NextFin

BofA Walks Away from Moonshot AI IPO as US Scrutiny of Chinese AI Deepens

Summarized by NextFin AI
  • Bank of America withdrew from discussions to serve as overall coordinator for Moonshot AI's Hong Kong IPO, signaling that US national-security scrutiny of Chinese AI firms is now outweighing deal fee income.
  • Moonshot AI is targeting a Q1 2027 Hong Kong listing at a roughly $50 billion valuation, aiming to raise up to $5 billion, with ARR growing from $300 million to $1 billion and on track for $2 billion by December 2026.
  • The exit reflects a structural regime shift, not a cyclical hiccup: US lawmakers and the White House have launched investigations and alleged covert model distillation, forcing banks to price political risk into cross-border AI transactions.
  • Remaining sponsors are CICC, Deutsche Bank and Goldman Sachs; further US bank withdrawals or a pricing below $40 billion would confirm the structural-risk thesis, while a deal at or above $50 billion would refute it.

NextFin News - Bank of America has dropped out of discussions to serve as an overall coordinator for Moonshot AI's Hong Kong initial public offering, a significant shift for one of the most anticipated technology listings of the coming year, as the Chinese artificial-intelligence company confronts intensifying national-security scrutiny in the United States. The talks never produced a formal mandate, and the bank is no longer part of the potential roster of lenders working on the deal, according to people familiar with the matter.

The withdrawal matters because it is not a routine reshuffling of a syndicate. Moonshot AI — the Beijing-based maker of the Kimi chatbot and the Kimi K3 open-weight model — has been racing toward a first-quarter 2027 listing in Hong Kong at a valuation of roughly $50 billion, after closing its final private funding round. Bank of America had been in discussions for the role of overall coordinator alongside sponsor banks China International Capital Corp., Deutsche Bank AG and Goldman Sachs Group Inc., and the company has already filed confidentially for the listing. Losing the US bank at this stage signals that the political cost of shepherding a Chinese AI champion onto a public market is starting to outweigh the fee income.

This is not a cyclical hiccup in dealmaking; it is a structural break in how cross-border capital flows around frontier AI. The mechanism runs through Washington, not Wall Street: as US investigators treat Chinese AI models as national-security assets rather than commercial products, American financial institutions are being forced to price political risk into transactions that, on paper, look like straightforward fee business. The immediate question for investors is whether Bank of America's exit is an isolated case or the first of several withdrawals that could compress Moonshot's valuation and narrow its path to a public market.

The Deal, the Bank, and the Timing

The facts, as they stand, are stark. Moonshot AI has been working toward a Hong Kong IPO that would raise as much as $5 billion, with early-look investor meetings potentially beginning as soon as this month. The company closed its last private round at a valuation of about $50 billion — up from $31.5 billion in a funding round this summer, and a steep climb from the $35 billion valuation it carried after raising $3.5 billion in July 2026.

Bank of America's role was to have been central. As overall coordinator, the bank would have managed the syndicate, steered the investor roadshow, and anchored the book-building process. That role never materialized into a signed mandate, and the discussions have now ended. The remaining sponsor banks — CICC, Deutsche Bank and Goldman Sachs — leave the deal with a distinctly non-US shape.

The timing is the telling detail. The withdrawal comes less than three months after Moonshot's Kimi K3 launch in July 2026, a release that put the company at the center of both commercial excitement and regulatory alarm. Kimi K3, a 2.8-trillion-parameter mixture-of-experts model with a one-million-token context window, was billed by Moonshot as the largest open-weight model ever released, with full weights made available in late July. The model's capabilities — and its open distribution — are precisely what drew the attention of US lawmakers.

On April 29, 2026, House Homeland Security Committee Chairman Andrew Garbarino and House Select Committee on China Chairman John Moolenaar announced a joint investigation into the national-security and cybersecurity risks posed by PRC-developed AI models, naming DeepSeek, Alibaba, Moonshot AI and MiniMax. The committees followed with letters to companies using or exposed to these models: Anysphere, the developer of the Cursor coding platform, and Airbnb. On July 31, 2026, a similar letter went to DoorDash after it acknowledged using a Moonshot AI model for lower-level work to cut costs.

The pressure is not confined to Congress. On July 22, 2026, the director of the White House Office of Science and Technology Policy stated that the US government has evidence that Moonshot AI may have operated a covert platform to conduct large-scale distillation against US AI models — the extraction of capability from American frontier systems through coordinated querying. Moonshot has pushed back specifically against the distillation allegation. Even so, the allegation, if it hardens into policy, would put any US financial institution facilitating Moonshot's public listing squarely in the crosshairs of Washington.

Why Banks Walk — and What It Means

The Mechanism: Political Risk Becomes Deal Risk

The surface reading is simple: Bank of America exited because the deal got politically complicated. The deeper mechanism is that the economics of cross-border investment banking have changed structurally. A mandate fee on a $5 billion IPO is large, but it is a one-time gain. The potential costs — congressional inquiries, regulatory examinations, reputational damage, restrictions on US government business, and the risk of being associated with a company accused of extracting capability from American AI systems — are recurring and potentially open-ended.

For a systemically important US bank, the asymmetry is unfavorable. The fee income is shared across a syndicate; the political exposure is borne alone. When the subject is a Chinese AI company at the center of an active congressional investigation, the rational calculus is to walk away before signing. Bank of America's exit is therefore best read not as a judgment on Moonshot's fundamentals, but as a repricing of the political risk premium attached to Chinese AI transactions.

The billions of dollars American companies invest in foundational research, compute infrastructure, and security engineering is being undercut by a sustained extraction campaign conducted at a fraction of the cost of independent development.

That line, from the April letter sent by Garbarino and Moolenaar to Anysphere, is the lens through which Wall Street is now viewing Chinese AI mandates. It is not the language of commercial competition; it is the language of national security. And once a transaction is framed in those terms, the fee is no longer the deciding variable.

Cyclical or Structural: This Is a Regime Shift

Is this a cyclical fluctuation that will revert once attention moves on, or a structural shift that will persist? The evidence points firmly to structural. Three tests separate the two.

First, a cyclical deal hiccup would be driven by temporary market conditions — weak investor demand, a bad earnings quarter, a volatile index. None of those apply. The driver is an active congressional investigation and an executive-branch allegation, both of which create durable policy risk rather than transient sentiment.

Second, a structural shift requires a change in the rules of the game that will not self-correct. The US treatment of Chinese frontier AI as a national-security issue, rather than a commercial technology category, is such a change. It is backed by a bipartisan congressional probe, a White House statement, and a pattern of letters to US companies that use Chinese models. None of these reverse on their own; they escalate or harden.

Third, the history that once made these deals routine no longer applies. For years, US banks routinely coordinated Hong Kong listings for Chinese technology companies. That playbook assumed a stable separation between commerce and national security in the AI sector. That assumption has been withdrawn.

The conclusion is uncomfortable for anyone betting on a smooth path to Moonshot's listing: the friction is not a wave that will pass. It is a new current that will shape every Chinese AI capital-raising event for the foreseeable future.

The Second-Order Effect: Pricing, the Syndicate, and the Market

The first-order effect is obvious: Moonshot lost a top-tier US coordinator. The second-order effects are more consequential, and they are what the market has not fully priced.

First, the remaining syndicate is now composed of CICC, Deutsche Bank and Goldman Sachs — a mix that is strong but politically exposed in different ways. Goldman Sachs, as a US bank, still faces the same scrutiny that pushed Bank of America out. If Washington's pressure continues, Goldman's continued participation becomes its own question mark. A further withdrawal would force Moonshot to rely on a narrower, more regionally concentrated syndicate, which typically means a smaller, more Asia-centric investor base and a lower clearing price.

Second, the valuation math is beginning to strain. Moonshot's annual recurring revenue has grown rapidly — from $300 million as of June to $1 billion currently, and the company is on track to reach $2 billion by December, according to people familiar with the matter. Even at $2 billion of ARR, a $50 billion valuation implies a 25x revenue multiple. That is demanding for any software company, and it is being asked of a company whose largest addressable market — the United States — is closing to it on national-security grounds. The multiple was underwritten on a world of open global growth; the world that now exists is one of fragmentation.

Third, the signal to other Chinese AI companies is chilling. Moonshot is not an obscure startup; it is one of China's most celebrated AI labs, backed by Alibaba, Tencent and 5Y Capital. If its path to a public market is being narrowed by US political risk, then every other Chinese AI company with cross-border ambitions faces the same headwind. The Hong Kong market may be booming — it raised HK$210 billion across 87 listings in the first half of 2026, the strongest first half in five years, and PwC expects HK$380 billion for the full year — but a deep primary market cannot fully substitute for a closed American one when the technology and capital are globally contested.

The Counter-Thesis

The strongest argument against this reading is that Bank of America's exit is a deal-specific negotiation outcome, not a geopolitical signal. Mandates are won and lost on fee terms, relationships, and internal bank priorities every day. Moonshot's confidential filing is still in place, its valuation has kept climbing, and the Hong Kong market is in the middle of a genuine boom. In this view, the company will list successfully in the first quarter of 2027 with a strong syndicate, and Bank of America's absence will be a footnote in the prospectus rather than a theme of the roadshow.

There is force in that argument. Hong Kong's IPO market has real momentum: in the first quarter of 2026 it was the world's top venue by funds raised, with 40 new listings bringing in HK$110.4 billion, the second-highest quarterly volume on record, and technology, media and telecommunications accounted for 55 percent of fundraising. Proceeds reached $41.2 billion through mid-August, more than double the figure from a year earlier. HKEX Chief Executive Bonnie Chan noted in August 2026 that more than 100 companies had listed that year, with proceeds exceeding $40 billion — already ahead of the $37 billion raised in all of 2025. Liquidity is not the problem.

But the counter-thesis rests on a distinction that is eroding: the line between a commercial listing and a national-security event. Bank of America did not lose a fee negotiation; it walked away from discussions entirely, and the stated context is US scrutiny of a Chinese AI company. When the reason a bank declines a mandate is the nationality and technology sector of the issuer, the event is political by definition. A strong Hong Kong market can absorb a smaller deal; it cannot neutralize a US government investigation into the issuer's core technology.

The falsifying signal is specific and observable: if Moonshot AI prices its Hong Kong IPO in the first quarter of 2027 at or above the $50 billion private valuation, with a fully allocated book and no further withdrawals from US banks in the syndicate, then the political-risk premium described here has not materialized, and Bank of America's exit was indeed deal-specific. If, instead, the deal is delayed, downsized, priced below $40 billion, or loses another US bank, the structural reading is confirmed.

What to Watch

The near-term watchlist is narrow and concrete. First, whether Goldman Sachs remains in the syndicate as Moonshot moves toward investor meetings — a second US bank exit would be a decisive escalation. Second, whether the House committees' investigation produces formal findings or policy recommendations targeting Moonshot specifically, rather than the sector generally. Third, whether the company's revenue trajectory holds — the $2 billion ARR target for December 2026 is the fundamental anchor that keeps the valuation defensible.

Scenarios diverge sharply. In the base case, Moonshot lists in Hong Kong in the first half of 2027, but at a valuation below the $50 billion private mark, with a syndicate led by CICC and Deutsche Bank and a US bank on the sidelines. In the upside case, the company hits its revenue targets, US scrutiny stays at the sector level rather than targeting Moonshot directly, and the deal clears near $50 billion on the back of Hong Kong's deep liquidity. In the downside case, further US bank withdrawals force a delay or a materially downsized offering, and the valuation resets toward the $31.5 billion to $35 billion range of earlier private rounds.

Short term, the news is a sentiment overhang on Moonshot and on Chinese AI listings generally. Medium term, the fundamentals — revenue growth, model capability, domestic market share — will reassert themselves, because Moonshot remains one of China's strongest AI franchises. Long term, the structural question dominates: a global AI industry in which the largest private market, the United States, and the largest growth market, China, are financially decoupled will produce two pricing regimes, not one. Moonshot's IPO will be priced in one of them.

For investors, the takeaway is not that Moonshot is a bad company — the funding history and revenue growth argue the opposite. The takeaway is that the best-funded, fastest-growing Chinese AI company can still be stopped at the border of the American financial system. Bank of America did not price the deal out; Washington priced it out. That is the new reality of frontier-AI capital markets, and it applies to every company on both sides of the line.

Explore more exclusive insights at nextfin.ai.

Insights

Why did BofA drop Moonshot IPO role?

What is Moonshot AI current valuation?

When is Moonshot Hong Kong IPO planned?

Who are Moonshot IPO sponsor banks?

What is Kimi K3 model capability?

Why does US probe Chinese AI firms?

How does distillation extraction work?

How does political risk change deals?

Is this shift cyclical or structural?

What happens if Goldman also exits?

How does US ban affect Moonshot value?

Is Hong Kong IPO market booming now?

What is Moonshot revenue growth rate?

Who backs Moonshot AI privately?

What defines frontier AI market pricing?

How does decoupling impact AI pricing?

How does this compare to DeepSeek probe?

Can HK market replace US capital access?

What are risks for Chinese AI listings?

How do US banks price political risk?

Search, ask or analyze anything...