Alibaba Is Selling 710 Million New Shares to Fund AI. Is Baidu Next?

Alibaba proposed on Sunday 23 August a Hong Kong placement of 710 million new ordinary shares at HK$112.70 — HK$80.0 billion, about US$10.2 billion — with all net proceeds going to chips, AI infrastructure, and model development and deployment. It is the largest primary follow-on ever by a Hong Kong-listed company on the company's and the press's own characterisation, and the third largest in the world this year after Alphabet and Intel. The shares are unregistered and offered offshore, so US investors cannot take part.

Three days earlier, our June-quarter analysis ended by asking whether capital expenditure had a stated ceiling — RMB 67.7 billion in one quarter, up 75%, with no guided envelope. Management has now put a number on the table. It bounds the funding, not the spending — and it prompts the question every Chinese platform holder asked within the hour, which is whether Baidu is next. The short answer, worked below, is no: Baidu is running the same build at a sixth of the scale, has three times the runway at its own burn rate, and has already paid for it with borrowed money.

The points

• Shares before: about 18.24 billion ordinary, the count behind the 2.28 billion ADS carried on our BABA page, at eight ordinary shares to the ADS. Third-party weighted-average counts run higher, 18.6 to 19.2 billion.
• Shares issued: 710 million ordinary, which is 88.75 million ADS-equivalent.
• Shares after: about 18.95 billion ordinary, or 2.37 billion ADS.
• Dilution: 3.89% of the pre-issue count, 3.75% of the enlarged one. On the higher third-party counts it is 3.7% to 3.8%. Nothing in the argument turns on which end is right.
• Price: HK$112.70 per ordinary share, which is HK$901.60 per ADS — about $115 depending on where in the 7.75–7.85 peg band the Hong Kong dollar settles. The 3.6% discount the terms quote is to Friday's ADR close near $119.34, not to the Hong Kong line's HK$123.00.
• In the June quarter Alibaba bought 13.4 million ordinary shares back for $162 million. This placement issues 53 times those shares and takes in 63 times those dollars, at an average buyback price of about $97 per ADS against roughly $115 here.
• It is a proposed placement announced on a Sunday. Size, pricing and completion can all still move; the book was reported oversubscribed and upsized.

What $10.2 billion buys

Alibaba reports in renminbi, and the dollar figures below are the company's own convenience translation in the June-quarter release, at RMB 6.7851 to the dollar. That is the right rate for comparing this raise to that quarter and is not a spot rate.

The largest follow-on in the history of the Hong Kong market covers one quarter of Alibaba's capital expenditure. Not one year — thirteen weeks, at a rate up 75% year on year that our analysis measured at 2.5× the quarter's entire adjusted EBITA and 25.2% of everything the company billed. Hold June's rate for four quarters and the capex bill is about $40 billion; this raise is a quarter of it.

The company did not need the money

Alibaba held $69.9 billion of cash and liquid investments at 30 June — enough, on our analysis, for roughly ten quarters of outflow at the June rate. The raise adds 14.6% to that, which at the same outflow rate is about six more weeks.

A company with ten quarters of funding does not sell equity at a discount to buy an eleventh unless it expects the outflow to grow. That is our inference and not the company's statement: management has given no capex ceiling, and has described the build as multi-year with a claimed two-to-three-year payback at current AI-product margins. The competing reading is that this is opportunistic — a book that was oversubscribed and upsized, taken up by sovereign wealth funds and long-only institutions, from a company that had been repurchasing stock at lower prices. Both survive the announcement. What the announcement settles is that the capex line is now large enough to fund externally by a company that could have paid cash.

So is Baidu next?

Not on these numbers. Baidu is the closest comparison available — a Chinese platform whose AI build blew up its cash line in the same June quarter, reported two days before Alibaba's — and the answer is that it is running the same play at a sixth of the scale, from a much softer cash position, and it has already funded it.

• Baidu's capex is one sixth of Alibaba's. RMB 11,390M against RMB 67,678M — though Baidu's own line tripled year on year and rose 93% in a single quarter, which is a faster ramp than Alibaba's +75%.
• Its cash pile lasts three times as long. RMB 283.1 billion ($41.7B) against an RMB 7,954M quarterly outflow is 35.6 quarters, close to nine years. Alibaba's much larger RMB 474.5 billion against a much larger outflow is 10.6 quarters. The company with the smaller balance sheet has the longer runway, and it is the one that has not raised.
• Baidu already raised the money — as debt. Short-term loans went from RMB 7.6 billion to RMB 26.3 billion and long-term loans from RMB 3.4 billion to RMB 20.8 billion since 31 December 2025, about RMB 36 billion of new borrowing, producing a RMB 17.4 billion financing inflow in the June quarter alone. Lease right-of-use assets doubled to RMB 18.6 billion — datacentre capacity.
• It is still buying its own shares back, $259 million since the start of the year under a $5 billion authorisation. Companies about to issue equity do not usually run both at once, and Alibaba's own placement issues 53 times what it repurchased in the quarter.
• An Alibaba-sized raise would be a third of Baidu. $10.2 billion against a market value of about $32.5 billion — 345.5 million diluted ADS at $94.23 as of 18 August — is 31% of the company. The proportionate gesture, one quarter of capex, is $1.7 billion.
• The capital-markets plumbing is being laid, which is not the same as a deal. Baidu's Hong Kong dual-primary conversion goes to an EGM on 26 August, and the meeting's mandates are reported to include a general authority to issue up to 20% of shares and a 10% repurchase mandate. Those are standard Hong Kong listing mandates. Twenty percent of $32.5 billion is about $6.5 billion — a ceiling on what could be issued without a further vote, not an intention to issue it.

The honest version of the answer: nothing on the tape says Baidu is raising equity. What the two companies share is the thing the score cannot see.

Neither raise shows up in the Rule of 40

Equity issued to build infrastructure lands on the financing line. So does debt drawn to build it. Neither touches free cash flow, so the Rule of 40 records neither Alibaba's $10.2 billion nor Baidu's RMB 36 billion of new loans — while recording every dollar of capex that both of them pay for.

That is why the score on these two looks the way it does. Alibaba joined coverage on 17 August at about +1 and printed −1.9 for the June quarter against the +0.9 to +7.2 band our preview had drawn; Baidu printed −24.3, eighteen points worse than the quarter before, on a free-cash-flow margin of −25.4%. In both cases the gap is capex, and in both cases the funding that makes the capex possible is invisible to the measure. The score tells you what the build costs. It does not tell you who can pay for it, and this week that was the more useful question.

What to watch

1. Whether the placement completes on these terms. Everything above is arithmetic on terms announced, not cash received.
2. The September-quarter capex line. The raise is sized at one quarter of June's. A materially larger print makes the funding logic read as a floor; a flat one favours the opportunistic reading.
3. Whether a spending number ever gets bounded. The question from three days ago is still open, and a funding envelope is not a capex envelope.
4. Whether the buyback continues alongside the issuance. Issuing at about $115 per ADS after repurchasing at about $97 is a statement about which price management thinks is which, and the next quarterly disclosure shows both sides.
5. Baidu's 26 August EGM and the mandates it approves. A 20% issuance authority is a ceiling, not a plan, but it is the instrument any future placement would use.
6. Whether Baidu keeps funding the build with debt. RMB 36 billion of new loans in six months at a company holding RMB 283 billion of cash is a choice about which lever to pull, and the next borrowing disclosure shows whether it is still the preferred one.
7. Whether the Kunlunxin chip unit files in Hong Kong. The listing is reported as a plan and nothing has been filed; a subsidiary listing raises money for the build without diluting Baidu shareholders, which is a different answer to the same question.

Sources and provenance: the placement terms are from Alibaba's announcement of 23 August 2026 and its term sheet, reported the same day by Reuters, Bloomberg and the South China Morning Post; the "largest-ever" and Alphabet/Intel rankings, the oversubscription and the buyers, the $119.34 Friday ADR close and the HK$123.00 Hong Kong close are press and company characterisations, as is management's payback commentary. Both companies' June-quarter figures — capital expenditure, free cash flow, cash and investments, Baidu's loan balances and the RMB 6.7851 and 8:1 conversions — are as each reported, captured in our Alibaba and Baidu June-quarter notes; Alibaba's cash balance is as of 30 June, so the +14.6% is arithmetic on a stale denominator. Prior-quarter ratios and Rule of 40 scores are from our Alibaba and Baidu analyses. Baidu's 20% issuance and 10% repurchase mandates and the Kunlunxin listing are reported rather than filed. Every ratio, dilution and runway figure here is ours: dilution on the 2.28 billion ADS this site carries (third-party counts of 18.6–19.2 billion give 3.7–3.8%), Baidu's market value at $94.23 as of 18 August on 345.5 million diluted ADS, and both runways at each company's own June-quarter outflow held flat.