG: 0700) saw capital expenditure reach RMB 52.784 billion, up 176% year-on-year, primarily directed toward AI computing power procurement and infrastructure construction. As a result, single-quarter free cash flow fell to negative RMB 13.8 billion — the first time Tencent has posted negative quarterly free cash flow since it began disclosing the metric in Q2 2014. Meanwhile, on June 17, Tencent issued RMB 15 billion in dim sum bonds and USD 2.45 billion in dollar bonds in a single round.

Alibaba Group Holding Ltd (HKG: 9988) recorded capital expenditure of RMB 67.678 billion in Q1 of its fiscal year 2027 (calendar Q2 2026), up 75% year-on-year, with quarterly free cash flow falling to negative RMB 44.67 billion. On August 23, Alibaba announced plans to place HKD 80 billion in new shares, with all proceeds dedicated to AI construction — its first new-share placement since its Hong Kong listing in 2019.

Huatai Securities noted in its report: "Q2 2026 has already shown a trend of accelerating capex and rising external financing demand." This closely mirrors the overseas evolution path — except China remains at an earlier stage.

The US Blueprint: Three Stages Completed, Risk Is Migrating

Overseas AI financing has followed a clear three-step progression:

Stage One (2022-2023): The model validation period, where funding was dominated by strategic equity and cloud resource commitments. Investments by Microsoft into OpenAI, and by Google and Amazon into Anthropic, typify this phase. At the time, the six tech giants (Microsoft, Alphabet, Amazon, Meta, Oracle, and Nvidia) were still "cash cows," with combined operating cash flow rising from approximately USD 199.8 billion in 2019 to around USD 368.2 billion by 2023.

Stage Two (2024-2025): Capex began to consume free cash flow, and investment-grade bonds evolved from a low-frequency tool into a regular funding source. The six giants' capex-to-operating-cash-flow ratio rose from approximately 36% in 2019 to around 60% in 2025. In 2025 alone, they issued over USD 100 billion in investment-grade bonds — nearly three times the annual average from 2020 to 2024.

Stage Three (2026 to present): Full capital-stack mobilization. As of July 2026, the six giants had already issued approximately USD 220 billion in bonds this year, more than double their full-year 2025 total. Financing tools have expanded horizontally into equity, hybrid instruments, GPU asset-backed securities, project finance, SPVs, and revolving credit facilities — while vertically, different layers of repayment priority have been carved out.

The Huatai Securities report notes: "Risk is shifting from balance sheets into contracts, guarantees, and structures." A typical example is CoreWeave, which packaged computing power assets into collateralizable, ratable financing vehicles based on GPU hardware and long-term customer contracts — its USD 8.5 billion delayed-draw term loan ultimately secured an A3 investment-grade rating from Moody's. Another example is Nvidia, which provided approximately USD 250 billion in financing guarantees to OpenAI, transforming from chip supplier into credit provider.

China-US Gap: Not Just Stage, But Structure

The Huatai Securities report highlights three major structural differences:

First, the stage of development differs. Domestic leading tech companies' capex remains largely covered by internal funds. In 2025, eight major internet firms including ByteDance, Alibaba, and Tencent had combined capex of approximately RMB 455 billion — about one-sixth of the US six major cloud companies in the same period. The overall internal coverage ratio is approximately 90%, with the three major telecom operators near 100%.

Second, the investment entities differ. Overseas, market-driven hyperscale cloud companies dominate; domestically, private internet companies and central enterprise telecom operators lead, with third-party IDC providers and local state-owned computing platforms playing supporting roles.

Third, the financing structures differ. Internet giants cover approximately 50% of needs through internal cash flow, with debt financing dominated by offshore bonds (55%) and offshore syndicated loans (31%), while onshore bonds account for only about 1%. The three major telecom operators rely primarily on lease liabilities and have not yet issued bonds. Third-party IDC providers depend mainly on bank loans and are beginning to explore multi-tier REITs. Semiconductor companies also rely primarily on bank loans.

The Huatai Securities report points out that the constraints on onshore bond issuance are multi-dimensional: "The bulk of AI capex goes to high-end chip procurement, which is currently settled predominantly in US dollars... covering foreign-currency expenditures with onshore RMB financing creates FX conversion costs and cross-border capital risks." Additionally, Tencent, Alibaba, and others use red-chip VIE structures, creating institutional frictions for onshore bond issuance. Domestic private-enterprise bonds also tend to have relatively short tenors, with a thin buyer base for long-dated issues.

Core Calculations: Annual New Onshore Bond Supply of RMB 18.3-52.2 Billion

Huatai Securities estimated the financing needs for domestic AI construction from the investor side, covering 8 internet platforms, 3 telecom operators, 2 third-party IDC providers, and 4 semiconductor companies. Three scenarios were set with 2026 as the base year:

Across the three scenarios, cumulative capex from 2026 to 2030 would be approximately RMB 5.56 trillion, RMB 6.52 trillion, and RMB 7.95 trillion respectively, corresponding to annual averages of roughly RMB 1.11 trillion, RMB 1.30 trillion, and RMB 1.59 trillion. Based on these figures, cumulative funding gaps over 2026-2030 would be approximately RMB 1.08 trillion, RMB 1.94 trillion, and RMB 3.27 trillion, with cumulative financing demand of approximately RMB 1.27 trillion, RMB 2.15 trillion, and RMB 3.5 trillion respectively.

The report states: "Based on assumptions of 2%, 10%, and 20% compound annual growth rates for capex from 2026 to 2030, cumulative domestic AI construction financing demand is estimated at RMB 1.27 trillion, RMB 2.15 trillion, and RMB 3.5 trillion. AI financing demand will primarily be channeled through bank loans and onshore/offshore bonds. Across the three scenarios, cumulative new onshore bond supply from 2026-2030 would be approximately RMB 91.3 billion, RMB 158.7 billion, and RMB 260.8 billion. The domestic bond market is large enough to absorb this — the issue is insufficient financing demand, and supply growth represents more of an opportunity for investors."

Onshore bank loans account for over 70% of total financing through these two channels and remain the primary means by which the domestic financial system absorbs AI financing demand. Based on the financing structure assumptions above, the share absorbed by onshore bonds would gradually increase over the projection period, with average annual new supply of approximately RMB 18.3 billion to RMB 52.2 billion.

By entity type, internet platforms remain the largest source of onshore bond supply across all scenarios, contributing RMB 38.7 billion to RMB 123.2 billion, or 42%-47% of the total. The three telecom operators show the greatest elasticity, with their share rising from 7% to 23% in the high-growth scenario.

Where the Opportunities Lie: Four Product Categories Worth Watching

The Huatai Securities report notes that AI financing in the near term brings more structural change than aggregate impact.

Tech-Innovation Bonds: These are the best fit for AI financing needs. Since the launch of the "Bond Market Technology Board" in May 2025, tech-innovation bonds with labels such as "computing infrastructure," "token computing power factories," and "storage-computing integration" have been issued. However, new issuance remains concentrated in traditional sectors such as power, banking, and construction, with electronics, computers, and communications still underrepresented. The report concludes: "As hard-tech companies increase their debt financing, the 'tech content' of tech-innovation bond issuers is expected to improve."

Panda Bonds: These can absorb RMB financing needs from red-chip Chinese entities and foreign issuers. From January to July 2026, foreign-issuer panda bond issuance already exceeded full-year 2025 levels, with net financing of RMB 70.35 billion and the foreign share rising to 54%. Foreign tech companies have not yet participated, but if foreign AI supply-chain firms combine RMB financing with onshore computing investments, both the scale and issuer structure of the panda bond market stand to benefit.

Multi-Tier REITs: These address the exit and recycling of existing computing assets. In August 2025, the Nanfang Wanguo Data Center REIT and Nanfang Runze Technology Data Center REIT launched as the first batch of data center public REITs nationwide, raising a combined RMB 6.9 billion — both have since announced expansion plans. In the institutional REITs space, products from Century Internet, GDS Holdings, and Aofei Data are already in existence, with cumulative issuance of RMB 10.767 billion from March 2025 to August 2026.

Dim Sum Bonds / Chinese Dollar Bonds: The offshore market can absorb long-duration demand. In June 2026, Tencent issued a 30-year dim sum bond, with sovereign wealth funds and insurance companies subscribing for 47% of the issue — exceeding bank participation. The report notes: "Long-dated onshore credit supply remains scarce. If high-grade tech companies expand long-tenor credit issuance, it would help absorb insurance capital allocation demand."

Key Variables: Chip Bottlenecks and Institutional Loosening

The Huatai Securities report makes clear that a meaningful increase in onshore bond supply depends on two key variables:

First, progress in domestic advanced chip localization. Due to current export restrictions, the computing scale that Chinese AI companies can actually deploy is limited by chip availability — even if companies are willing to take on debt, capital cannot be fully converted into effective investment. The report states: "If breakthroughs are achieved in key technologies such as domestic GPUs, part of the financing demand currently absorbed by offshore syndicated loans and offshore bonds could shift to onshore channels." Approximately RMB 470 billion to RMB 1.2 trillion of offshore bond financing demand from internet platforms within the projection period has the potential to shift onshore.

Second, improved convenience for internet platforms issuing bonds onshore. This includes reducing institutional frictions for red-chip VIE structures, improving credit enhancement mechanisms for private-enterprise bond issuance, and cultivating a domestic buyer base for long-duration bonds.

The report concludes: "The shift in financing structure from equity-dominated to a mix of equity and debt will become an important variable for China's credit bond supply."