At the Intersection of the Technological Singularity and an Industry Inflection Point, Place a Firm Bet | Unity Ventures View
The Four Most Important Trends in Early-Stage Tech Investment
"Singularity" and "inflection point" — these two words define our era. The former brims with boundless imagination about the future, representing a discontinuous, explosive leap. The latter is more real and urgent, signaling the obsolescence of old paradigms and the establishment of new rules.
When these two concepts intertwine, how do entrepreneurs find the optimal fit between technology and application? How can capital place the defining bet on the future?
Xiao Wang, founder of Unity Ventures, shared his views at the "Jazzy Gravity X China Tech Industry Investment Conference," the "NetEase Future Conference," and the "Unity Ventures AI Hardware Salon." He believes the next breakthrough in AI will come from significant improvements in end-to-end service capabilities within closed-loop scenarios, as well as opportunities in AI hardware, intelligent manufacturing, and globalization.
Wang believes that entrepreneurship is always something you can start right now. Regardless of the environment, there is always a right time to start a company. The same holds for investing — maintain confidence and patience at all times. "We have been on the path of early-stage tech investing, steadfastly supporting entrepreneurs during the riskiest and most challenging early phases."
The emergence of DeepSeek has brought greater clarity to the stalemate in US-China tech competition, strengthening market confidence in tech companies. This year, tech investment in China has grown rapidly. Here are several trends worth watching.
Intelligent Applications: From Conversation to Execution, Agents Unlock Broader Horizons
The key to the second half of AI lies in the deep integration of technology deployment and commercial scenarios, and China is well-positioned to lead innovation in this wave.
AI has evolved from early large language models chasing parameter scale to a diverse technical system integrating language, speech, video, and world models. Its core capability has shifted from chat and Q&A to executing specific tasks. The next AI breakthrough will come from significant improvements in end-to-end service capabilities within certain closed-loop scenarios.
AI coding does exactly this — completing code services end-to-end. It has two characteristics: first, as AI develops, it will certainly expand from coding into other domains, because coding has determinism and is relatively easier; second, the next breakthrough will happen in relatively closed-loop scenarios with deterministic inputs and outputs. We've invested in a video editing project, which has deterministic elements.
Following this logic, scenarios with deterministic inputs and outputs that can achieve a complete service闭环 are the most likely next breakthrough points.
Coding is already very clear, with revenue growing rapidly. Breakthroughs across various domains will likely follow. Unlike previous understanding, this won't be limited to a single industry — because large models possess generalist capabilities, breakthroughs may happen simultaneously across multiple scenarios. We are on the eve of an application explosion.
AI Hardware: From the Digital World to the Physical World
Large models have reached a new critical juncture. Through data collected by sensing devices, they are combining knowledge and know-how from the digital world with hardware and everyday scenarios, gradually entering the physical world.
Only when agents can truly interact with, plan within, and autonomously complete tasks in the physical world can we approach true AGI. This stage still requires time, depending on improvements in multi-agent capabilities and the maturation of hardware and humanoid robots.
The specific path requires entering from niche scenarios, combining hardware with AI to better perceive the world and establish data闭环, thereby unleashing "intelligent value" over the long term. Unlike software, high-standard delivery capability is crucial for AI hardware. Only by converting every effective interaction into replicable experience can unique data advantages and delivery capabilities be formed.
What must be firmly believed is that with China's unparalleled supply chain advantages, and once large model infrastructure construction is complete, China has the soil to surpass other countries and give birth to AI hardware giants.
A New Paradigm of Globalization: Born Global
Large models represent a structural opportunity, but other domains still hold promise — intelligent manufacturing, the globalization of Chinese enterprises, and more. The companies we invest in do business not just in China but globally. The globalization capabilities of Chinese startups and the momentum of Chinese companies' global布局 have both shown clear growth.
The traditional path was to first establish oneself in the China market, then replicate capabilities and teams globally. Now, day one is global — whether through the multilingual capabilities inherent to large models, or through global resources and product capabilities built into the team from the start.
This globalization capability stems mainly from the fact that Chinese startups stand at a high position across the value chain — from industrial and supply chains to technical capabilities and engineer dividends.
China's innovation and technical capabilities have reached a relatively high level, allowing us to tackle global markets from the outset. The space between serving several billion people globally versus 1.4 billion in China alone is fundamentally different.
Just as the United States has trillion-dollar companies, will China see trillion-RMB companies become trillion-dollar ones? This is a distinctly real opportunity. The opportunity for Chinese entrepreneurs to build globally is significantly greater than before.
Disruptive Innovation: Controllable Nuclear Fusion, Quantum Computing, and More Are on the Way
A society's progress requires both foundational accumulation of relatively mature technologies and zero-to-one disruptive innovation. Only when 1% disruptive innovation points the way, combined with 99% engineering realization and systematic construction, can society achieve true leapfrog development.
Disruptive innovation means there is no existing path to follow. It requires major technological changes where technical depth matters, alongside commercial重构. Whether in nuclear fusion, quantum computing, brain-computer interfaces, robotics, or large models, this characteristic exists.
For investment institutions, this means paying attention to disruptive innovation with foresight — which also places higher demands on our ability to judge technological trends. We are not betting from behind, but in the first round, the second round, the earlier the better. We are willing to place bets when entrepreneurial teams face the greatest risk and the greatest difficulty.
For an early-stage institution like ours, as long as something is imaginative, the technology has potential, and the team has the foundation and capability to support that vision, we will support them early and help them advance to the next round.
(Based on remarks by Xiao Wang, founder of Unity Ventures)
