Alphaist: A Dollar-Denominated Fund Born in 2025, and Its Ten-Year Bet

By Lili Yu

This year, after leaving 5Y Capital, investor Zhe Chen made a major life decision: launching Alphaist Partners, an early-stage USD fund. In today's climate, this choice seems almost absurd. The venture capital market of the past two years is widely seen as the harshest, most unforgiving winter in memory.

Fundraising has ground to a halt. In the first eleven months of 2024, the number of new VC/PE funds established in China dropped 46% year-over-year. To make matters worse, geopolitical tensions and various restrictions have led roughly 95% of USD LPs to steer clear of China entirely.

During overseas fundraising, one USD LP told Chen that his new fund hit virtually every red flag on their list: first, it invests in China; second, it focuses on early stage; third, it's a first-time fund. Beyond that, Chen is the textbook definition of a "three-no" GP: he's never been a GP before; none of his portfolio companies have exited; and apart from one finance hire, he has no team to speak of — a bona fide solo VC.

But here's the thing: there's always that remaining 5% that changes everything.

In the end, Chen raised tens of millions of USD from a group of entrepreneurs, fellow investors, and overseas institutions. Alphaist Partners has already completed its first investment and is advancing several new deals.

Against the pervasive pessimism that hung over the industry last year, Chen remains unfazed. He believes venture capital will always endure, and that the current vintage will prove to be the best of the decade before and after it.

His confidence stems from his belief that a new innovation cycle has begun. Just as startups need new paradigms, the investment industry needs new people and new paradigms to break old habits. Besides, he's the "barefoot" type — meaning he can afford to be fearless.

Before launching his fund, Chen spent seven years at 5Y Capital focused on AI and robotics investments, backing companies like HAI ROBOTICS, AgileX Robotics, Lightelligence, and Horizon Robotics.

Earlier still, as an electrical engineering student and later engineer, he spent nearly eight years in core R&D — first at BlackBerry working on smartphone communications, then at Google on Google Glass, specializing in embedded systems and signal processing.

In July 2023, after reading Waves's interview with Wenfeng Liang, he and his 5Y colleague Kaiyan He were electrified. Having deeply studied Frank Wang and the DJI ecosystem of entrepreneurs, Chen felt the article "instantly triggered a familiar radar." So our conversation began with the recently debated question: "Why did Chinese VCs miss Liang?"

The following is Chen's account, edited for clarity:

Why Chinese VCs Missing Liang Was Almost Inevitable

When the market is full of investors chasing buzzwords, inflating bubbles, and herding around "star teams," it's perfectly normal for an entrepreneur who's genuinely obsessed with an ideal — even to the point of paranoia — to be underestimated or simply not understood.

Internet-era investors are wired to look at DAU, retention, and revenue. They're accustomed to growth, subsidies, and deal-making because internet businesses have two defining traits: marginal costs approaching zero, and network effects that create winner-take-all platforms once user scale is achieved.

Investors also carry the assumption that large models would replicate the data flywheel of search and recommendation engines from the internet era — that more users and more usage would make models smarter. So a cluster of large model startups burned through cash on promotion, chasing DAU and retention. Some investors even scoffed at open-source model strategies for lacking a business model.

But DeepSeek shattered these assumptions: in the face of genuine intellectual power, feedback from average users is worthless; when intelligence is far from saturated, users migrate rapidly to more advanced models; and open-sourcing a superior model built the fastest-growing, most valuable developer ecosystem globally.

DeepSeek and DJI represent something fundamentally different from internet-era entrepreneurship. They're no longer simple competitions over business models and user numbers, but embody a new paradigm of technology-driven entrepreneurship.

When technology-driven ventures achieve breakthroughs or step-function advances, traditional internet VCs' fixed notions about business models and moats become irrelevant. What truly matters is whether a company is focused and efficient, and whether it can be the first to cross the critical technology threshold. Money can't solve problems at this stage — in fact, it can backfire when investors pressure companies to chase superficial metrics.

But I have a particular radar for entrepreneurs like Liang. At our first meeting last year, I immediately drew parallels to Wang. Having spent seven or eight years as an engineer at BlackBerry and Google, and having known many DJI people, I recognized that same pure, obsessive engineering temperament.

Liang's thinking on talent and R&D — his focus on recruiting and cultivating fresh graduates, his pursuit of originality over imitation — mirrors DJI almost exactly.

I once invested in a DJI alum who was building something new. I happened to know a top-tier hardware product manager who'd done something similar, and offered to make an introduction so he could learn from past experience. The founder immediately refused: "Why would I listen to how someone else did it, instead of figuring it out myself from scratch?"

This is why I particularly admire extreme, pure, first-principles entrepreneurs for this paradigm. They habitually filter out 99% of noise and avoid copycat shortcuts, focusing their innovation on critical problems — and this aligns deeply with my own philosophy.

In fact, many of technology's greatest companies — Apple, Cisco, NVIDIA, Qualcomm abroad; Huawei and DJI in China — raised very little capital, or didn't need VC money at all.

Some peers say VCs missed DeepSeek because it didn't fundraise. Others say when you find a great company, you should beg to invest. I don't think either mindset is right. The reason you'd have to beg is because the company sees no value in you beyond the cash.

So I think more about how I can better serve underestimated and misunderstood entrepreneurs. Beyond instinctive empathy for idealists and my understanding of technology-driven entrepreneurship, I want to help founders solve thorny problems and provide support when they truly need it — like DJI in 2010, or High-Flyer before 2015.

Why Launch a New USD Fund Right Now?

My confidence comes from two shifts: the changing paradigm of technology entrepreneurship, and the current market environment.

Technology entrepreneurship is distinctly different from the internet era. Internet startups were characterized by low complexity and high uncertainty. The business model was simple, trial-and-error was efficient, and once validated, victory came from pouring in money for growth.

So two types of investors thrived in the internet era. One was the philosopher, the deep thinker who could reason top-down with exceptional clarity — understanding network effects, marketplace dynamics, why Uber and Didi were two-sided networks while ofo and Mobike were time-sharing rentals, why the former could monopolize and the latter couldn't.

The other was those skilled at winning deals or chasing hot projects. Because of winner-take-all dynamics, momentum investing was consistently effective in internet investing.

But in the technology innovation era, none of this logic holds. The analogies, induction, and deduction that generalist internet VCs excel at can't make sense of early-stage DJI, Unitree, or DeepSeek.

Technology companies have high complexity and high uncertainty. Their accumulated moats can't be abstracted into a single model, and companies may undergo multiple step-function leaps. They can't be analyzed top-down, nor can they win users and market share simply by burning cash.

Today's technology entrepreneurship demands sophisticated technical taste from investors. First, you need clear-eyed understanding of technological trends and their pace. My engineering career illustrates this: BlackBerry overestimated user attachment to physical keyboards and underestimated 3G and app store adoption; Google Glass overestimated optical and display technology progress and underestimated user stickiness to smartphones as the super-terminal.

Second, don't oversimplify, but don't mythologize technology either.

A common mistake I see when investors evaluate robotics projects is failing to understand the difference between a demo and a product — lacking appreciation for engineering complexity and technical reliability. At the same time, there's no single killer move or insurmountable technical barrier. The real moats are always the least technically glamorous things: network effects, scale, brand.

How do you cultivate good technical taste? Get deep into the trenches, understand technology's essence, listen to engineers.

How did I know DeepSeek was exceptional? When DeepSeek Coder launched in late 2023, my 5Y colleague Kaiyan He immediately heard from AI coding entrepreneurs how it surpassed the best CodeLlama and GPT-3.5 Turbo at the time.

The current market environment, meanwhile, is when the best entrepreneurs and VCs are truly rewarded.

When markets are hot — 2020, 2021 — platform funds could invest in 100 companies a year. How do you cut through that noise? When there's no beta, or even negative beta, coverage strategies are guaranteed to lose money. This is the best environment for alpha investing.

I believe 5Y's concentrated, high-conviction approach is the truth in VC. Everyone knows 5Y led Xiaomi, Kuaishou, and Horizon Robotics at the angel round, but few realize how much they continued to invest in subsequent rounds.

My approach will be the same. Over my career, I haven't made many investments, but in over 80% of them, I led with a board seat. For strong companies, I'll lead consecutive rounds. With AgileX Robotics, I led two consecutive rounds and invested across three rounds total — and when I finished the third round, their lawn mower robot hadn't even reached mass production, let alone generated revenue. A year later, it became the industry's leading company.

The overseas institutional LPs backing me similarly believe in alpha strategies. They know peers are talking ABC (Anything But China), they know many GPs will struggle to raise future funds, they know competition for quality assets will decrease — and these are precisely why they're willing to back a first-time fund investing in China today.

I've noted their commonalities: they've all made serious money, and most have invested in 5Y Capital and Capital Today — both extremely alpha-oriented funds that delivered tens of X returns. So they know this strategy can work.

I'm deeply grateful for 5Y's influence. Beyond teaching me concentrated conviction, Qin Liu told me when I started that VC's purpose is to "inspire entrepreneurial spirit and create long-term value." When I felt lost in investing, he told me VC's core is "finding your own Lei Jun."

Thus, Alphaist's values are truth-seeking, passion, and long-termism.

Truth-seeking means returning to first principles and reality, with fewer analogies, inductions, and deductions. Truth is hard in private markets — why not chase consensus deals? After a few rounds, valuations rise, while disproof cycles are long.

Then there's the principal-agent problem in VC: bosses aren't on the front lines, relying entirely on junior staff for information. When DeepSeek V2 broke out last year, a friend at a leading fund told me, "Make sure your boss never finds out Liang met with some investors in early 2023, or he'll get slaughtered."

As for long-termism, two recent moments stand out: once, when I mentioned to Kathy Xu that DJI was a great company but unfortunate for investors since it wouldn't IPO, she immediately countered, "Why should a great company exit?" Another time, discussing fund terms with an LP, I said my standard term was twelve years, and he replied, "Twelve years is too short."

Many don't understand why I'm choosing to start out on my own now.

My answer to LPs during fundraising: in the next ten years, China will produce ten companies like DJI, and I want to partner with four or five of them at the early stage.

As a Robotics Investor, Why Haven't I Invested in a Single Humanoid Robot Company?

The answer is simple: I believe humanoid or general-purpose robots are still in their infancy. The hardware, software, and application scenarios all need considerable time to develop, and the current market hype doesn't change my assessment of the difficulty and timeline.

The reason humanoid robots are needed is that they're designed for human living environments. And humans, as highly evolved creatures, possess finger, torso, and muscle flexibility, dynamics, and power density far beyond what today's batteries, motors, and gearboxes can achieve.

Either we invent entirely new transmission and energy methods, or we make precise tailoring of applications and scenarios with current technology — otherwise, humanoid robot hardware today struggles to meet human-scenario demands.

On the software side, humanoid robots require algorithmic complexity far exceeding autonomous driving. Autonomous driving started with the 2005 DARPA Grand Challenge and took 20 years to begin commercialization. Given similar requirements for reliability and long-tail scenarios, humanoid robot software will likely undergo a comparable journey.

Application scenarios present an even greater challenge, as various specialized robots will compete with humanoid forms. The smartphone-versus-feature-phone analogy investors love to use is inappropriate: early smartphones were primarily a software revolution with limited hardware structure and cost changes, requiring little sacrifice from users to capture value.

When Tesla announced its China entry in 2014, that year's global Model S sales were just over 300,000. Tesla had already solved the science and engineering problems, commercial scale was established, and a hundred Chinese EV companies quickly emerged — yet the market didn't fully explode until 2020-2021. Has humanoid robotics reached that point?

Recently an investor asked me, "Haven't humanoids and embodied AI all been invested out? What else is there in robotics?" It reminded me of early 2024, when a friend at another fund said, "We led the angel round of one of the Six Little Tigers large model companies, invested in that hot AI application, invested in that AI infra play — our AI allocation is complete, what else is there in AI?"

If everyone consensus-bullish on something from day one, either I'm too late to the insight, or it's not suitable for startups.

The best entrepreneurial opportunities start out disliked by most, with few competitors, and big companies not paying attention — that's how you capture maximum returns when you succeed. When I met Xingxing Wang fiddling with AlienGo in 2019, he'd already persisted for three years in a direction no one cared about. No investor would have imagined that Unitree, selling a few dozen robot dogs annually to geeks and research institutions, would one day become one of the world's leading humanoid robot companies.

Additionally, there's a VC industry adage that first-round valuations over $100 million rarely produce outsized outcomes — and in my observation, this holds for both China and the US.

Every investor knows the Gartner Hype Cycle. The question is: can you participate in that initial wave? If you're positioned as an early, small fund, getting into hot teams at angel and exiting at Series B for a 10x return in a few years — that's fine. But if your goal is to empower young entrepreneurs and build something truly great over the long term, the choice between participating in a greater-fool game versus focusing on what you're convinced of but others dismiss becomes clear.

That said, robotics is indeed a major direction for Alphaist. It fits my investment aesthetic: a wave lasting over ten years, unclear path, small initial market size, and difficulty for giants to commit serious resources. This is precisely where early-stage VC opportunity lies.

Finding Entrepreneurs Who Create Something from Nothing, Who Insist Against the Odds

I don't want to invest in resource-arbitrage or "do it again" entrepreneurs. I want to find those who create something from nothing, who insist against the odds.

I invested in an early robotics project where the founder, by any conventional measure, was an amateur — no engineering background, no product or mass production experience, couldn't even assemble a credible team. What moved me was the founder's courage to keep tearing himself apart, his all-in passion, and his vision ahead of its time.

Deconstruct the most exceptional companies and you'll find their success was never an inevitable outcome of prevailing conditions, but rather the result of extraordinarily capable entrepreneurs making the impossible possible.

Take NVIDIA's GPU work — early on, it wasn't particularly glamorous. Couldn't Intel have done it? CPUs were bigger than GPUs, and Intel looked down on it. But Jensen Huang was driven by a dream of general-purpose computing, not just graphics rendering — building the software stack, building CUDA, accumulating for over a decade before AI's explosion.

In 2015, Horizon Robotics was the earliest AI chip company in the market. Kai Yu, an algorithm scientist, had never made chips and had no automotive industry experience. What drove him was a vision far ahead of its time and sheer determination. When the company was still chaotic, he insisted on tackling the hardest problem: autonomous driving chips. Many people, investors included, asked: with such a strong algorithm team, why not do pure software like facial recognition? But Yu insisted that algorithms alone would build no moat.

Truly great entrepreneurs are those who do the right thing when the team wavers, investors doubt, and the market gives negative feedback — this requires tremendous inner strength.

DJI's core values include "striving for perfection with passionate determination" — immense passion, extreme pursuit, and doing everything possible to fulfill long-term mission. This is my portrait of the ideal founder.

Alphaist aims to be the earliest, most long-term, and most influential partner for such entrepreneurs.

Professor Zexiang Li was magnificent in this regard — when Wang was most doubted and his team was falling apart, Li steadfastly supported him, providing lifesaving investment and introducing students to join. Without that support at that moment, the company might not exist.

Elon Musk also had one trusted VC friend: Antonio Gracias at Valor. In 2007, when the Roadster faced severe cost overruns and the company neared bankruptcy, Gracias took on massive supply chain cost reduction work. In 2018, during Model 3's production hell, he camped out at the factory with Musk, sleeping just a few hours daily, directly helping solve manufacturing problems.

2025 will be excellent timing. The democratization of AI, led by DeepSeek, brings massive variables. I eagerly anticipate it spawning many new companies, much as iOS did back in its day.

Whether robotics, smart hardware, or autonomous driving, all are rapidly landing and scaling.

Looking back from 2035 at what this decade will have brought, I bet AI and robots will be everywhere, and a new generation of entrepreneurs will build more DJIs and DeepSeeks.

By then, we'll be grateful we lived through the best vintage.

Image source: Unsplash