Qiming Venture Partners' Wang Shiyu: AI App Explosion Needs a Huge Business Model Overhaul
The real explosion of consumer-facing AI applications may require a massive shift in business models to make it happen.
Editor's Note: As AI sweeps through the global tech industry, with foundational models and compute infrastructure iterating at breakneck speed, consumer-facing applications and commercialization have lagged behind — scalable, sustainable C端 AI products remain elusive. At the 20th ChinaVenture Annual Investment Summit, Wang Shiyu, partner at Qiming Venture Partners, zeroed in on the AI consumer赛道. Drawing on the firm's decade-plus of full-chain AI investments, he clarified two innovation directions — incremental and disruptive — and dissected the real-world challenges of application deployment. He argued that a true explosion of AI applications will require a fundamental transformation in business models. This article presents an edited transcript of his keynote, sharing Qiming Venture Partners' analysis and investment thinking on the AI application layer, republished with permission from the firm's WeChat account.
Over the past two years, the most certain wave in global tech has been the AI explosion. From ChatGPT igniting widespread attention, to the intense "hundred-model war," to rapid iteration at the compute and model layers — technological dividends are being unleashed at an unprecedented pace.
Yet in this wave, compared with the swift advances at the model and infrastructure layers, AI's progress at the application layer, especially on the consumer side, has consistently seemed half a step behind. Despite continuous technical breakthroughs, truly scalable C端 AI products generating sustainable commercial value remain scarce. Is this due to immature technology, or is it business models that urgently need transformation?
Addressing this question at the 20th ChinaVenture Annual Investment Summit, Wang Shiyu, partner at Qiming Venture Partners, delivered a keynote titled Opportunities and Challenges for Consumers in the AI Wave, systematically outlining the firm's investment strategy and thinking on AI applications and the consumer端.
Wang Shiyu, Partner at Qiming Venture Partners
Wang first identified two parallel opportunities on the consumer端: incremental innovation and disruptive innovation, with the latter holding the potential to fundamentally reshape supply-demand dynamics. At the same time, AI consumer端 investment faces three major challenges — AI has proven itself in "save time" scenarios, but its performance in the much larger "kill time" market remains to be seen; marginal costs are no longer zero; and in the face of giant competition, how to escape the black hole effect.
Finally, he predicted that a true explosion of C端 AI applications may require a massive transformation in business models to accompany it.
The following is an edited transcript of his remarks, prepared by ChinaVenture:
It's an honor to be here at ChinaVenture's annual summit to share Qiming Venture Partners' investment strategy and thinking on AI, particularly the AI consumer端 and application端.
Let me briefly introduce Qiming's AI footprint. We began laying groundwork in this major赛道 very early — as far back as 2013, when it wasn't yet called generative AI, but rather NLP, machine learning, deep learning, reinforcement learning, and other technical directions. Over the past decade-plus, we've deployed capital across more than 100 AI projects, with cumulative investment of approximately RMB 12 billion or equivalent in USD, of which over twenty have successfully listed or grown into unicorns.
Our AI investment framework divides into three layers: infrastructure, models, and applications. In large models, we invested in Zhipu AI; in chips, Biren Technology; in the model layer, StepFun... Some of these companies have already listed, others are at the Pre-IPO stage.
Market consensus is relatively clear on the infrastructure and model layers, where we've been actively deploying. What hasn't yet formed broad consensus is precisely the application layer. Today I want to focus on sharing our early thinking and positioning on applications.
In C端 or application端 AI investing, I'll share from two angles: first, opportunities for the consumer端 in the AI wave; second, challenges faced in application端 investing amid this wave.
01/ Two Types of Innovation Opportunities for Consumers in the AI Wave
First, on opportunities. We believe two types of innovation are happening in parallel right now.
The first is incremental innovation. Its drivers include new intelligence (generative AI) and new data — through new intelligence infrastructure (new models or new agents), we activate previously underutilized existing data, creating entirely new experiences. For example, we already have positions in AI+social, AI+gaming, AI+office meetings or transcription, and fitness-related industries.
Take AI+social as an example. We invested in a dating and marriage platform called "Qianshou," founded by Wang Yu, formerly the founder of Tantan. What role does AI play here? Traditionally, matchmakers provided one-on-one services, but at extremely high cost. AI enables intelligent data-driven precise matching and deep service for platform users, making what was once expensive nearly free or extremely low-cost. This is a textbook case of incremental innovation.
The other type, which we're actively evaluating and which is currently unfolding but hasn't yet exploded or formed market consensus, is disruptive innovation. I believe the underlying driver of disruptive innovation is more about fundamentally transforming supply-demand relationships.
For example, we're looking at AI+education and AI+healthcare. In AI education, we invested in Yuaiweiwu, founded by a former co-founder of Gaotu Techedu. In education, applying new intelligence can deliver what were previously costly one-on-one or small-class educational experiences at very low or even zero cost to all users, fundamentally liberating teacher productivity. In the previous mobile internet or internet wave, the biggest cost for education platforms was teachers. With AI and new models, including "teacher + AI," the supply side of teachers undergoes a fundamental transformation.
On supply-demand transformation, there's a good case study from the mobile internet era: DiDi. DiDi started as merely an information matching platform, connecting taxis with riders — but if that were all, the market size would have been extremely limited. Beijing had roughly 60,000–70,000 taxis, so nationally perhaps a bit over 1 million, making the total addressable market roughly $200 million in commission revenue. That would have been the ceiling. But as we all know, DiDi far exceeded this scale because it later added express cars, premium cars, and other vehicle types. DiDi transformed the supply side, not just the information matching.
Therefore, truly disruptive innovation through technology or new intelligence often creates massive opportunity by fundamentally transforming supply-demand relationships — and we believe AI holds this potential in education and healthcare.
02/ Three Major Challenges for AI on the Consumer端
In application端 and consumer端 AI investing, we also encounter challenges.
On April 22, the Nasdaq hit another record high, driven primarily by optical communication modules, storage, and continuously rising GPU chips — these are highly certain, consensus opportunities in the AI mega-wave. But there's also an awkward reality: AI on the application端 always seems to fall slightly short, or everyone is still searching.
Currently, applications divide broadly into two categories: "Save time," mostly helping users improve efficiency, such as tools for individuals or small-to-medium enterprises. The other is "Kill time," entertainment, with Douyin as the quintessential example.
If we follow the experience of the previous mobile internet era, from the C端 perspective, the entertainment market far exceeds the efficiency market — this should be intuitive. But we're also seeing that in the United States, many major tech companies are conducting large-scale layoffs. Many institutions, including VCs and PEs, have raised a point: AI's biggest opportunity right now isn't about how much value it creates on the C端, but that it has already reached the labor market — a trillion-dollar-level market in the United States.
The second challenge is that marginal costs are no longer zero. In the mobile internet era, early investment advice was typically to acquire customers as quickly as possible — rapid user acquisition was a critical prerequisite for success. So investors then heavily weighted DAU and MAU; monetization could come later, because at sufficient scale, monetization would naturally follow. This was the iron law of the internet from that era.
But in the AI era, user growth means increased token consumption — marginal cost is no longer zero. Early AI-native products must not only pay "toll fees" to internet giants for customer acquisition, but also pay token costs for user usage. And most of these costs cannot be covered by user payments. For example, with an AI version of Douyin, where users swipe through short videos, it's difficult to charge them per video. This is a fundamental difference between the AI era and the mobile internet era.
The third challenge: in the AI era, a frequently heard term is "democratization." AI democratization creates opportunities not just for individuals, but for enterprises as well. When facing giant competition, the user data they hold serves as critical production material or competitive moat. This means that when startups build new AI application端 products, a crucial point is to differentiate from giants. Currently, more vertical, independent, private, and exclusive data can somewhat create commercial viability and defensibility.
A popular saying in recent years: "AI is eating software." In the first half of this year, a major trading theme in U.S. equities was AI replacing traditional software, causing software stocks to perform very poorly. A recent prominent example: when Anthropic launched Claude Design, Figma — a top-tier software in the traditional designer space — dropped 10% that same evening. Such cases have been numerous on the tools side over the past six months. A single large model feature iteration can likely overshadow all the engineering efforts you made over six months or even the past year.
My personal prediction: the explosion of C端 AI applications may need to be accompanied by a massive transformation in commercial models. Take the recently hot AI short dramas as an example. AI short dramas have dramatically slashed production costs compared to live-action. A recent benchmark I heard: a roughly ten-episode live-action short drama previously cost about RMB 200,000 to produce; using AI, a comparable-quality short drama might require only three people, about one to two weeks, at a cost of RMB 8,000 — this is unquestionably a crushing advantage over traditional production costs.
But whether AI short dramas can support an independent new platform remains highly contentious. Whether producing short dramas or short videos, the current business model is selling the duration of the drama or video to short video platforms for revenue. AI short dramas, AI comic dramas, AI short videos — after users create videos, the optimal monetization path remains selling to Douyin, Hongguo, and Kuaishou. This is the most practical and profit-maximizing choice; this hasn't changed. Under this premise, can so-called AI-native content support a platform? I believe this requires a transformation in business models to sustain.
Edited by | Wang Manhua
Source | ChinaVenture
PAST REVIEWS
Qiming Honors | Qiming's Wang Shiyu Named to Cyzone 2024 Under-40 Investors List
Founded in 2006, Qiming Venture Partners currently manages 11 USD funds and 7 RMB funds, with total assets under management reaching $9.5 billion. Since inception, the firm has focused on investing in early and growth-stage outstanding enterprises in Technology and Healthcare.
To date, Qiming Venture Partners has invested in over 580 high-growth innovative companies, of which more than 210 have listed on the NYSE, NASDAQ, Hong Kong Exchanges and Clearing Limited, Shanghai Stock Exchange, and Shenzhen Stock Exchange, or exited through M&A and other means. Over 80 portfolio companies have become recognized unicorns or super-unicorns.
Many Qiming Venture Partners portfolio companies have grown into the most influential companies in their respective fields, including Xiaomi (01810.HK), Meituan (03690.HK), Bilibili (NASDAQ:BILI, 09626.HK), Zhihu (NYSE:ZH, 02390.HK), Roborock (688169.SH), Hesai Technology (NASDAQ:HSAI, 02525.HK), UBTECH (09880.HK), WeRide (NASDAQ:WRD, 0800.HK), HyperStrong (688411.SH), Insta360 (688775.SH), Unisound (09678.HK), Biren Technology (06082.HK), Zhipu AI (02513.HK), Gan & Lee Pharmaceuticals (603087.SH), Tigermed (300347.SZ, 03347.HK), Zai Lab (NASDAQ:ZLAB, 09688.HK), CanSino Biologics (688185.SH, 06185.HK), Schrödinger (NASDAQ:SDGR), MicroPort EP MedTech (688617.SH), Sanyou Medical (688085.SH), Amoy Diagnostics (300685.SZ), GenScript ProBio (688520.SH), Insilico Medicine (03696.HK), Hope Medicine, Yuanxin Technology, MediLink Therapeutics, LaNova Medicines, StepFun, and others.