Zhang Lu, Founding Partner of Fusion Fund

AsianFin -- The landscape of artificial intelligence investment is undergoing a quiet but profound transformation.

In Silicon Valley—the beating heart of global AI innovation—a two-track evolution is underway. On one path, tech giants like OpenAI, Google, xAI, and Meta are locked in a high-stakes race, channeling hundreds of billions of dollars into developing ever more powerful foundation models. Their focus: building general-purpose systems with unprecedented reasoning, language, and multimodal capabilities.

But on the other path, a different story is unfolding. Leaner, domain-specific models—designed for speed, efficiency, and practical deployment—are quickly gaining ground. From healthcare to finance, these smaller-scale models are winning favor in high-value sectors due to their lower operational costs and faster time-to-market. What once seemed like a side narrative is now commanding the industry’s full attention.

Perhaps it’s not that AI is evolving too fast—it’s that we’re standing too close to see the shape of what’s emerging.

That shape became clearer in May. At the Open Agentic Web Conference on May 20 (Beijing time), Microsoft declared that “AI agents are the new infrastructure of the internet,” while unveiling more than 50 new products in a single announcement.

A day later, Google used its I/O 2025 Developer Conference to emphasize that AI is transitioning “from passive tools to active agents,” reinforcing the central role of agent-based systems in its vision of the future. Meanwhile, Nvidia continues to reshape the underlying compute landscape with its Blackwell Ultra platform, redefining the architecture of AI workloads at scale.

Together, these developments point toward a major industry shift—from the “war of a hundred models” to the next wave of platform-level innovation, with AI agents at its core.

“This is more than a trend. It’s becoming industry consensus,” said Zhang Lu, a prominent Silicon Valley investor, Founding Partner of Fusion Fund, and Visiting Lecturer at Stanford University. “AI agents are emerging as the next-generation general-purpose platform—comparable to what PCs and the internet were in earlier eras. This is the beginning of true human-machine collaboration.”

In light of this transition, AsianFin sat down with Zhang Lu for an in-depth conversation on “The Latest Trends and Opportunities in AI Innovation and Venture Capital in Silicon Valley for the Second Half of 2025.”

What remains for startups in a market dominated by giants? Which vertical applications still offer greenfield opportunities? And how can investors assess the evolving structure of the AI value chain?

Zhang’s perspective offers a roadmap to what comes next. As foundational models hit scaling limits, the smart money may well shift toward the agent layer and industry-specific solutions—where software meets real-world need, and where the next generation of unicorns could quietly take shape.

The following is the full transcript of the conversation between AsianFin and Zhang, edited for brevity and clarity:

The More Vertical, the Greater the Opportunity

AsianFin: In the first half of this year, Silicon Valley has been highly active in the AI sector. As someone who has been closely following investments in Silicon Valley, could you summarize your investment observations for the first half of the year?

Zhang: The first half of this year has brought a steady stream of exciting news, including both the exploration and breakthroughs of startups in AI, as well as the continuous product launches from major tech companies. Overall, we are currently in a phase where AI-driven innovation is experiencing an explosive boom, and the industry’s momentum remains strong.

Compared to previous years, this year’s discussions have focused more on the practical implementation of agent-level applications and the innovative integration of traditional industries with artificial intelligence. This shift highlights the enormous potential of AI technology in the future market.

Take the technology and IT sector as an example: its overall market size accounts for about 9% of the US GDP. However, the impact of this wave of artificial intelligence goes far beyond the tech industry—it is rapidly expanding into traditional sectors with vast amounts of high-quality data, such as services, healthcare, and financial insurance. The integration potential between these industries and AI is immense, and it is expected that the market influenced by AI will expand from the original 9% to as much as 50%-60% of GDP. Such a massive potential market presents unprecedented opportunities for industry players.

As early as two or three years ago, I mentioned in conversations with investors and entrepreneurs that many opportunities actually lie in the hands of large companies, as they possess high-quality data. There is already a difference in market size between the consumer (C-end) and business (B-end) markets, and large enterprises like Salesforce and Microsoft have strong ecosystems and platform advantages. However, certain vertical segments—such as financial insurance, logistics, and healthcare—actually offer more entry points for startups, enabling them to pursue more focused and in-depth innovation between the model and application layers.

At the same time, many major tech companies are actively building AI ecosystems, aiming to make it easier for startups to achieve technological and product innovation on their platforms.

I am very optimistic about the future development of the open-source ecosystem. In fact, I mentioned a year or two ago that the rapid iteration of the open-source community has a tremendous driving effect on the entire innovation ecosystem. It not only lowers the cost of innovation but also accelerates the interaction and agility of smaller models.

Looking back at the first half of this year, investment in the AI sector has shown a clear trend: a large number of AI application companies are returning to business fundamentals. By business fundamentals, I mean focusing on revenue growth, industry collaboration, and how to achieve efficient and rapid co-evolution with customers and the industries they serve.

Companies that excel in this area have achieved rapid annual revenue growth of 20 to 40 times while keeping their workforce size essentially unchanged. This is thanks to their deep integration of AI automation capabilities into internal processes—a phenomenon that is far from isolated.

As early-stage investors, we can clearly sense that artificial intelligence is being rapidly embraced at the application layer and has entered an accelerated development phase. New technologies are truly demonstrating their ability to integrate into industries and are creating exponential momentum across various sectors.

What’s most exciting about AI today is no longer its valuation, but its ability to deliver real-world implementation and value creation.

AsianFin: Can we understand this as AI reconstructing industries?

Zhang: I prefer to describe the role of AI as empowerment.

I often use an analogy to help people understand AI’s impact on industries: it’s similar to how computers became widespread across different sectors. At first, being able to use a computer was a competitive skill, but later it became a basic ability everyone was expected to have. Computers also replaced some human labor in many industries and reshaped workflows. The current development of artificial intelligence is very much like the transformation brought by computers back then.

This analogy may not be entirely precise, but it helps people gain a deeper understanding of how to apply AI, how to view AI, and how to prepare for its full integration.

Take the tech sector as an example: major technology companies like Microsoft, Google, and Meta have been laying off employees in recent years. However, these layoffs aren’t due to poor business performance, but rather because companies are concentrating more resources on core R&D teams to develop more AI tools. This has enabled them to achieve threefold, fivefold, or even greater productivity, reducing their reliance on large workforces.

Marc Benioff, founder of Salesforce, has also publicly stated that they are not hiring new staff this year. The reason behind this is their launch of the Einstein GPT model for enterprise clients, which can boost output by two to three times or more without expanding headcount.

He once put forward a viewpoint that I deeply agree with: in the future, CEOs and leaders will be managing not just human labor, but also digital labor, with artificial intelligence as the driving force behind digital labor. This shift is fundamentally restructuring the infrastructure of many industries. AI may not replace all human labor, but it will inevitably reshape workflows.

For employees with strong leadership skills or high-quality output, AI can further amplify their abilities, providing powerful support to their work and enabling exponential growth. On the other hand, for positions with lower output and highly repetitive tasks, AI may become a more cost-effective alternative.

This is precisely the dual impact of artificial intelligence on both industry restructuring and company organization, a phenomenon particularly evident in the technology sector.

AsianFin: You just mentioned that in specialized sectors such as finance, insurance, logistics, and healthcare, there are more opportunities for entrepreneurs. Could you elaborate on that?

Zhang: In fields like healthcare, finance and insurance, as well as logistics and supply chain, artificial intelligence has already demonstrated broad application potential.

In the healthcare sector, AI currently plays a role mainly in assisting and empowering the industry. This field naturally possesses vast amounts of high-quality data, which creates favorable conditions for model training and industry adoption. Compared to the sheer volume of data, data quality is even more critical for improving model performance and industry adaptability.

At present, the industry is gradually shifting from general large language models (LLMs) to more lightweight, efficient, and easily deployable vertical small models at the edge. These models are better suited to meet the healthcare sector’s stringent requirements for privacy protection, low-latency response, and high reliability—especially in high-risk scenarios such as diagnosis and treatment, where the margin for error must be nearly zero. Building dedicated, high-quality data warehouses is a key pathway to improving model accuracy.

At the same time, the introduction of AI is significantly reducing the R&D and application costs of medical innovation. For example, one cell therapy company we invested in has developed a highly vertical AI small model focused on AI-assisted applications in its treatment area. Another company is focused on DNA sequencing data, building dedicated analytical models. What these companies have in common is that, by leveraging high-quality data and AI, they have achieved rapid product iteration and precise industry adoption.

In the financial sector, we invested in a startup dedicated to automating commercial paper issuance (Commercial Paper Issuance, or CP) with artificial intelligence. The company uses traditional AI methods such as reinforcement learning to fully automate the commercial paper issuance process. With a team of just seven people, they have already signed multiple Fortune 500 clients, with single transaction amounts reaching billions of dollars. Although their fee rates are low, the sheer scale of transactions has brought the company annual revenues in the tens of millions of dollars, and their overall financing needs remain minimal.

The key characteristics of these types of applications are: standardized processes, high repetitiveness, and large volumes of data. These make them ideal scenarios for AI to achieve efficiency gains and cost optimization. Although the financial industry is subject to strict regulation and has a complex structure, each segmented process contains ample opportunities for AI to integrate data and automate tasks.

In the logistics, supply chain, and manufacturing sectors, AI is deeply integrating with robotic systems. In industrial settings, robots are usually non-humanoid, such as robotic arms and automated handling equipment, gradually forming a more intelligent industrial operation system.

Finally, let’s talk about space technology. This is one of the areas I am personally very optimistic about, and it has also been a key focus of my investments in recent years. To some extent, space technology shares similarities with manufacturing or industrial automation—the key lies in how to choose the right edge devices to deploy AI.

In space scenarios, these edge devices are satellites. Each small satellite can carry AI models, becoming an intelligent node in space. Within this ecosystem, we can comprehensively deploy artificial intelligence at the infrastructure, model, and application layers. More importantly, the data collected from space is of extremely high quality and has broad potential application value. Although this field is still emerging, its integration with AI is accelerating, and the future prospects are vast.

Finding the Key Player

AsianFin: “Investing in AI is actually about investing in an ecosystem—finding the Key Player in the ecosystem to drive AI applications, rather than just investing in the applications themselves.” This is a very interesting perspective. Could you elaborate on this?

Zhang: AI applications are just the surface; what truly supports their development is the entire complex ecosystem behind them—from infrastructure, models, and architecture to the data layer. When the ecosystem is well integrated, the result is lower costs and higher efficiency.

Therefore, I believe that when evaluating an AI project, you can’t just look at whether the application itself is “well done”; you also need to deeply analyze whether the underlying ecosystem architecture has the capability to integrate. For example, Google’s self-developed TPU chips were created to fill a gap in its ecosystem at the computing layer, enabling full-stack integration from chips to applications and thereby reducing system-level costs. If a company relies on external vendors at every layer of its ecosystem, it will be very difficult to optimize its overall cost structure.

To elaborate further, the concept of “ecosystem” also means you need to pay attention to the entire value chain: who are the key players, who controls the data entry points, who has the capability to optimize models, and who can efficiently connect technology with real-world scenarios.

Model training itself is a highly complex systems engineering process. It’s not just about “feeding data”—it’s a continuous cycle of feedback, optimization, and iteration. These seemingly “underlying” core capabilities are actually the decisive factors that determine whether applications can be successfully implemented.

Therefore, when we make investments, we typically break down these questions repeatedly: Which roles are critical to this AI application? Can we integrate these key elements into our own system? Are we able to build our own data resources and data structure systems? Ultimately, can we optimize the overall application performance and significantly reduce operational costs?

Finally, I’d like to add an emerging phenomenon in the new ecosystem: Many tech companies today can be roughly divided into two categories. One category consists of truly innovative technology firms; the other is relatively “traditional” tech companies. The latter have strong customer resources and channel networks, but their technological innovation capabilities are relatively weak. These companies are now generally willing to engage in “joint sales” collaborations with startups.

So at this stage, the “ecosystem” encompasses not only foundational elements like technology, data, models, and applications, but also new collaboration mechanisms, new types of channel partners, and paths for joint innovation. These key players often play a decisive role in the success or failure of an AI startup.

I recommend that entrepreneurs approach AI projects with an “ecosystem mindset”: don’t just focus on what you’re doing yourself—think about who the “key variables” are within your ecosystem. Have you truly built a complete solution that reduces costs, increases efficiency, and meets industry needs?

This is precisely the core logic we have consistently adhered to throughout our long-term investment, observation, and support of AI startups.

AsianFin: You also invested in Jia Yangqing’s project (Lepton AI). Would you consider him a key player as you described?

Zhang: Absolutely. Their team is innovating in the field of AI infrastructure, especially excelling in GPU optimization. The company achieved rapid revenue growth after just its first round of financing, while keeping capital consumption extremely low.

In other words, their revenue is growing at a pace that even surpasses the rate at which they are burning through their funding—a phenomenon that is extremely rare among startups.

Jia himself, along with his team, has made significant contributions to the open-source ecosystem. They not only develop products but also actively participate in building the entire open-source ecosystem. From an ecosystem perspective, their influence within the open-source community makes them a very crucial ecosystem builder, which is exactly what I mean by a key player and a typical representative of this group.

AsianFin: Why are you always able to identify key players?

Zhang: Every quarter, we publish one or two in-depth industry research reports, many of which are publicly available on the Fusion Fund website. Recent reports include topics such as “AI Healthcare 2.0,” “Physical AI,” “AI Infrastructure,” “AI Fintech,” and “Space Technology.” Currently, we are also preparing a report related to the open-source ecosystem. These reports usually require two to three months of research, with the core goal of clarifying the boundary between signal and noise and identifying the key players in the industry ecosystem, including data providers, investors, strategic partners, and more.

Our research relies on an industry network that we have built up over many years. For example, we have established a CXO network consisting of 45 CTOs (Chief Technology Officers) from Fortune 500 companies. At the same time, we have built a network of technical experts, with members from leading organizations such as OpenAI and DeepMind, as well as AI leaders from institutions like NASA.

In addition, we have formed a super founders network with 62 successful serial entrepreneurs. Many of these members have worked closely with me since their early startup days, so when they launch new projects, they often reach out to us first, which enables us to get involved with the most promising innovations at the earliest stages.

We have also independently developed an AI analyst named Ada, which has been assisting us in data analysis and information screening since 2018, greatly improving our team’s efficiency.

It is precisely by leveraging this industry network and technical tools that we are able to efficiently identify early-stage high-potential companies and, with our deep industry insight, accurately find outstanding entrepreneurs worthy of support.

The Springtime for Serial Entrepreneurs

AsianFin: With the advent of AI, has the threshold for entrepreneurship been raised or lowered for founders?

Zhang: This question can be viewed from two perspectives.

From the standpoint of rapid innovation and swift product launches, AI has definitely lowered the barrier to entrepreneurship. In the US, some small business founders may not have strong programming skills themselves, but with AI programming tools, they can quickly build various applications and bring them to market for commercial operations.

On the other hand, because the pace of industry iteration has accelerated and the threshold for large-scale innovation has dropped, market competition has become much fiercer. In this environment, some companies may only be able to sustain themselves at the cash flow level. To capture a larger market share and grow into companies worth billions or even tens of billions, the challenge is indeed greater than before.

Innovative companies generally fall into two categories: one is those that develop well and become cash flow businesses, continuously generating considerable revenue. While these companies are stable in their operations, they may not necessarily achieve significant scale expansion. The other category is companies that are investable by VCs (venture capitalists). After investors come on board, they grow together with the company and help it rapidly develop into an industry giant worth billions or even tens of billions.

AsianFin: What is the typical profile of an AI entrepreneur in Silicon Valley?

Zhang: One of the most notable features of Silicon Valley is the high diversity of its innovation ecosystem.

For example, about 40% to 50% of current Silicon Valley residents are first-generation immigrants. These individuals were not born and raised in the US; instead, they came here to study or work and later chose to start businesses, especially among unicorn companies, where over 60% of founders are first-generation immigrants.

We’ve observed some interesting trends. In both the AI infrastructure and application layers, entrepreneurial teams are trending younger; at the same time, the number of serial entrepreneurs with industry experience continues to rise, giving them more advantages in vertical AI applications.

Looking back at our investments from 2015 to 2018, about 50% to 60% of founders were first-time entrepreneurs, while around 40% were serial entrepreneurs. However, in the past three or four years, we’ve seen a clear shift: now, in the companies we invest in, about 60% to 70% of founders are successful serial entrepreneurs.

AsianFin: What do you see as the core competitive advantage or key success factor for AI startups?

Zhang: The competition in today’s artificial intelligence sector is no longer just about models—it’s a contest of data and cost. Data is crucial in two main ways: its quantity and its quality.

If a startup can acquire or build proprietary, high-quality data assets, it will have a significant edge in both model development and real-world applications. The higher the data quality, the smaller the model required, which in turn reduces computing power consumption and energy costs, further improving overall profit margins.

Take, for example, the commercial paper (CP) company I mentioned earlier. The founder has a strong background in the financial industry, which allows them to access critical industry data and train models based on that data. This capability is often found in founders with industry experience or a track record of serial entrepreneurship.

Additionally, the ability to control costs is also a key factor for success in AI entrepreneurship. Only when costs are manageable can AI achieve large-scale, widespread industry adoption. Entrepreneurs who are adept at cost optimization tend to bring products to market faster, facilitate scalable applications, and ultimately generate substantial economic value.

Qualified Entrepreneurs

AsianFin: How do Silicon Valley VCs view the current hot fields of embodied intelligence and the low-altitude economy?

Zhang: In my view, every emerging tech trend goes through a phase where the signal-to-noise ratio is low—that is, there’s far more noise than genuine signals. This happens because the market is big enough to attract a flood of capital and entrepreneurs, leading to a natural round of elimination and selection through market mechanisms. This phase is inevitable and cannot be skipped.

Whether you’re an entrepreneur or an investor, the key is whether you truly understand which direction of innovation you value most within this trend. Over the past decade, we have never chased after the hottest trends, yet many of the sectors we entered early on have eventually become mainstream. This is thanks to our consistent methodology and systematic ecosystem analysis capabilities.

Before deciding whether to enter a particular field, we first build a comprehensive ecosystem map, identifying the key players, technological foundations, application requirements, cost structures, and more, followed by in-depth research and analysis. If we ultimately determine the opportunity is big enough, we will move forward with conviction—even if the market isn’t hot yet.

Almost all trending topics are born this way: entrepreneurs and capital flood in at a certain point in time, valuations soar rapidly, and similar cyclical booms occur every two or three years. The ones that can truly weather these cycles and survive are those companies that have had a clear vision and long-term perspective from the very beginning.

A few days ago, I shared a video with my team—an interview with Google co-founder Larry Page from the year 2000. At that time, he described the company’s vision as building a search engine that could answer any question—essentially what today’s AI aspires to be. Doesn’t this sound remarkably similar to current large language model products? In fact, the birth of the Transformer model originated from Google Research.

Take Elon Musk, for example. Although public opinion about him is now mixed, his visions for SpaceX and Tesla were established as early as twenty years ago and have been consistently pursued ever since. This shows that entrepreneurs who can truly endure through cycles often set clear, long-term goals when their companies are still small and their direction has yet to be recognized by the market.

Of course, a vision may not be realized within five or ten years, and a company must also have a viable business model and survival capabilities. But this clear, long-term vision is the fundamental driving force that keeps a business moving forward. Larry Page once said that such a vision, on an “intellectual level,” inspired his deepest motivation—and that is the true starting point of great entrepreneurship.

AsianFin: You just mentioned that having a clear vision is an important trait for entrepreneurs. If we set aside the industry and focus solely on the “person,” what kind of entrepreneur do you think is more likely to succeed?

Zhang: I believe the first and foremost quality is having a clear and unwavering long-term vision. This is not only a matter of strategic judgment, but also the core driving force behind a company’s sustained progress.

The second is exceptional resilience. Entrepreneurship is destined to be tough—it means high pressure, intense commitment, sacrificing social life and personal well-being, and even questioning oneself when the project shows no progress. Without resilience, it’s hard to go far. Sometimes, I deliberately “scare” founders by asking: Are you truly prepared to face ongoing loneliness and challenges?

The third is outstanding leadership. A truly effective founder must be able to attract top talent who are willing to “follow” them, especially in a place like Silicon Valley where competition for talent is extremely fierce.

These are the three core qualities I value most: a clear vision, strong resilience, and powerful leadership. The entrepreneurial journey is destined to be a lonely one, but it is precisely because of this that it highlights courage and value.

Being Acquired Is Just One Option

AsianFin: I’ve noticed that many of the examples you gave are in the B2B space, and you personally seem to focus more on B2B as well. Is this because the B2B ecosystem in Silicon Valley has a particular advantage?

Zhang: Over the past decade, we’ve invested in almost no B2C projects, and have always focused on the B2B sector.

One very important reason is that Silicon Valley truly has the world’s most mature and high-quality B2B startup ecosystem. This is reflected not only in the caliber of entrepreneurs, but even more so in the maturity of the entire business environment. For example, enterprise clients in Silicon Valley generally have a strong willingness to pay and are ready to spend on high-quality technology and services.

In addition, large companies in Silicon Valley typically have clear budget allocations—especially the budgets controlled by CTOs, which are often used for exploring cutting-edge technologies, external strategic partnerships, and potential technology acquisitions. This is crucial for startups.

For instance, these companies might invest in ten startups at the same time. Perhaps five of them won’t continue the partnership the next year, but they’re willing to bear this trial-and-error cost. More importantly, they’re open to giving feedback and co-creating through iterations. For example, if you give them a small order worth $7 million, once the results are validated, they can help you scale it across the entire company. This kind of mature business ecosystem allows startups to focus on technological and product innovation, rather than expending a huge amount of energy on survival validation and customer education in the early stages.

The second reason is closely related to the characteristics of the AI era. The emergence of AI has not only lowered the barriers to entrepreneurship for startups, but has also enhanced the capabilities of large tech companies. Giants like Google, Meta, Microsoft, and Salesforce are advancing technology at a pace that far exceeds market expectations.

In this context, if a startup wants to launch a product in the B2C market, it will inevitably face direct competition from these tech giants. More realistically, high-quality consumer data is mainly in the hands of big companies like Google and Meta, putting startups at a natural disadvantage in terms of data resources. I’m not saying there are no opportunities in B2C, but rather that the uncertainty and competitive pressure startups face in the consumer market are much greater.

Some investors even suggest that in the consumer market, perhaps 90% of the opportunities will ultimately be seized by big companies, leaving only 10% for startups. In the B2B market, however, even if large companies take half the market share, the remaining 50% still leaves significant room for startups. At the root of this is the control and distribution of data.

In addition, exit strategies in the B2B market are much more diverse. Besides IPOs, mergers and acquisitions account for a very high proportion of exits. Personally, I have experienced an acquisition exit as an entrepreneur. These types of acquisitions are often completed quickly, occur frequently, and cover a wide range of deal sizes—from tens of millions to several billion US dollars. According to statistics, about 80% of company exits in the market are realized through mergers and acquisitions. This provides us, as early-stage investment institutions, with more diversified and efficient exit options.

Therefore, taking into account the overall ecosystem, business logic, and my own experience, I believe we are better suited to invest in the B2B sector, especially since I myself started my entrepreneurial journey in the B2B space.

I have always firmly believed that true large-scale implementation and transformation of technology within industries is driven by innovation in B2B scenarios.

AsianFin: Large tech companies also invest in startups. Does this create competition with early-stage investment firms like ours?

Zhang: Actually, it doesn’t. On the contrary, it’s a very close and collaborative relationship.

For example, at NVIDIA’s annual conference this March, NVIDIA officially launched the DGX Cloud project, inviting only five venture capital firms as core partners, including A16Z, Insight Partners, and Fusion Fund. Portfolio companies of these partners, if selected after applying to the program, can receive a certain amount of free priority computing power and supporting resources. In reality, large tech companies hope to work with top VCs to secure promising innovative startups early on and provide them with various forms of support.

In recent years, our collaborations with these major companies have become increasingly close, and some of our portfolio companies have even been acquired by NVIDIA. On one hand, these tech giants build their ecosystems through strategic partnerships; on the other hand, they are accelerating mergers and acquisitions to integrate innovative technology resources, because their competitors are equally strong and they must rapidly expand and strengthen their own capabilities.

As you mentioned, investing in startups is not only about building an ecosystem, but also an effective way to bring outstanding companies under their umbrella through acquisitions.

AsianFin: Are entrepreneurs in Silicon Valley more open to acquisitions?

Zhang: That’s absolutely true. But it’s not just about having an open mindset—the terms of acquisitions in Silicon Valley are themselves very attractive, as tech companies here are willing to make strategic acquisitions at relatively high valuations.

Many people might assume that acquisitions are priced based on revenue multiples—five times, seven times, or ten times revenue—but the logic behind strategic acquisitions is actually quite different.

Take one of our portfolio startups as an example: its annual revenue is less than $10 million, yet it is currently in acquisition talks with a major publicly listed company. The reason the latter is interested in acquiring it is that the startup’s solution is highly compatible with its own product ecosystem. Once integrated, it could rapidly drive sales and generate billions of dollars in new annual revenue for the listed company. Against this backdrop, paying $500 million to acquire a company that has only been around for three years and has yet to reach $10 million in annual revenue is entirely justifiable.

From an entrepreneur’s perspective, such an offer already represents a highly attractive, high-multiple valuation. However, some serial entrepreneurs still feel the valuation is on the low side and hope for an even higher price. This clearly demonstrates the significant premium that technological innovation itself commands.

Therefore, for many entrepreneurs, mergers and acquisitions are not a last resort, but rather a way to realize value at different stages of development by optimizing equity holdings and exit strategies. For investors, it is likewise an ideal path to achieve high-multiple returns.

In reality, some companies are better suited to become a product line within a larger platform, rather than insisting on growing into an independent, publicly listed company.

Investing Accurately, Exiting Smoonthly

AsianFin: Silicon Valley is seen as the birthplace of venture capital and innovation. How do local investors view innovation cycles and long-term returns?

Zhang: Many people mistakenly believe that VCs in Silicon Valley are willing to accept extremely long investment cycles—say, 50 years—but that’s not actually the case. Most professional VC funds have a typical cycle of 10 to 15 years, with exits usually concentrated within 5 to 10 years. This is mainly determined by the fund’s own structural setup.

Of course, top institutions like Sequoia Capital in the US are experimenting with new operating models, such as holding shares in outstanding companies for the long term through “evergreen funds.” Even after a company goes public, they continue to participate as major shareholders in its long-term development. However, this does not mean that exit timelines can be extended indefinitely.

That said, compared to other regions, capital in Silicon Valley is indeed more “patient.” Among our fund’s limited partners, 70% come from long-term capital sources such as sovereign wealth funds, university endowments, pension funds, and insurance companies. They support our focus on early-stage investments and care more about long-term returns. This capital structure shapes our investment strategy, making us more inclined to back early-stage, disruptive innovation projects.

The reason we focus on early-stage investments is not only because we excel at building technology and driving early commercialization, but also because we have a rational understanding of the exit cycles and risks associated with early-stage projects.

We typically categorize projects into two types: one consists of deep innovation projects that are long-term, technology-driven, and have a lengthy commercialization cycle—within each fund cycle, we only invest in a select few of these. The other type includes projects with a clear commercialization path and the potential for rapid revenue growth in the short term, which make up a relatively larger proportion. This portfolio mix helps us balance long-term value while maintaining overall return stability.

From a broader perspective, the innovation cycle in Silicon Valley often follows a three-step loop: foundational technology innovation, technological application innovation, and business model innovation.

Take the development of the internet as an example—it was the evolution of underlying technologies that fueled the emergence of a multitude of new business models. The current wave of artificial intelligence, at its core, is part of a broader process of industrial digitalization. AI is not an isolated technology; it is closely linked to foundational infrastructure such as low-cost sensors and data collection networks. Only when the underlying infrastructure is robust can high-quality data be gathered, which in turn drives AI model training and algorithm iteration, ultimately enabling commercial implementation.

This is the essence of innovation in Silicon Valley: foundational technology drives breakthroughs in applications, which in turn lead to the restructuring of business models, forming a continuously evolving ecosystem.

AsianFin: We just discussed the issue of returns, and you mentioned that many projects have achieved very high multiples. What’s the key secret behind this? Is it about investing early, or about exiting well?

Zhang: Both are crucial. As early-stage investors, timing is extremely important—if you invest too early, the wait can be too long; if you invest too late, valuations are high and returns are limited.

But even more important is not just investing accurately, but also managing well. We devote significant effort to post-investment management, serving as board members in 60%-70% of our portfolio companies, maintaining frequent communication with founders, and being deeply involved in strategic planning and key milestones. For first-time founders, for example, we provide systematic training materials, assist them in building board structures, developing fundraising strategies, and better planning their financing and equity mechanisms.

In addition, our CXO network—which includes 45 CTOs from Fortune Global 500 companies—helps entrepreneurs accelerate product validation, connect with customers, and drive strategic partnerships. Even when acquisition offers arise, we assist them with evaluation, negotiations, and even bring in multiple bidders to optimize exit pricing.

So, achieving high returns on projects is not just about investing early; it’s even more about our systematic selection process and solid post-investment empowerment capabilities—areas where our team truly excels.

AsianFin: You’ve been investing for many years and still maintain such energy. How do you keep your passion alive?

Zhang: I’m truly passionate about this career. From the very beginning, I treated founding this company as my own “second entrepreneurial journey,” with a very clear goal: to build a globally leading venture capital firm and make a real impact in Silicon Valley, the world’s innovation hub.

The core driving force behind my ongoing commitment is the desire to bring about real change in industries through technological innovation. In my early days, I chose to get hands-on and start a company myself; now, as an investor, I hope to empower the ecosystem and indirectly drive technological progress and business growth. To date, we’ve invested in nearly a hundred companies. We’re not just shareholders—we’re deeply involved in the board decisions of many companies, truly becoming their long-term value partners.

With my technical background, I’ve always been curious about cutting-edge technology. Engaging daily with outstanding founders and learning about the latest tech trends is energizing in itself—I don’t even need coffee to stay alert. For me, this job is full of fun and challenges, and my motivation comes entirely from within, not from external pressures.

In addition, I consciously protect my passion. For example, we never blindly chase trends; instead, we focus on areas we genuinely believe in, where we have deep understanding and strengths. This sense of conviction helps us stay clear-headed and passionate in our investments, without being swept up by external anxieties.

Our entire team is driven by passion as well. We’re not just pursuing financial returns—we care more about creating long-term value. Financial returns are just one way to quantify our achievements. What truly motivates us is the real and lasting impact we create by driving innovation and supporting outstanding entrepreneurs.