AI Unicorn Factory

@Wu Ruirui

The Script

How do you manufacture an AI club deal?

First, you need a senior executive from a major tech company. If from ByteDance, ideally a post-90s 4-1 or post-98 3-1 level; if from SenseTime, given the formidable SenseTime Mafia, the predecessor in that role should preferably be a star founder too—then you can be crowned "the next so-and-so."

The entrance must be expensive. Best to start at $80 million valuation—this isn't a normal angel round figure, but abnormality is precisely what attracts.

Market whispers must circulate. Sequoia, Hillhouse Capital, IDG, Monolith are in; they're already queuing for the third round; valuation about to double; "Mama Wu" loves it...

Here's a critical point: be famous, yet maintain absolute mystery. The founder shouldn't show their face in the market; the product stays under wraps until enough money is raised. As for company goals and how they'll get there, most people can only rely on hearsay or speculation—like looking at a card not yet turned over.

Sometimes, the founder, investors, media, and market engage in a game of active and passive dynamics. Is this what people call "playing hard to get"? In any case, amid an atmosphere of FOMO and mystery, if you can complete the first one or two rounds at maximum speed and maximum valuation, then even without visible business progress, subsequent rounds typically close quickly too—and at multiples of the valuation.

Congratulations, the script is basically set. Funds never considered top-tier, those who only resolved to invest in AI from 2025 onward, major companies whose core businesses might be disrupted by AI... all will willingly pay top dollar for this script.

Club deals originally referred mainly to PE-stage investments, usually because the capital required was too large for any single fund to swallow, necessitating syndication to piece together allocation. But interestingly, club deals are now appearing in VC too, and in an almost standardized, industrial assembly-line fashion—the contrast is striking.

Yet this is precisely what's happening in AI right now.

Many feel simultaneously excited and fearful; eager to participate yet dismissive.

The Loop

There is a loop that exists.

"Loop" literally means a circle. The term, as elsewhere learned it, comes from several VC investors, referring specifically to certain small circles.

One VC fund investor said that early last year, he learned early on that a high-potential founder was leaving to start a company, but couldn't get in touch at all—friend requests went unanswered. Even when he found the founder's "big brother" to make an introduction, the other party remained silent. He was puzzled—given his institution's brand power, such situations were rare. It wasn't until a colleague found an FA that the founder finally connected with the colleague through the FA's arrangement.

He gradually understood: the other party was following the FA's instructions. "But I'm not in that FA's loop."

Though from an FA's core duty—helping companies raise capital efficiently and strategically—this may not be a problem. Yet an interesting phenomenon is that over the past year or two, multiple AI star projects have been essentially monopolized by a handful of FAs. And flipping through their investor lists, one finds striking similarity.

The so-called loop is really just a circle of friends.

One FA founder told us that many senior executives from major companies leave without a clear direction, let alone a product demo. And when VC decisions must rely more on judging people, yet valuations are this expensive, investment has already shifted from rational decision-making to psychological warfare. "Without someone championing it inside the fund, the valuation could never rise this high."

This FA said that at such moments they need to collaborate with key investors: those with sufficient sway and a solid track record, deeply trusted by their bosses; who also understand managing partners' pain points—what kind of founders they prefer, which peer funds' competition they care about, whether they're valuation-sensitive or want to swing for the fences—to tell the story well internally.

More importantly, they must hold firm at investment committee meetings. For instance, as we understand it, several AI companies were questioned early on in fund ICs for being "not solid enough," "unclear thinking," or having "mismatched maturity and valuation," but the sponsoring investors were extremely persistent—some had projects rejected, then the deal owner pushed them back to the IC again.

More than one founder and investor has told us that when an important young investor is unusually insistent on a project, partners sometimes tend to support it. This has strong present-tense logic: partly trust, partly an implicit insecurity about what's "new"—"they think young people understand AI better, that maybe they've seen something the partner hasn't."

So to some extent, FAs and star fund investors, surrounding a few early-stage founders with good optics, form what's called the loop. elsewhere has spoken with many investors; one feeling is: those inside it smile without speaking; those outside feel some indignation, yet also some envy.

Over the years, the divergence between dollar and RMB funds has grown increasingly clear. AI applications prefer dollars; embodied intelligence needs massive RMB; counterparties for similar projects have become very fixed small circles. Within small circles, information and sentiment transmit rapidly. Thus certain names always appear alternately, repeatedly, on the funding lists of those star companies.

Perhaps no one loses, and for a time this may simply be the correct, with-the-times rule of the game. As long as the game continues.

Rainmakers

In investing, there's a concept called Rainmaker. Usually refers to those few largest-scale, most influential funds whose decisions and moves carry symbolic weight for market direction. Like Sequoia for early-stage; Tiger Global for mid-to-late stage.

But today, FAs seem to have shifted from behind-the-scenes matchmakers to something like Rainmakers themselves.

In a rapidly changing early-stage industry, many unusual behaviors can be explained. But we remain curious: why now?

The core is who holds the best assets. In today's market, if you're a high-potential founder hoping to raise substantial money, you'll most likely choose an FA. Founders come from similar companies, or studied under the same famous advisor in academia, and there's a covert comparison in fundraising: "I'm a 4-1, how could my valuation be lower than a 3-2?"

Of course there are two types of companies that generally don't use FAs: either large-scale teams with professional investors already in place, like Kimi; or serial successful entrepreneurs like Guo Lie.

The FA industry itself is quite characteristic of Chinese venture capital. But such high-density involvement in AI companies' development processes—it's hard to say whether this represents market maturity or immaturity.

Many decent-quality founders maintain good working relationships with FAs. In industry lore, one star company and one FA that rose in the AI era have a mutually celebrated success story.

This company once teetered on the brink, with barely enough cash left for salaries. FAs that swarmed when it was hot scattered one by one; only one persisted in "pushing every possible institution," ultimately finding the company lifesaving funding, after which it quickly entered a rapid growth channel.

After this, the founder introduced multiple former colleagues interested in starting companies to this FA. Several star projects later, the latter quickly built its reputation, with many of its client companies raising funding at breakneck pace.

Bao Fan of China Renaissance once said, each generation has its own affairs. In the AI era, not just tech and business companies are being reshuffled, but funds too, and FAs especially.

Today the industry has already seen talk of the "AI FA Three Giants"—they entered the industry in the mobile era, witnessed cycle after cycle, but were then too junior. Even in 2023, large-scale model projects that could raise big money didn't fall to small FAs. It wasn't until AI applications took off.

So the loop, one might also say, is each generation having its own comrades-in-arms.

The Catfish

Over the past few years, AI's momentum becoming a prairie fire and the emergence of a new generation of VCs have happened almost simultaneously.

One generation of entrepreneurs makes one generation of investors; the new generation of GPs also believes that this AI era's Zhang Yiming should be discovered by their AI-Native fund. These leaders are generally post-85s or even post-90s.

Their venture funds have younger founders, faster decision-making, yet also possess the mature experience of large funds—for established funds, it's as if catfish have entered the industry.

One investor told us that for most funds, the month around New Year is dead time—vacation starts a week before New Year's Eve, and after reopening there's another week or two for annual planning, during which projects are set aside. But this year, one fund's partners started meeting founders on the third day of Lunar New Year, still sending term sheets during holiday. He was anxious, having to contact founders of projects not yet pitched, asking if any investors had launched surprise attacks during the break.

Speed also comes from organizational flatness. In large funds, after rounds of meetings, giving a TS within a week of seeing a project is already considered fast; in some new funds, half a day suffices. One investor said they downplay hierarchy internally and externally, uniformly calling everyone "investor"; corporate culture emphasizes Mafia-style mutual backup.

Under pressure, some established fund leaders have returned to the battlefield. Among several top-tier dollar funds, 70s-generation GPs who had slowed their pace in the post-pandemic era began intensively looking at projects again. One fund limited each project to one hour of presentation because every hour from morning to night was already booked with meetings; another fund reportedly held 16 IC meetings in one week.

Peer pressure objectively exists. Within funds, competition and information barriers exist between groups and individuals; between funds, established leaders and new funds with 20-year age gaps, old rivals who've competed neck-and-neck at the same stage for years—all are roiling under the surface. At the intersection of these mutual measuring gazes, investment decisions for star projects get made.

So this has also created a peculiar phenomenon: many wonder why numerous AI and embodied intelligence companies show virtually no business progress, yet funding keeps climbing?

Partly it stems from jostling between funds.

No wonder Freda Duan, partner at Altimeter Capital, previously said that VC is consensus investing.

Role Models

Synchronously with all this: more investors are joining companies they've invested in.

For instance, the most discussed recently is a top-tier fund investor who recently joined an AI star company in a key role.

This is entirely not an isolated case. According to elsewhere's knowledge, there have been at least four or five similar cases recently. All had invested in the company from their original funds, then chose to join, deeply participating in the company's fundraising and go-to-market; though a minority are responsible for technical R&D.

Many young investors view them as role models.

elsewhere previously wrote about a widespread round of promotions in China's VC industry not long ago. To some extent, this was also fund management's means of countering this trend.

Investor is, of course, a promising career. But it's quite grueling. Over the past two decades, VC's power structure has been relatively rigid, with little mobility among core personnel. This is also what makes the industry unfriendly to young people.

Choosing to join a company means potentially faster growth rate and financial returns. This is perfectly understandable. But for an investment decision, if it contains the investor's own career expectations, is this still a pure investment decision?

The Result

Today's venture financing exuberance surpasses even 2021's peak. But AI isn't just the venture industry's proposition—it's a broader societal one, and thus it—including its capital phenomena—seems exempt from all questioning.

Yet Matrix Partners' Huadong Wang recently posted something to this effect on social media: does anyone still remember the SaaS and innovative drug FOMO of 2020-21?

AI's prospects probably cannot be analogized to SaaS and innovative drugs at all, but compared to Chinese VC's craziest era—the mobile age—does this generation of AI entrepreneurs need such massive capital?

Mobile-era capital arms races stemmed largely from competition-driven mutual attrition; but at the current stage, except for similar situations in major companies' and model companies' AI wars, most startups don't face such scenarios—so what's the value of money far exceeding actual need?

All gifts are priced.

One investor expressed to us that regarding their star portfolio company, market speculation about its business is everywhere—even interviewed candidates get approached for intelligence. Normal product launches always involve trial-and-error iteration, but under this pressure, founders don't dare mass-market, continuously pushing new products through internal testing only.

And looking back, among companies that raised big money in the past two years—elsewhere will do a roundup later—how many have launched products that users buy into? One founder even used investment funds to buy wealth management products: before finding direction, ensuring investors don't lose principal.

Borrowing a lyric: So sad, this isn't the result I wanted.

Cover image: Adolph Menzel, $2, 1875, The Alte Nationalgalerie