Alibaba's recent share placement has sparked widespread skepticism about the economics of China's AI capital expenditure cycle, but JPMorgan believes the market's response far exceeds what the actual cost warrants.

On August 24, Alibaba completed a placement of 710 million new shares at HK$112.70 per share, raising HK$80 billion (approximately US$10.3 billion), with the issue price set at an 8.4% discount to the previous trading day's closing price. The following day, Hong Kong-listed shares tumbled 8.5%, while Tencent and the Hang Seng Tech Index fell 2.2% and 2.1%, respectively. JPMorgan estimates that only about 0.3% of Alibaba's share price decline on that day can be attributed to the placement discount itself; after stripping out peer-driven contagion effects, the abnormal repricing borne by existing shareholders amounts to roughly US$19 billion—approximately twice the size of the funds raised.

JPMorgan believes Alibaba's AI investment cycle may run longer than previously expected, but the economics of the associated computing power remain attractive, and the funding risks across the three major cloud service providers differ materially. The bank maintains "Overweight" ratings on Alibaba, Tencent, and Baidu, and advises investors to buy Alibaba shares ahead of the September quarter earnings release.

Market Pricing Implies "Value Destruction," JPMorgan Calls It Excessive

JPMorgan views the placement mechanism itself as financially straightforward. With 710 million new shares priced at HK$112.70, the theoretical ex-rights price stands at HK$122.63, implying a mechanical value transfer to new subscribers of only about US$900 million (approximately 0.3%). However, the closing price on August 25 fell to HK$112.50, meaning the market pushed Alibaba's pro-forma market capitalization further down from roughly US$286 billion (after including the US$10.3 billion raised).

JPMorgan notes that attributing the entire abnormal price movement to this placement would imply a valuation loss of approximately RMB185 for every RMB100 raised; even if all proceeds were completely destroyed, it would only result in a market value reduction of about US$10 billion.

According to the bank's annual model, every RMB100 invested in infrastructure generates a positive net present value of RMB36 (RMB22 under a 60% infrastructure allocation baseline). Therefore, JPMorgan interprets the bulk of this abnormal share price movement as a repricing of the duration and scale of Alibaba's broader capital expenditure program, rather than a rational response to the placement itself.

Four Tests to Deconstruct Placement Economics

JPMorgan breaks down the placement-related debate into four independent tests, arguing that conflating these issues is the primary cause of market pricing confusion.

Can computing power create economic value? Yes. The bank's model assumes RMB1 of annual revenue per RMB1 of capital expenditure at 100% utilization, a 45% incremental EBITDA margin, and 5-year straight-line depreciation. The calculation yields a 22% project internal rate of return, a 2.9-year payback period, and RMB36 of net present value (at a 10% discount rate)—broadly consistent with management's stated "payback in under 3 years and comprehensive ROIC above 13%." The limiting factor lies in the use of proceeds: Alibaba described the purpose of this fundraising as "full-stack AI capabilities" including infrastructure, but spending on model training and application inference currently lacks auditable returns, so the bank uses 60% allocated to infrastructure as its base case.

Does the placement enhance earnings per share? It depends on the infrastructure allocation. In scenarios where infrastructure accounts for 80% or more, the placement accretes EPS starting from the second year; under the 60% base case, EPS dilution is projected at approximately 0.5% in fiscal year 2028, turning roughly neutral thereafter. The mature-stage ROIC threshold required for EPS accretion by year three is approximately 15%.

Is equity the cheapest financing option? No. Substituting debt financing would save approximately 3 to 3.5 percentage points of EPS drag across all scenarios, as RMB bond coupons of 2% to 3% are far below the 12% earnings yield. JPMorgan believes the rationale for equity financing lies in its provision of perpetual capital without refinancing needs, offshore currency for overseas procurement, and maintained rating headroom—but these benefits do not come without cost.

Does it increase intrinsic value per existing share? No. Using HK$205 as the pre-transaction fair value, post-transaction per-share value declines by 0.9% and 1.2% under 100% and 60% infrastructure allocation scenarios, respectively. Only when the pre-transaction fair value is below approximately HK$153 (at 100% allocation) or HK$137 (at 60% allocation) would the placement add value to existing shareholders.

Three Funding Paths, Each with Its Own Cost

JPMorgan outlines the distinct funding logic that Alibaba, Tencent, and Baidu each employ to address AI capital expenditure, pointing out that each path carries corresponding shareholder costs, and not all costs are equally visible.

Alibaba faces the largest structural funding gap. Based on RMB200 billion in capital expenditure, the annual internal funding shortfall is approximately RMB100 billion; if annualized from the June quarter's capex pace to RMB270 billion, the gap widens to RMB170 billion. This placement covers roughly 8.4 months of the gap (base case) or 5 months (stress case). The company also holds approximately RMB150 billion in discounted liquidity and RMB borrowing capacity at 2% to 3% interest rates, so liquidity and solvency are not critical risks. The bank expects more debt financing and potential cloud-segment fundraising over the next 12 months, but a second placement at the parent level would be viewed as a negative signal of deteriorating returns.

Tencent can largely self-fund its incremental capital needs. Annualized first-half operating cash flow stands at approximately RMB310 billion, while the bank forecasts full-year capital expenditure of RMB200 billion, leaving the company broadly balanced. Tencent also holds RMB487 billion in listed investments and RMB388 billion in unlisted investments, totaling an RMB875 billion portfolio that serves as a significant funding buffer. Its cost is implicit: foregone portfolio returns, tax frictions on asset disposals, and market impact. JPMorgan sees parent-level risk as near zero.

Baidu has the weakest internal funding capacity. June quarter operating cash flow was only RMB3.4 billion, while capital expenditure reached RMB11.4 billion, with the funding gap primarily filled through short-term and long-term loans. The parent company's low valuation and reluctance to dilute existing shareholders push the equity financing path to the subsidiary level—namely, Kunlun Chip. The cost of this path is a reduction in the parent's ownership stake in AI assets, though the scale is constrained by Kunlun Chip's fundraising capacity.

Capital Expenditure Trajectory and Earnings Absorption Capacity

JPMorgan maintains its fiscal year 2027 capital expenditure forecast for Alibaba at RMB200 billion, with RMB270 billion as a stress scenario. The bank notes that the financing event occurred five days after earnings, indicating that management expects capital expenditure to remain elevated for more than one procurement cycle, which alters the assessment of investment cycle duration but does not change specific annual figures.

Returns have a scale ceiling. Under unchanged unit economics, larger project scale creates more value, but the risk lies in supply expansion depressing pricing: 80% utilization combined with a 10% price cut corresponds to approximately 11% returns, while 70% utilization with a 20% price cut leaves only about 3%. This is precisely why RMB270 billion is viewed as a risk scenario rather than an upside scenario.

At the earnings absorption level, JPMorgan calculates that total depreciation from one year of base-case capital expenditure represents approximately 28% of Alibaba's expected fiscal year 2027 net profit, 13% for Tencent, and 23% for Baidu. However, incremental EBITDA exceeds depreciation from the first full year onward, so the primary risks stem from EBITDA realization, utilization, and pricing, rather than the depreciation burden itself.

Upcoming Catalysts and Risk Boundaries

JPMorgan lists several key near-term milestones. The August 26 placement settlement removes technical uncertainty; Alibaba's September quarter results are the primary checkpoint, with key metrics including whether external cloud revenue growth exceeds 50%, whether AI lab losses narrow quarter-over-quarter, and whether capex pacing aligns with the financing scale. Tencent's Hunyuan 4 and Baidu's Kunlun Chip launches serve as respective milestone validations for each company.

The bank also specifies risk triggers that would change its view: if Alibaba's three-year capital expenditure commitment significantly exceeds RMB380 billion without a corresponding upward revision to cloud revenue guidance; if raised funds remain largely parked in cash rather than fixed assets after more than 12 months; if AI lab losses fail to decline quarter-over-quarter in the September quarter; if computing power prices continue to slide as supply loosens; or if another parent-level equity placement occurs within 12 months—any single trigger would weaken the current investment thesis.

Alibaba's shares currently trade at just 12 times fiscal year 2027 earnings and 8 times fiscal year 2028 earnings. JPMorgan believes the market has over-discounted the negative signals from this placement and advises investors to buy on weakness ahead of the September quarter earnings release.