Vol.67 The BP for a Hot Project May Be Fake

Summary

• The value of the Manus episode lies not in the financing rumor itself, but in the fact that a BP deemed potentially fake when Manus was asked to verify it continued circulating among real investors for an extended period. The $100M Benchmark financing rumored in May was not confirmed by Butterfly Effect; the two BPs, one aimed at dollar institutions and the other at RMB funds and state-owned capital, were also deemed potentially fake when verified. 张楠’s view: “Even if it is wrong, some form of consensus has already formed behind it.”
• The documents look more like sales materials assembled by intermediaries to raise money than a financing narrative produced by the Manus team. Both BPs are garish, awkwardly written, and heavy on market data, trends, and use cases, but light on endogenous information about the team, company history, next steps, financing cadence, and thinking. 庄明浩 said a PPT generated by an off-the-shelf AI from a simple prompt “should be far more professional than the version we are looking at now.”
• Wild BPs often revolve around allocations in hot projects, while layers of intermediaries defer the liquidity problem into LP ownership and repayment risk. Even if local state-owned capital or an intermediary secures an allocation, it may not have the cash on hand and must raise from private capital or individual LPs; the ultimate investor, investment vehicle, and portfolio company may be separated by “three to five layers.” Similar SPV transactions are active around OpenAI, SpaceX, and Anthropic, but how indirect holders realize their returns after exit remains “a massive landmine.”
• AI startups’ choices between dollar capital and state-owned capital now go well beyond ordinary financing considerations, implicating model sources, domestic filings, and the China-US environment at the same time. Manus’s overseas product uses more than a dozen models including Google’s, while its domestic product previously worked with Tongyi to meet filing requirements; yet once the capital structure touches state-owned capital, overseas observers may still classify it as a state-backed project. 庄明浩’s conclusion was blunt: “Unless you decided on day one that you would stand firmly on one side, the middle ground has no meaning whatsoever.”
• BP leakage serves at least 3 purposes—fundraising, competitive warfare, and survival—but the Manus case points to a fourth: raising money purely by exploiting information asymmetry. Meituan- and Bilibili-style leaks involve reselling allocations; Uxin Used Cars-style leaks attack a rival by contrasting 2015 and 2016 metrics; Monster Charging-style leaks use 30% gross margins and 50%-80%, or even 90%, store revenue shares to call for a cooler market. If someone collected money using a Manus allocation supposedly earmarked for domestic entities but judged unlikely to close, then redirected investors to another asset after claiming the deal had failed, the conduct could approach fraud.
• For founders caught in the moment, silence or denial remains the most practical strategy, but any company aspiring to become a platform cannot avoid public opinion and interest coordination forever. Popular To C products depend on attention while lacking sufficient narrative power; 庄明浩 called the combination of politics, regulation, capital, and public communications an entrepreneur’s “hexagonal test.” 蒲凡 believes that anyone aiming to become Jack Ma or the Zuckerberg of a new era must learn to deal with public opinion, officials, and complex interest groups. “You can avoid it, ignore it, or stay silent for now—but you cannot stay silent forever.”
• The key test of a BP’s authenticity is not polished design, but whether it contains personalized facts that only the company could have generated. A real BP must explain what the company does, how far it has progressed, why it is raising now, how the money will be used, and what state it hopes to reach; extensive discussion of the industry, returns, exits, and transaction structure is more characteristic of intermediary sales material. Information remains the primary asset in the private market, while the more advanced skill of a seasoned investor is expectation management—朱啸虎 demonstrated it after the investment by saying “even at RMB250B, you still can’t get an allocation” and “nobody is selling.”

Deep dive

1. Two denials did not end the rumor; they turned a false consensus into information

• The episode starts with 2 Manus BPs: one supposedly aimed at dollar institutions, the other at RMB funds and state-owned capital, accompanied by claims that the financing was nearly finalized. When Manus was asked to verify them, it denied the financing report and said the 2 BPs might also be fake; the $100M Benchmark financing rumored in May was likewise not confirmed by Butterfly Effect.
• 张楠’s media-market view is that the private market is built on a large volume of information that cannot be fully verified. No comment, denial, and confirmation are all information in themselves. Even a false report deserves further questioning: “Why did a consensus diametrically opposed to reality form? And who believed it?”
• 蒲凡 cited Norway’s sovereign wealth fund as a contrast: funds required to disclose their entire portfolios are structurally hostile to the opacity of the private market. Private-market risk is difficult to quantify and can only be assessed by cross-checking information—both the industry’s operating model and a longstanding source of criticism.

2. Flashy peripheral narratives expose a BP that did not come from the team

• 张楠’s first reaction to the documents was that they were “too knockoff”: chaotic colors, little design sense, and awkward language. 庄明浩 added that even prompting an off-the-shelf AI to “create a financing PPT for Manus” should produce something more attractive and professional.
• A genuine company BP should present the team, development history, next steps, financing cadence, and strategic thinking. 庄明浩 believes these “personalized and endogenous” details are precisely what investors cannot obtain from public reports.
• Instead, the 2 documents mainly copy market size, industry trends, technical principles, and peripheral use cases, like a newcomer stuffing public reports into a template. They “describe a very peripheral story,” which is enough to conclude that they are not the team’s core financing materials.

3. Allocations in hot projects have become commodities, with BPs serving as fundraising manuals

• When 庄明浩 was building Panda from 2015 to 2019, particularly during the concentrated fundraising period in 2016 and 2017, he encountered similar documents almost every day: unfamiliar asset managers promoting “a leading game-livestreaming project” on social media, promising an “8% annualized guaranteed return,” and attaching redacted screenshots of an internal BP.
• The team could only follow the exclusive FA layer by layer, asking which investor had been contacted, whether its money came from an already-raised fund, state-owned capital, a large LP, or retail investors, and whether another round of fundraising would be needed. But when cash is scarce, “getting the money takes priority by far over all these messy details,” so startups sometimes have little choice but to look the other way.
• 蒲凡 said that tracing through the intermediary in this case revealed a genuine entity controlled by a provincial State-owned Assets Supervision and Administration Commission, with the relevant institutions also appearing on the records; the underlying capital, however, was largely from individual LPs. Many companies associated with those individuals had been deregistered or flagged for abnormal operations, and some of the related business owners were subject to restrictions on high consumption. “Real information” in a document does not automatically make the transaction real.
• 庄明浩 reconstructed one possible mechanism: a fund with local state-owned-capital attributes first secures an allocation in a hot project but may not have enough cash on hand, so it raises from private capital. The true investors, the vehicle holding the money, the fund exercising the allocation, and the portfolio company may be separated by “three to five layers.” The BP is the material needed to sell that allocation to the next layer down.

4. The private market is moving from active retailization to passive retailization

• In the market’s previous boom, retailization reflected the scarcity of quality allocations: even bank storefronts could display allocations in 360 and Meituan. Today, funds lack liquidity and are being forced to seek private capital. The fact that TME’s acquisition of Ximalaya prompted investors to say “thank you for injecting liquidity” shows that exits have become a systemic problem.
• SPV transactions involving OpenAI, SpaceX, and Anthropic have also been active in the US over the past 2 or 3 years. Buyers trade interests in the entity one layer above the investment vehicle without changing the portfolio company’s disclosed shareholder structure. Like S-fund transactions, this may provide liquidity to the market, but it pushes the problem down the road: whether indirect LPs will be protected when the real exit arrives remains a huge question mark.

5. AI financing choices are constrained simultaneously by models, filings, and geopolitics

• 庄明浩 believes this AI cycle is especially difficult to navigate: dollar financing raises concerns about hard-tech exposure and China-US relations, while RMB or state-owned capital may affect overseas business. Even a face-swapping app carrying an AI label is difficult to separate completely from the broader policy environment.
• Manus’s overseas product uses more than a dozen models including Google’s, while its domestic product previously worked with Tongyi. Bringing a domestic product officially online requires model and algorithm filings, so the model source and operating process may both need to be reconfigured; this is not simply a matter of choosing the best technology supplier.
• The outside world, however, will not recognize this middle ground. Once a domestic AI company has a local artificial-intelligence guidance fund or university-linked fund among its shareholders, overseas reports may classify it outright as state-backed. Agent and open-source-model developments have also moved extremely fast over the past 2 months. Whether Manus and Tongyi reflected hesitation over execution, process, or strategy remains unknown, as the episode explicitly noted.

6. Silence may be the only short-term option, but public narrative management is a long-term requirement

• Faced with wild BPs, 庄明浩’s practical answer was limited to “say nothing” or “choose denial”; any further explanation could create new problems on either side. The team can later tighten its checks on funding sources, ability to fund, and the need for additional fundraising, but choosing the RMB market still makes complete avoidance difficult.
• Bilibili’s BP leak years ago caused an uproar because the community was then viewed as an “ideal country,” making the BP and fundraising themselves seem almost heretical. Today, old shares, state-owned capital, cross-border structures, and regulation have been layered on top. Founders face a “hexagonal test” far more complex than the product.
• 蒲凡 said the public asks what value Pop Mart creates for society, while it accepts 韦神’s genius image through details such as his clothes, water bottle, and near-total lack of material desires. Technology startups, by contrast, struggle to persuade the public with an unfulfilled vision alone. “You have no narrative power whatsoever,” making silence a form of temporary defense.
• 刘强东 both emphasizes that JD.com has paid “over RMB100B” in social insurance and housing provident funds over the years and acknowledges that a company must be profitable while employees remain exhausted. That reflects a mature entrepreneur integrating social value and commercial reality into one narrative. 蒲凡 believes that if the goal is to become Jack Ma or the Zuckerberg of a new era, dealing with public opinion, officials, and complex interest groups is mandatory coursework.

7. BP leaks typically serve fundraising, competitive warfare, and survival

• The first category is fundraising: an investor capable of securing an allocation packages it as a new product for resale and prepares an explanatory document for the new buyer. 蒲凡 places Meituan’s Pre-IPO, early Bilibili, and Manus in this category, which is why the documents “look real, but feel a little fake when you look closely.”
• The second category is competitive warfare: the Uxin Used Cars whistleblower produced 2 BPs supposedly from 2015 and 2016, contrasting sharply different transaction volumes under the same definition to prompt the market to suspect data fabrication. Uxin subsequently reported the matter to the police.
• The third category is survival: a leaked Monster Charging BP said the business model worked with gross margins of about 30%, but store revenue shares had already reached 50%-80% and even 90% in popular areas. After the document surfaced, public opinion turned toward calls for a rational cooling of the shared power-bank industry; the leak itself served an agenda-setting function.

8. The Manus case may also be a pure fundraising trap exploiting information asymmetry

• 庄明浩 noted that Manus originally had a dollar structure and judged it “unlikely to take state-owned capital,” yet the wild documents listed domestic entities as the entire investment target. That structural contradiction is more concerning than the crude PPT.
• One possible path is for an intermediary to collect large sums from individuals using a Manus allocation, then claim the investment failed and redirect the money into another project. If a similar company’s founding team took a hard line, reporting the matter to the police on suspicion of fraud would not be impossible.
• More discouragingly, such an amateurish document did circulate for a long time among real investor communities, showing that it worked on at least some people. 蒲凡 believes that if it is ultimately proven to be a pure scam, it would mean the movement of talent toward state-owned capital and inland markets did not bring the expected rapid professionalization with it.

9. Historical losses are narrowing the channels, but the hottest projects remain vulnerable

• 庄明浩 estimates that retailization and channel-based fundraising in the private market peaked around 2018-2019. LPs who entered during the 2014-2015 “mass entrepreneurship and innovation” period have now seen the ultimate returns from blind pools, special-situation funds, and allocations in star projects. The lessons of historical losses are forcing the market to contract.
• Only the hottest sectors still have the “qualification” for such problems. Last year, documents for multiple rounds of secondary shares in Zhipu also circulated, showing a project “raising new shares while old shares trade on the side.” The documents were relatively formal and the route more clearly defined. With an estimated 50 funds already on the cap table, the company could hardly investigate every transaction and effectively chose to let the trading continue.
• 张楠 speculated that the party circulating the Manus BP may have spoken with the team when the product exploded in March. 庄明浩 confirmed that this was only an individual guess and called the judgment “especially hindsight-driven”; the episode did not verify it further.

10. A real BP answers company questions before industry questions

• 庄明浩’s framework is direct: a BP must explain what the company does, what state it has reached, why it is raising now, where the money will go, and what it hopes to become after the financing. The more central the content, the more it should consist of company-specific facts that cannot be templated.
• If large portions discuss industry size, potential returns, exit routes, and transaction structure, the producer is more likely to be an FA, asset manager, or allocation seller. Such material is not worthless; it simply serves a different audience from a company BP. As for an obviously shoddy version, most people can treat it as entertainment rather than a research document.

11. Information judgment and expectation management are the investor’s long-term assets

• 蒲凡 summed up the discussion in one line: “The core asset in the private market is information.” Risk can never be fully determined or quantified, so investors can only continuously exchange, verify, and interpret information. How false information forms and who believes it are themselves usable information.
• 张楠 used Xiaohongshu’s valuation as an example. The valuation of the GSR portfolio implied by Bloomberg-related reporting was about RMB180B, but they heard that 朱啸虎 was quoting RMB250B externally—and that even at that price, no allocation was available. 朱啸虎 said only “nobody is selling,” managing scarcity and price expectations at the same time.
• 庄明浩 emphasized that this is genuine post-investment capability: not merely recruiting, managing supply chains, or making introductions, but helping a company manage market expectations from the perspective of capital and transactions. The right to define the narrative is like the exploration and development capability of an oilfield; it determines whether potential value can be realized, while the final answer remains that “self-strength makes everything strong.”
• ByteDance’s old shares were once among the most liquid assets in China’s quasi-private market, and traders no longer needed to remake a BP. Xiaohongshu is approaching a similar state. Platform-scale companies have long operating lives, relative stability, and potentially existing revenue and profits, so their trading looks more like PE and can be priced more on the accounts than on the story.

12. The paradox worth tracking is that even a project with no shortage of buyers still needs a BP

• By the episode’s consensus, Manus sits in the AI Agent sector, which has both beta and strong consensus, and has exceptionally high name recognition. It should not need a BP to re-explain the project or lack buyers. Yet wild documents circulated widely, and “everyone wanted to see them.”
• 蒲凡 left 3 questions unresolved: Is the growth rate of AI products being overestimated? Has the project been packaged too vaguely? Has the capital market’s patience been exhausted to the point where every project must be explained from scratch? The episode did not treat these questions as answers, but as clues for future observation.
• The team’s earlier project exited successfully, and market feedback indicated that the founder was not short of cash, making 张楠 even more surprised by the Benchmark financing rumor. At the end of the episode, 张楠 again said Manus had indeed taken Benchmark’s money, but did not clarify whether this was the $100M financing reported in May; the parent company had not previously confirmed that rumor. What the wild BP ultimately revealed was not a secret about Manus, but how hot-project allocations, scarce liquidity, and an appetite for information can combine to create a business.