Holding nearly RMB 475 billion in cash and liquid investments, with operating cash flow up 11% year-on-year, Alibaba announced plans over the weekend to place HK$80 billion in new shares. The immediate reaction might be: is Alibaba short on cash? But that is not the most accurate question. What investors should really ask is: when AI capital expenditure—large outlays for long-term assets like chips, servers, and data centers—begins to clearly exceed operating cash flow, should the company draw down existing cash, take on more debt, or let new shareholders share the investment risk?
Choosing a different financing method reflects not just whether cash on hand is sufficient, but also management's judgment on the cost of capital, future returns, and current valuation. The HK$80 billion placement is not a "proof of cash shortage" but rather a test of capital allocation: Alibaba is turning the technological race in AI into a cash flow and return-on-investment question that shareholders must jointly bear.
Four Seemingly Contradictory Numbers
In the June 2026 quarter, Alibaba's revenue reached RMB 268.953 billion, up 9% year-on-year, while operating cash flow was RMB 22.945 billion, up 11%. However, capital expenditure in the same quarter hit RMB 67.678 billion, up 75%, resulting in a negative free cash flow of RMB 44.670 billion. At quarter-end, the company still held RMB 474.505 billion in cash and other available liquid investments. Three days later, Alibaba announced a proposed placement of HK$80 billion in new shares in Hong Kong, explicitly stating that all net proceeds would be used for full-stack AI capabilities and AI infrastructure.
+RMB 22.9 billion operating cash flow: daily operations still generate cash, up 11% year-on-year; -RMB 44.7 billion free cash flow: after deducting operating-related capital investments, a clear cash gap emerges; RMB 67.7 billion quarterly capex: roughly 2.95 times the quarterly operating cash flow; RMB 475 billion in cash and other liquid investments at period-end: substantial, but not all of it can be distributed to shareholders.
To simplify the three cash flow terms: operating cash flow shows "how much cash daily business brings in"; capital expenditure shows "how much is being spent for future growth"; and free cash flow, which is the former minus the latter, is closer to the cash truly left after investment. These four numbers coexisting is not contradictory. Operations are still profitable, but capital expenditure is consuming cash at a faster pace; the balance sheet remains solid, but funding needs for the coming years are being pulled forward. The placement is not about "whether we can pay bills today," but about who will bear the next phase of AI investment without quickly eroding the cash buffer.
Capital Expenditure Won't Immediately Squeeze Profits, But It Will Immediately Drain Cash
Media reports often simplistically attribute Alibaba's 75% drop in net profit to "excessive AI capital expenditure." This makes some intuitive economic sense but is not accurate from an accounting perspective. When purchasing servers, chips, and data center equipment, cash has already flowed out, but this expenditure typically first enters the balance sheet as fixed assets and does not hit the income statement all at once in the current period. Depreciation is then recognized gradually as the equipment is used. Therefore, the decline in current net profit and the deterioration of free cash flow must be explained separately.
Paying for AI equipment leads to an immediate cash outflow in investment activities, then forms fixed assets on the balance sheet, and only later enters the income statement through gradual depreciation. The income statement answers: how much was earned this quarter? The decline in net profit mainly stems from shrinking operating profit, goodwill and investment impairments, EU fine provisions, and reduced book gains from investment disposals and asset price changes (fair value gains). Capital expenditure itself does not write down profits in full at once. The cash flow statement answers: how much was spent this quarter? Operating cash flow is positive, but purchases of operating-related property and equipment reached RMB 67.660 billion, directly pushing free cash flow into negative territory. Cash pressure is exposed earlier than the current income statement.
Alibaba's quarterly operating profit fell from RMB 34.988 billion to RMB 15.161 billion, with operating margin dropping from 14% to 6%; GAAP net profit fell from RMB 42.382 billion to RMB 10.444 billion. After temporarily removing share-based compensation, investment gains and losses, goodwill and intangible asset impairments, non-GAAP net profit still fell 38% to RMB 20.715 billion. Non-GAAP can be understood as "removing some one-off and non-cash noise first, then looking at how much the main business earned," but it is not cash, nor does it mean the excluded costs don't really exist. Even under this measure, core earnings remain under significant pressure, just not as severe as GAAP profits show.
Product development expenses rose from RMB 15.001 billion to RMB 22.529 billion in the quarter; property and equipment depreciation, impairments, and related land lease costs increased from RMB 6.891 billion to RMB 11.814 billion. The former more reflects current R&D and infrastructure operating investment, while the latter begins to show past capital expenditure gradually entering the income statement. The economic cost of capital expenditure has not disappeared; it has simply shifted from "spending cash all at once today" to "recognizing expenses gradually over the next few years."
From RMB 22.9 Billion Operating Cash Flow to Negative RMB 44.7 Billion Free Cash Flow
Operating cash flow of RMB 22.945 billion being higher than GAAP net profit of RMB 10.444 billion cannot be directly interpreted as "stronger cash profitability." Operating cash flow adds back depreciation, impairments, share-based compensation, and other expenses that did not involve cash outflows in the quarter, and it is also influenced by payment timing—when customers pay and when the company pays suppliers. Therefore, single-quarter operating cash flow should be viewed together with profit, changes in receivables and payables, and multi-quarter trends. Free cash flow goes a step further, testing whether cash generated from daily operations can cover the real money spent for future growth.
This bridge shows that Alibaba is not "failing to generate cash from operations," but rather that operating cash flow is insufficient to cover the current level of investment intensity. In the quarter, for every RMB 1 of operating cash generated, approximately RMB 2.95 was simultaneously invested in capital expenditure, with the gap filled by existing cash, asset disposals, borrowings, convertible securities, or new equity financing.
More notably, Alibaba's cash and other liquid investments stood at RMB 520.824 billion at the end of March 2026, but fell to RMB 474.505 billion by the end of June—a quarterly decline of RMB 46.319 billion, roughly matching the free cash flow gap. If we assume outflows continue at RMB 44.670 billion per quarter, existing liquidity could cover about 2.7 years. However, this is only a rough extreme scenario: server purchases won't be identical each quarter, and e-commerce collections have peak and off-peak seasons.
With Cash Available, Why Issue New Shares Instead of Using It?
"Having cash" and "should spend all the cash" are two different things. Alibaba has previously committed to investing at least RMB 380 billion in cloud computing and AI infrastructure over the next three years—a plan larger than the total related investment over the past decade combined. Using the exchange rate at end-June, HK$80 billion is roughly RMB 69.5 billion, only about 18% of the RMB 380 billion plan, and roughly enough to cover one quarter of capital expenditure at current intensity.
Preserving a strategic safety cushion: chip procurement, data center construction, and power resources require advance commitments. If the company relied entirely on cash on hand, consecutive quarters of free cash flow gaps would quickly weaken its ability to respond to competition, regulation, and M&A opportunities. Matching long-term AI infrastructure investments with equity that never needs to be repaid: returns may take years to materialize and are uncertain. Debt requires periodic interest payments and principal repayment at maturity; equity financing has no fixed repayment date and is better suited for investments with slow, volatile returns. Capitalizing on the current capital market window: when the market is willing to pay a higher valuation for AI growth, issuing shares can reduce cash burn. However, whether this is favorable for existing shareholders depends on whether the issuance price is below the company's intrinsic value.
Thus, the placement can be understood either as prudent or as management picking a more favorable financing window. If management believes the stock is severely undervalued yet still issues a large number of new shares at a discount, it means transferring part of future value to new investors. If the stock already fully reflects optimistic AI expectations, swapping highly valued shares for capital that never needs to be repaid could be a worthwhile deal. The key is not that "placement is necessarily negative" or "financing is necessarily positive," but rather comparing two costs: on one hand, approximately 3.5% equity dilution—simply put, the same profit must now be distributed across more shares, reducing each existing shareholder's share; on the other hand, the cost of continuing to burn cash, increasing debt, and missing the AI infrastructure window.
Is AI Investment Burning Cash, or Already Generating Returns?
Looking only at free cash flow would underestimate the operational improvements underway. Alibaba's AI cloud and computing services revenue reached RMB 48.437 billion, up 45% year-on-year; segment adjusted EBITA rose from RMB 2.419 billion to RMB 5.628 billion, up 133%, with margin improving from approximately 7.2% to 11.6%. "Adjusted EBITA" is the profit measure the company uses to assess the day-to-day operating performance of a business segment—roughly, setting aside interest, taxes, and some one-off or non-cash items to see whether the business itself is profitable. These numbers show that AI infrastructure is not just about spending; cloud commercialization is happening, and the benefits of unit cost declines as scale expands are beginning to show.
But the other side of full-stack AI remains expensive. AI lab and application revenue was only RMB 3.338 billion, up 16% year-on-year, yet under the same segment profit measure, losses expanded from RMB 3.224 billion to RMB 13.861 billion, mainly driven by AI capability investment and rising inference costs for Tongyi applications. Inference cost is the computing cost incurred each time a model processes a user request. In other words, the cloud and computing layer is already showing profit improvement, while the model and application layer continues to burn profits at scale.
This shifts the investment judgment from "is there demand for AI" to three more rigorous questions: Can new computing capacity maintain high utilization? Can cloud business profit growth outpace depreciation growth? Can application losses narrow as user scale, payment rates, and inference costs improve? Only if these three are progressively delivered will today's capital expenditure transform from "cash consumption" into "high-return reinvestment."
How Much Dilution Will the HK$80 Billion Placement Cause? What Has the Current Stock Price Already Priced In?
As of August 23, Alibaba's official announcement only confirmed the proposed placement of HK$80 billion in total, without disclosing the final share count and price. Reuters obtained a term sheet showing a proposed issuance of 710 million shares at HK$112.70 per share. The following dilution calculation is therefore a scenario analysis based on the term sheet, not the final transaction result.
EPS is earnings per share—"how much profit is allocated to each share on average"; diluted EPS also factors in potential shares that could be added. The company used approximately 19.324 billion weighted shares in calculating non-GAAP diluted EPS for the June quarter. Using this as an approximate denominator, the additional 710 million shares represent 3.67% of existing shares, or approximately 3.54% of post-issuance shares. If we consider only the increase in share count, temporarily ignoring the future earnings the proceeds might generate, EPS would decline by about 3.5%. In the quarter, the company repurchased only 13.4 million shares, so the proposed issuance is roughly 53 times the quarterly repurchase volume.
Approximately 3.5%: if profits remain unchanged for now, the profit per existing share declines by roughly this proportion; approximately RMB 69.5 billion: proceeds converted at the end-June exchange rate, equivalent to about 1.56 times the quarterly free cash flow gap; approximately 53 times: the proposed issuance relative to shares repurchased in the June 2026 quarter.
Starting with the simplest absolute valuation anchor: Alibaba's quarterly non-GAAP diluted EPS was RMB 1.07, or approximately HK$1.23. Multiplying a single quarter by four gives approximately HK$4.92; assuming the placement completes with profits unchanged, EPS would be roughly HK$4.75. In other words, HK$4.75 is just a working draft of "one full year based on current quarterly performance post-placement," not a company forecast. At the August 21 Hong Kong closing price of HK$123, the stock trades at approximately 25.9 times this rough annualized profit. That is, investors are willing to pay about HK$25.9 for every HK$1 of such annualized profit. The market clearly does not treat Alibaba as a mature e-commerce company without growth; it is paying in advance for AI profit growth.
A "reverse valuation" approach: instead of guessing a target price, ask the reverse question—since the current price is HK$123, if investors want a 10% annual return, how much must Alibaba's EPS grow to in three years for the current price to make sense? The calculation starts from the post-placement rough annualized EPS of HK$4.75; the current price is the August 21 closing price of HK$123; assuming investors want a 10% annual return and sell after three years at different P/E multiples. The calculation excludes dividends and does not treat a single quarter multiplied by four as company guidance.
Assuming the market gives a 22x P/E in three years and investors want a 10% annual return, the current HK$123 price requires Alibaba's EPS to reach approximately HK$7.44 in three years, implying average annual growth of about 16%. If profits stay at current levels and the valuation falls to 18x in three years, the current price cannot deliver sufficient returns. If AI cloud profits continue high growth, application losses narrow, and EPS grows around 20% annually, the current valuation could be digested. The term sheet price of HK$112.70 demands slightly less: under the same 22x P/E and 10% return requirement, EPS only needs to grow at an average of about 12.8% over three years. This explains why a discounted placement might attract long-term capital—new investors buy cheaper and require lower profit growth from the company; this discount is borne by existing shareholders through issuance discount and reduced ownership percentage.
Data verification note: Reuters reports the term sheet price of HK$112.70, stating a 3.6% discount to the recent closing price; however, public market data shows Hong Kong shares closed at HK$123 on August 21, implying a discount of approximately 8.4% by direct calculation. Since the company's official announcement has not disclosed the price, this article uses HK$112.70 and 710 million shares only as a provisional scenario, with the discount calculated independently using the clearly stated HK$123 closing price.
What Does This Financing Really Represent? Three Sets of Evidence to Wait For
Hypothesis one: proactively preserving firepower. Demand is confirmed, new computing capacity utilization is high; financing is simply to avoid burning through cash too quickly. Verification signals: whether cloud revenue, segment operating profit, and operating cash flow growth can consistently outpace depreciation and capital expenditure. Hypothesis two: capitalizing on the valuation window. Management believes issuing shares now costs less than continuing to burn cash or taking on debt. Verification focus: whether returns from new AI investment exceed the minimum return shareholders require for risk, and whether buyback policy changes significantly. Hypothesis three: internal cash is tightening. AI investment payback is slower than expected, application losses and capital expenditure remain high for an extended period, and external financing becomes necessary to maintain competitiveness. Verification signals: continued free cash flow deterioration, declining cash, rising debt, and additional financing rounds.
In coming quarters, more than revenue growth, the metrics worth tracking are: AI cloud revenue and segment operating profit (adjusted EBITA): can revenue growth remain above capital expenditure growth, and can margins keep improving; AI lab and application losses: are inference costs declining, and can revenue growth catch up with investment growth; capital expenditure and depreciation: as depreciation accelerates into the income statement, can cloud profit improvement offset new costs; operating cash flow and free cash flow: does operating cash flow growth come from real profits and customer collections rather than short-term factors like delayed supplier payments; cash, debt, and subsequent financing: is HK$80 billion enough to support the stated plan, or will a second round of equity or debt financing emerge; buybacks and new shares: if the company issues discounted shares while buying back at high prices, the capital allocation logic needs re-examination; incremental ROIC: for every RMB 100 newly invested in AI, how much after-tax operating profit is ultimately generated; if it remains below the cost of financing for a long time, larger scale may become increasingly uneconomical.
RMB 475 Billion Is a Safety Cushion, Not the Answer
Alibaba holding nearly RMB 475 billion in liquidity yet proposing a placement cannot be simply explained as "short on cash," nor can the necessity of financing be dismissed based on cash on hand. The most reasonable interpretation is: AI investment has already extended from R&D expenses on the income statement to equipment on the balance sheet, and further penetrated into a free cash flow gap; the company is proposing to share this high-intensity investment period with new equity. Current evidence is not one-directional. AI cloud revenue and profits are growing rapidly, indicating commercialization is happening; but expanding model and application losses, with capital expenditure reaching nearly three times operating cash flow, also show returns have not yet covered full-stack investment. If the HK$80 billion placement completes as planned, it will ease short-term cash pressure while handing approximately 3.5% static dilution to existing shareholders.
Therefore, the real question investors need to answer is not "does Alibaba have money," but: over the next three years, can AI investment keep non-GAAP EPS growing at roughly 16% or more annually and ultimately turn free cash flow positive again? If the answer is yes, the placement may be using equity capital that never needs repayment to buy long-term growth; if the answer is no, it simply postpones the problem of insufficient cash by reducing the per-share ownership of existing shareholders.