After more than a decade in venture capital, Zhiyang (John) Gao, Managing Partner of Trending Capital, has spent most of his career wrestling with a single question: how hard technology — Artificial Intelligence above all — makes the leap from a capability in the lab to a product that actually gets widely used, and creates value, in the real world. From CICC Capital to founding Trending Capital, his lens on deals has kept converging in one direction: the technology matters, but what determines whether an AI company goes the distance is the kind of scenario it embeds that technology into.
In a recent interview, Gao discussed how his investment logic has evolved, where he sees the opportunity in the application layer, and his read on where the industry is heading.
Three companies, one shared logic
Looking back at the AI investments Gao has led — Neolix, QCraft and UniSound — outside observers tend to file them under different categories: autonomous delivery, autonomous driving, and intelligent voice. To Gao, their core is the same: none is “selling an AI model.” Each embeds AI capability into a real-world scenario with clear, must-have demand and a closed commercial loop.
UniSound, he notes, has taken speech and language-understanding capability into specific industries such as healthcare and IoT, and is now listed on the Hong Kong Stock Exchange. Neolix has narrowed autonomous driving down to the certainty of “last-three-kilometre” delivery, proving out a business model with a scaled fleet. QCraft, on one of the hardest paths in autonomous driving, has taken a pragmatic approach, steadily expanding the boundary of what it can actually deliver.
“When I was willing to back these companies relatively early, what I valued wasn’t how advanced their technical specs were — it was that the teams had all thought through the same question: who exactly is my AI for, at which step in the workflow, and how much cost does it save or how much value does it create?” Gao says. The more specifically a company can answer that, he argues, the more resilient it becomes.
Why he is now focused on the application layer
Over the past two years, the industry’s attention has concentrated heavily on foundation models. Gao does not dispute that breakthroughs in base models are the bedrock of this wave of AI — but from an investor’s standpoint, he is more interested in what gets built on top of that bedrock. In his view, model capability is fast becoming a kind of “accessible infrastructure”; what is genuinely scarce, and what genuinely forms a moat, is the ability to fuse that capability with a specific industry’s know-how, data and workflows into a product users cannot do without.
That, he says, is why his focus now sits at the application layer. The opportunity there, in his analysis, is that the marginal cost of the technology keeps falling while the depth of scenarios is almost limitless. Whoever can take AI in a sufficiently vertical domain from “usable” to “genuinely good” to “irreplaceable” will be the one to ride out the noise. He is more inclined to back teams that sit closer to the end customer and can convert the technology dividend into real revenue.
Four calls on where AI is heading
First, AI’s value is shifting from “generating” to “completing.” Early on, Gao notes, people marvelled that AI could write, draw and converse; what is more worth anticipating next is whether it can autonomously carry a complex task from start to finish. This leap from “answering questions” to “getting work done” will redefine the efficiency benchmark across many industries.
Second, the key to deployment lies not in the technology but at the boundary of trust and responsibility. In fields he has long followed, such as autonomous driving and healthcare, technical capability stopped being the only threshold some time ago; building a sustainable framework around safety, regulation and the allocation of liability is often the real variable that sets the pace of commercialisation. Understanding this, Gao believes, is fundamental to this kind of investing.
Third, China’s AI ecosystem has its own distinct advantages. Rich application scenarios, a complete industrial chain, and a market willing to iterate quickly allow many AI capabilities to find a route to deployment at remarkable speed. This capacity for “engineering-led deployment” is a piece of local competitiveness he increasingly values in cross-border investing.
Fourth, respect the cycle. Every wave of technology, Gao cautions, moves from overheating back toward rationality. An investor’s job, as he sees it, is not to chase the hottest narrative but to identify, amid the noise, the companies that can genuinely traverse the cycle and turn technology into sustained cash flow.
In closing
After years in AI investing, Gao has come to endorse a plain idea: great technology ultimately proves itself by being used. He hopes Trending Capital can keep standing alongside the founders willing to bring AI into industry and into daily life.
Zhiyang (John) Gao is Managing Partner of Trending Capital and of its onshore RMB fund-management arm, Ningbo Yiyi Fund Management Co., Ltd. Website: www.trending-capital.com.
