NextFin News - Alibaba Group fell 8% in Hong Kong trading Monday after pricing an HK$80 billion ($10.2 billion) share placement at an 8.4% discount to fund artificial-intelligence development, while SoftBank Group prepared a record 1 trillion yen retail bond sale for the same purpose and Bitcoin extended a rally that has lifted it more than $10,000 in two days. Three markets, three asset classes, one pressure point: the AI arms race has become a funding race, and investors are starting to charge for it.
The Three Moves, and Why They Belong Together
Alibaba priced 710 million new shares at HK$112.70 each in a placing expected to close August 26, according to the company's August 23 announcement. The price represented an 8.4% discount to the last close of its Hong Kong-listed stock and a 3.6% discount to Friday's close of its New York-traded depositary receipts. Shares dropped 8% in early Hong Kong trade Monday, with the American depositary receipts last changing hands near $120.
The deal is the largest primary follow-on offering ever by a Hong Kong-listed company and the city's biggest share sale since 2021, when Prosus sold $14.7 billion of Tencent Holdings shares. Globally, it ranks as the third-largest equity offering of 2026, behind placements by Alphabet and Intel.
The proceeds will fund AI development, extending a commitment Alibaba announced in October to invest 380 billion yuan ($56.54 billion) over three years in AI infrastructure. The capital raise arrives as profitability comes under pressure: for the June quarter, the company's net profit plunged more than 75% to 10.5 billion yuan ($1.6 billion), and it posted a free-cash outflow of $6.6 billion, reflecting rising spending on AI projects and computing infrastructure. Management said it has brought forward the projected payback period on AI investment to two and a half years from three years, citing surging demand for AI services.
In Tokyo, SoftBank Group is preparing roughly 1 trillion yen ($6.3 billion) of seven-year corporate bonds aimed at Japanese retail investors, its third retail issuance this year after April and June sales. The coupon is expected in the 4.3% to 4.9% range, with final terms set early next month. The sale would be the largest retail bond ever issued by a Japanese company, surpassing SoftBank's own 600 billion yen record set last year, and matches the 1 trillion yen record set by NTT Finance in 2020 when institutional bonds are included. SoftBank said the proceeds would repay existing corporate bonds and fund AI-linked investments, including a planned 10 trillion yen commitment to OpenAI by October and the acquisition of ABB's industrial robotics business.
Meanwhile, Bitcoin rose to $75,268 early Friday, up more than $10,000 in two days and crossing $75,000 for the first time since May. The cryptocurrency posted its strongest weekly gain since 2023, up more than 23% week over week, even as the S&P 500 fell 1.43% over the same period.
On the surface these are three unrelated stories. They are not. Each is a different answer to the same question: who will fund the AI buildout, at what price, and through which channel?
Equity, Debt, and Digital Gold: The Funding Channels Split
Alibaba chose equity. The 8.4% Hong Kong discount is not a rounding error; it is the price of speed. A block trade of this size cannot be worked slowly through the market without leaking, so management accepted dilution to lock in HK$80 billion immediately. The choice of Hong Kong over New York is also deliberate: it taps local demand for a Chinese technology name while sidestepping U.S. regulatory friction, and it aligns the funding base with the market where the core commerce business is listed.
SoftBank chose retail debt. Japanese households hold vast pools of savings, and with the Bank of Japan keeping yields near historic lows even as long-dated government bonds elsewhere surge, a 4.3% to 4.9% seven-year coupon from a familiar corporate brand is attractive to retail investors who rarely underwrite credit spreads the way institutions do. But the timing is pointed: SoftBank is asking households to fund an AI bet at the same moment institutional credit markets are showing strain.
Bitcoin chose to be the escape valve. Investors who do not want to fund the AI buildout at all — or who believe the funding will ultimately debase the currencies they hold — parked capital in the one asset whose supply cannot be expanded to meet demand.
The mechanism linking all three is capital intensity. Artificial intelligence has moved from a research budget line to a balance-sheet event. The players that want to compete must raise money, and they are raising it through whatever channel is cheapest or most available. The channel choice is itself information: equity issuance signals management thinks shares are fundable even at a discount; retail bond issuance signals institutions are being bypassed; a crypto rally signals a segment of the market is hedging the monetary consequences of the whole exercise.
"We're going to increase the size of the buyback. It could be more than $4 billion per issue," Treasury Secretary Scott Bessent said on August 20, a day after his department announced it would at least double the size of its liquidity-support buybacks of 10-year, 20-year and 30-year bonds, from $2 billion to at least $4 billion per session.
Credit Markets Are Starting to Push Back
The backdrop is a bond market that is no longer handing out cheap money without scrutiny. The 30-year Treasury yield recently touched its highest level since 2007, and U.S. federal debt passed $40 trillion. Credit-default-swap spreads on AI-linked names have widened materially this year. Nvidia's five-year CDS spread surpassed its late-July peak as SoftBank readied its bond issue, and newly issued bonds from Nvidia and SpaceX quickly slumped in secondary trading earlier this year, disappointing investors who typically flip such deals.
That is why SoftBank's retail channel matters more than the headline size. Retail investors price risk differently from institutions: they look at the coupon and the brand, not the spread. A successful 1 trillion yen sale would prove that Japanese household demand can absorb AI-infrastructure debt even as institutional markets grow more selective. A weak reception would send a different message: the cheapest funding source for AI is running dry, and companies will have to pay more for it.
The second-order effect is what most investors are missing. When tech giants and their financiers all raise capital at once, they are not just funding their own projects; they are competing for the same pool of savings. More supply of both equity and debt hitting the market simultaneously means prices adjust — dilution for equity holders, higher coupons for bondholders. This is not a liquidity problem at any single company. It is a repricing of what AI growth is worth across the entire complex.
The Treasury's intervention lit the fuse under Bitcoin and, indirectly, under the whole risk complex. On August 19, the Treasury Department announced it would "at least double" the size of its buybacks of long-dated bonds, in an effort to contain surging yields. Treasury Secretary Scott Bessent said current yields do not reflect market fundamentals and that liquidity in the 30-year bond is weak. The intervention worked in the near term: long-dated Treasuries rallied and yields pulled back. But it also carried a message that crypto investors heard loudly. When the fiscal authority steps in to suppress long-term yields while debt climbs past $40 trillion, the market begins to price what some economists call fiscal dominance — the government's borrowing needs subordinating monetary discipline. Bitcoin, long marketed as a hedge against currency debasement, rose on that logic.
The rally was amplified by policy optics. On August 20, President Trump hosted crypto executives at the White House and called on Congress to pass the CLARITY Act, industry-backed legislation that would establish legal guidelines for digital assets.
"We need Congress to take the next step by passing the CLARITY Act, a fair version of the CLARITY Act. It's a very, very powerful structured legislation which will keep us ahead of China, keep us ahead of everyone else," Trump said at the event.
The combination of Treasury action and regulatory optimism triggered a short squeeze. More than $1.25 billion of crypto short positions were liquidated in 24 hours, roughly $750 million of them bets against Bitcoin. Ether rose to $2,363, up more than 25% for the week, while Solana, BNB, XRP and Dogecoin gained between 6% and 19% in a day.
Cyclical Squeeze or Structural Repricing?
This is the call that determines whether today's moves are a buying opportunity or the first chapter of a longer repricing. Three pieces of evidence point to cyclical. First, the binding constraint is liquidity, not solvency: Alibaba remains profitable on a full-year basis, and SoftBank's bonds carry a coupon that retail investors can underwrite. Second, policy can relieve pressure quickly — the Treasury's buyback move pulled yields down within a day, and a Federal Reserve rate cut would lower the discount rate on the entire AI complex. Third, the moves are concentrated in funding channels, not in operating failures; no major AI player has missed on its technology roadmap.
But the structural counter-argument is harder to dismiss. The capital intensity of AI is not a one-year phenomenon; it is a multi-year buildout with uncertain returns. Every major player is raising money at once, which means supply of both equity and debt is hitting the market simultaneously. When supply surges and demand is finite, prices adjust — and the adjustment shows up as dilution for equity holders and higher coupons for bondholders. That is not a liquidity problem; it is a repricing of what AI growth is worth.
The most likely read is a cyclical wave riding on a structural shift. The funding pressure will ease when liquidity improves or when Treasury action calms the long end — but the underlying demand for capital will not disappear until the AI buildout matures. Investors should treat relief rallies as tactical, not as evidence that the repricing is over.
The Strongest Case Against This Read
The counter-thesis is that the market is overreacting to normal capital-raising. Alibaba's HK$80 billion placement is a small fraction of its roughly $2.4 trillion market capitalization; SoftBank has sold retail bonds successfully before, including a 600 billion yen deal last year; and Bitcoin's rally is a reflexive bounce off oversold conditions, not a regime change. On this view, the 8.4% discount is a standard block-trade concession, and the bond and crypto moves are noise around Nvidia's upcoming earnings.
That argument has force, but it rests on one assumption: that AI spending converts to cash flow on the timelines management has promised. Joseph Brusuelas, chief economist at RSM, framed the risk in a note this week, calling the Treasury's buyback expansion "what fiscal dominance looks like as the fiscal authority leans on the central bank to subordinate its goal of price stability to the government's borrowing and political needs." He added that such steps "will prove to be a temporary salve to an open financial wound of our own making." The wound, in his reading, is a deficit trajectory that requires $739 billion of new borrowing in a single quarter while consumer prices rose 3.4% over the year through July, well above the central bank's 2% target.
If Brusuelas is right about the fiscal backdrop, then the funding pressure is not cyclical at all — it is the market adjusting to a world where government borrowing crowds out private capital and forces higher returns everywhere. If Alibaba's 2.5-year payback claim holds, today's dilution is cheap. If it slips to four or five years, today's discount is the first of several.
The falsifying signal is concrete: Alibaba's next two quarterly reports on cloud revenue growth and free cash flow. If cloud revenue fails to sustain the triple-digit AI-driven growth the company has reported, or if free cash flow remains deeply negative through the December quarter, the structural-repricing thesis strengthens and the cyclical-relief case weakens. For the bond complex, watch the SoftBank retail sale's subscription rate when it prices early next month; a weak reception would confirm that the cheapest funding channel is closing.
What to Watch Next
Three catalysts will set the tone. First, Nvidia reports earnings on Wednesday, and its guidance will determine whether AI spending is still viewed as growth investment or balance-sheet strain. Second, the reception of SoftBank's retail bond — expected to price early next month — will test whether household demand can underwrite AI debt at these coupon levels. Third, whether Bitcoin can hold above $75,000 will determine whether the crypto rally was a short squeeze or the start of a sustained move. Broader macro risks also loom: oil slipped to about $91 a barrel amid falling U.S.-Canada trade talks, and the Treasury is set to outline further economic pressure on Iran.
Short term, expect volatility. Alibaba's 8.4% discount sets a reference point for other Chinese technology names considering raises, and any weakness in Nvidia could pull the entire AI complex lower. Medium term, the funding question dominates: companies with strong free cash flow can self-fund AI; the rest must dilute or borrow at higher cost. Long term, the winners will be those whose AI investments actually produce the cash flows they have promised.
The base case is that funding pressure eases into year-end as liquidity improves, but the cost of capital for AI-linked companies stays structurally higher than it was in 2023. The upside case is that AI revenue inflects faster than expected, validating today's spending and compressing spreads. The downside case is a slower payback that forces more dilution and pushes retail bond coupons higher.
The AI boom is no longer just a technology story. It is a capital-market story, and the market has just started charging for it.
Explore more exclusive insights at nextfin.ai.
