Translator’s Note: About the Author
José Maria Macedo, Founding Partner of Delphi Ventures, Founder and CEO of Delphi Labs. Delphi Digital, founded in 2018, is one of the most influential independent research institutions in the cryptocurrency and digital asset space, with Delphi Ventures managing over 200 portfolio projects. Macedo himself has been deeply involved in the cryptocurrency industry for many years. He studied Politics, Philosophy, and Economics (PPE) at the International School of Brussels, founded several profitable businesses during his studies, and received honors such as the Duke of York Young Entrepreneur Award. In recent years, Delphi Ventures’ investment focus has been expanding from Crypto to AI and deep tech. This article is Macedo’s summary of observations after conducting on-the-ground research in China’s AI ecosystem.

Recently, I spent two weeks in China, meeting with founders, venture capitalists (VCs), and CEOs of listed companies within the AI ecosystem. Before going, I was very bullish on this ecosystem, eagerly anticipating discovering world-class AI talent here at valuations far below those in the West.

But upon leaving, my perspective became more complex and nuanced: I became more bullish on Chinese hardware than I expected, but more bearish on software. At the same time, some of my observations about Chinese founders surprised me greatly.

The Founder Question

The great founders I’ve invested in all share a highly recognizable “gene”: independent thinking, rebelliousness, passion, and extreme persistence. They are not obedient; they constantly ask “why” and refuse to blindly follow established wisdom. Their decisions often seem inexplicable to outsiders but make perfect sense to them. They possess an uncompromising drive, often evident in their past experiences of near-obsession and pursuit of perfection. Among the vast sea of highly intelligent people VCs encounter daily, their “spikiness” makes them instantly recognizable.

However, many of the founders I met in China belonged to a completely different archetype—which surprised me.

They are undoubtedly outstanding: graduates of top universities, with impressive resumes from ByteDance or DJI, papers published in Nature, and multiple patents. Achievements that only the top tier of technical talent possess in the West are almost like a “ticket to entry” here. Moreover, they work harder than almost anyone I’ve ever seen. Meetings could be scheduled at any time, on weekends, even across different cities. One founder even met with us on the day his wife was giving birth!

But, that capacity for independent thinking, that rebellious spirit, that zero-to-one vision, was hard to find. These founders’ backgrounds are highly homogeneous, their business plans more risk-averse, their ideas often just premium V2s of existing products, rather than truly original bets. Considering the massive volume of technical talent China produces, I expected to hear more unprecedented, wild ideas here.

V2s refers to Version 2, meaning iterative upgrades of existing products; zero-to-one refers to disruptive innovation from scratch.

My interpretation is that China’s education system produces excellent talent but doesn’t leave enough room for “deviance”. This results in founders who are top-tier executors solving known problems, rather than those who can identify and solve a problem “nobody knew existed.”

VCs Are Reinforcing the Pattern

What’s more interesting is that local investors are exacerbating this phenomenon.

The core investment thesis for many Chinese funds is betting on excellent former employees from ByteDance or DJI—they value “pedigree” over “spikiness,” “halo” over “conviction”. The backgrounds of the VCs themselves reflect this: most come from big tech, consulting, or investment banking, mirroring the European VC scene a decade ago.

Ironically, looking back, China’s top founders—those who truly built era-defining great companies—often never worked at a big tech company:

• • Jack Ma was an English teacher who failed the college entrance exam twice.
• • Ren Zhengfei founded Huawei at age 43 after leaving the military.
• • Liu Qiangdong started JD.com by selling goods from a street stall.
• • Wang Xing dropped out of his PhD and started his entrepreneurial journey from day one.
• • Liang Wengfeng never worked anywhere else but his own company, yet built DeepSeek.

They were all rule-breaking outliers, people without shiny “credentials”—precisely the type who would likely be filtered out by the current investment system.

Finding these founders is where true alpha lies, but in my view, almost no one is paying attention to this group right now.

Alpha is a financial term referring to returns above the market average, here used to mean unique investment insight and returns.

Shenzhen and the hardware ecosystem

The most mind-blowing thing I saw in China wasn’t a startup pitch.

It was Shenzhen’s “underground hardware workshops”—where engineers systematically purchase high-end Western products, tear them apart, and reverse-engineer every component with extreme rigor. I left genuinely wondering if most Western hardware founders have any idea what kind of monster they’re competing against. The network effects here aren’t theoretical; they are physical, highly concentrated, and built up over decades.

The entrepreneurs we met backed this up with data: over 70% of hardware raw materials are sourced from the Greater Bay Area, and close to 100% from within China. This supply chain advantage allows their product iteration cycles to be far faster than anything Western hardware companies can achieve.

Most founders I met are replicating the “DJI model”: building consumer hardware in a vertical niche—like electric wheelchairs, lawn-mowing robots, next-gen fitness equipment—scaling to tens or hundreds of millions in revenue, then leveraging the accumulated customer base or underlying technology to expand into adjacent categories. Some companies are already far larger than you’d imagine. The most impressive company I saw was Bambu, a 3D printing company most Westerners have probably never heard of, reportedly with annual profits of $500 million and still doubling year-over-year.

Bearish on Chinese Software

I left more skeptical about investment opportunities in the Chinese software space than when I arrived.

At the model layer, China’s open-source models are indeed impressive—but closed models still lag significantly behind the West’s best, and that gap is likely to widen. The CapEx gap behind this is massive. GPU access remains constrained. And top Western AI labs are cracking down harder on model distillation. Revenue data highlights the disparity starkly: reportedly, Anthropic did $6B in revenue in February alone; while China’s best models have ARR in the tens of millions.

<block