He believes that the internet is a revolution in the "distribution mechanism," while AI is a disruption of the underlying technology.

Benchmark partner Eric Vishria recently had a fascinating conversation with Banana Capital partner Turner Novak on his podcast, The Peel, which I found to be an excellent interview.

In this conversation, Eric Vishria shared his views on entrepreneurship in the AI era, the types of founders they prefer, Benchmark's investment strategy, and how decisions are made internally.

Combining this with some of my previous articles about Benchmark might yield better results: "4x Valuation Increase in 2 Months, Benchmark Invests in the Fastest Growing AI Applications" and "Behind Benchmark's $500 Million Valuation Investment in Manus and Its Investment Strategy."

A few points he mentioned left a deep impression on me, such as the growth in the AI era completely surpasses traditional normal patterns; it is exponential growth that disrupts the SaaS companies' 3-3-3-2-2 growth rule.

Regarding founders, he focuses more on the founder's narrative ability, intellectual sincerity, and continuous learning ability. The ability of founders to construct a narrative for their companies is crucial. "In all scenarios, those who excel at continuously optimizing narratives ultimately prevail." At the same time, a core trait of successful founders is their learning ability: "Compared to initial experience, I care more about the founder's learning slope and first-principles thinking ability."

The best founders often possess two traits: extreme optimism and extreme skepticism—they are full of faith in their mission and business prospects but remain vigilant about everything else.

He believes that the internet is a revolution in the "distribution mechanism," while AI is a disruption of the underlying technology, closer to a "transistor" type of empowerment logic. The former addresses "connection efficiency," while the latter changes the "essence of creation."

Regarding Benchmark's own investment strategy and methods, they adhere to the core strategy of "seeking epoch-making companies and supporting the most visionary entrepreneurs." Benchmark's flat partner structure fosters high trust, ensuring that all members work together to support the invested companies.

He believes that the scarce resource is always "extraordinary companies," not capital. Therefore, Benchmark insists on a streamlined investment strategy; once they invest, everyone will deeply accompany the journey. He likens Benchmark to a communist collective of capitalists.

Notably, Eric Vishria led investments in two Chinese AI projects, Manus and Fireworks, with Fireworks already valued at $4 billion and an ARR exceeding $100 million: "VCs Are Starting to Do Roll-ups, This Chinese AI's Latest Valuation is $4 Billion with an ARR Over $100 Million."

Below, I have summarized this interview. Due to the length of the content, there may inevitably be some errors, but this conversation has provided me with a lot of inspiration:

1. Turner Novak: You mentioned that for every investment you've made, you almost know from the moment you engage with the project that this is a founder worth supporting. Can you elaborate on this?

Eric Vishria: I wouldn't say "I know immediately," but I can indeed build confidence quickly. Some VCs are naturally optimistic; we tend to be more easily excited, while others are more inclined to maintain a skeptical attitude. I belong to the former group, easily sparked by interest, and then adjust my judgment through due diligence and feedback from partners—that's roughly my working style.

Everyone has a different approach to "excitement," but for me, when I meet an entrepreneur, if I can quickly gain new insights—especially after spending a lot of time in a certain field, if something the entrepreneur says suddenly makes me realize: "They have unique insights." This moment of "epiphany" is crucial.

I encounter new companies, engage with people, and read a lot of information every day, but very few can present a viewpoint that refreshes my perspective. If someone can provide unprecedented insights or interpret the market from a completely new angle, that is an excellent signal for me. It’s important to clarify that this is usually unrelated to data metrics—I have almost never made an investment decision based on a number; the key lies in "insight," such as a unique understanding or analytical perspective on the market.

A long time ago, I collaborated with a designer from Ido who said something that has stuck with me: "Many times, the most important thing is not the solution, but correctly defining the problem." I believe this is akin to the "insight" in investing—when everyone is viewing the market or opportunity in a fixed pattern, if someone can propose a new approach, saying "we can enter this way," and it makes you feel "this logic makes sense," that cognitive collision becomes a key decision-making starting point, which contains an element of "insight."

The second thing that often impresses me is when you encounter a truly "learner" type of founder. I have previously described this feeling: some founders walk in as if they have a vacuum cleaner attached to your head, sucking all the knowledge out, right? When you have this feeling—like someone is directly downloading the content of your brain into theirs—you think: "Wow, if this person absorbs knowledge like this three times a week for ten years, that’s the knowledge equivalent of 1,500 brains. That’s incredible." You can almost tangibly feel this level of learning ability or knowledge compounding.

In fact, this makes me realize that compared to the "maturity" of entrepreneurs at the early stage of their startups (such as their understanding of company building), I care more about their "learning slope"—that is, their learning speed. If the learning curve is steep enough, they can quickly surpass everyone through compounding. Of course, this situation is rare, but when you occasionally meet an entrepreneur who possesses both "insight" and "learning ability," and you have a chemical reaction with them, it is truly exciting.

Some may ask: "How do you determine if someone is good at learning? After all, anyone can claim to be a good learner." This is indeed very subjective; it’s more of a feeling. Perhaps you can gauge it from the questions they ask: when you pose a question to them, can they grasp the deeper logic behind it? Are they digging for insights? Are they trying to get to the bottom of things? Are they thinking using "first principles"? For example, you might feel: "This person is reasoning from the essence."

Additionally, do they dare to ask "seemingly foolish" questions? Are they willing to question the most basic concepts? For instance, when you express a viewpoint, do they directly ask: "Why do you say that?" This combination of confidence and a willingness to humbly seek knowledge is often a good sign.

Another aspect is reflected in their "narrative evolution"—about the company's positioning, reason for existence, and winning logic, whether they can form a coherent story. Interestingly, we often make very early-stage investments at Benchmark, and in the projects I lead, perhaps one-third to half are the company's "first financing." These projects often give me a sense of "directly aiming for success," even though no one knows whether they will ultimately succeed—early-stage investments are inherently high-risk; some companies will succeed, and some will not, but that’s normal.

From my experience, even if the team is excellent and the market is right, external factors can also determine success or failure. But the most special thing about Silicon Valley is that one failure doesn’t mean much; it’s just "this time it didn’t work out," and that’s completely acceptable. When you hear a startup story, you think: "If this works out, it could really change the world." This possibility is the charm of early-stage investing.

Moreover, this story must be coherent enough. We have seen too many examples: although it’s uncertain whether it will succeed, the entire logical framework is tightly knit. Projects like this are always my favorite investment targets—the best founders and the most promising products and companies often possess this quality. When you introduce it to others, you might say: "I’m not sure if this will work, but they are doing this, this is their core argument, and this is the problem they are trying to solve." After saying this, you feel: "Hmm, this could be a brilliant idea."

Take Cerebras as an example; it is a company focused on AI chips and systems that recently filed for an IPO. Although it has gone through various controversies, what impressed me the most was that I started collaborating with its founder and CEO Andrew from the early days of the startup— the company was co-founded by five founders, and as a semiconductor company, it is not only capital-intensive but also faces many technical challenges. The problem is, I know nothing about hardware; I am completely an outsider.

That was in March 2016 when Andrew first came to us with the project. At that time, the semiconductor industry was far from being a hot topic; Nvidia's market value was less than $30 billion (now it exceeds $1 trillion), and Google had not yet released TPU, making the entire AI hardware field almost a blank slate. How did he impress us without any publicly successful cases?

The answer is simple: the first slide was an introduction to the team, and the second slide was also about the team. The founding team members were all serial entrepreneurs in the semiconductor and systems fields, with resumes that were "professional-level." Even though we were not clear at the time that this was a semiconductor company, just based on the team background, we could feel their professionalism.

Next, they presented their core argument: "GPUs are actually not suitable for deep learning; they are just 100 times stronger than CPUs." This viewpoint was provocative at the time—remember, this was before the Transformer era, and OpenAI had not yet been established. They then explained: "Why have graphics processors become the solution for AI or deep learning? Perhaps that’s not the case."

They elaborated on the characteristics of the workload and the solution they were going to build. You couldn’t help but think: "This might work; if successful, the value would be immeasurable." And the core of venture capital is precisely this "asymmetric possibility," right?

2. Turner Novak: Speaking of "narrative," Saji from Benchling mentioned that he learned the importance of company narrative from you. Why is narrative so critical?

Eric Vishria: I believe that as a founder or CEO, you are essentially "continuously telling a story." Engineers may scoff at "selling," thinking it sounds like peddling snake oil. But what I mean by "narrative" is not derogatory; it refers to the need to clearly articulate why the company exists, why you are the ones to do this, what the core issues are, where the competitive advantages lie, and how to win.

Whether you are facing customers, potential employees, partners, or investors, you are conveying this story and need to constantly iterate and refine it. This is one of the core responsibilities of a CEO. Excellent narrative ability is crucial because when smart people question "this logic has flaws," you will realize: "Oh, there is indeed a problem that needs to be corrected."

In reality, companies will always encounter various challenges—high customer concentration, stagnating growth, technological bottlenecks, etc. Some issues stem from reasonable factors that require clear explanation; others expose deeper hidden dangers that require reflection and resolution. The touchstone for all of this is the company's narrative logic. Ultimately, the CEO must lead and refine this narrative. I have personally witnessed a stark difference between two types of CEOs: one who diligently hones the narrative and another who dismisses it. In the end, the former often goes further.

3. Turner Novak: Do you know why some CEOs are more inclined to deeply refine their narratives while others are not? Is there a certain trait that makes you appreciate a particular CEO more or makes you realize they need to strengthen this capability?

Eric Vishria: I think this is somewhat related to "ambition"—if a person is extremely ambitious, the story they tell will inevitably be grand and can break down every layer of logic into every specific action at the moment, like peeling an onion.

Take Elon Musk as an example; he is a master of narrative. "Colonizing Mars" sounds far-fetched, but he has made countless people believe in this vision. More impressively, his narrative goes beyond "going to Mars" to "colonizing Mars," and even "making humanity a multi-planetary species"—the scale of this ambition is almost unmatched.

But he can break down the grandest vision into specific actions: we need to launch rockets → rockets must be reusable → develop multiple models of rockets → use rockets to transport satellites → develop robots to handle space tasks… His narrative consistently unfolds around core logic, even claiming, "We are not a car company; we are a battery company." The key is that he genuinely believes in this story, rather than merely persuading others, which is a perfect example of the essence of ambition and the power of narrative.

4. Turner Novak: Suppose you meet a founder whose new insights impress you, who is skilled in narrative, ambitious, and has an impressive team. How do you typically establish a connection with such a founder to seek collaboration opportunities? After all, you will be working together for many years; what is that relationship usually like?

Eric Vishria: For me, everyone has different motivations for being a VC. I was an entrepreneur, so my core motivation is "to collaborate with founders," which is also why I love this job. The chemistry with founders is crucial, and I spend a lot of time with them—at Benchmark, we often have cases where we have worked with founders for over ten years.

When I encounter such founders, I first delve into the business they are building: Why are they doing this? What is their motivation? How does it work specifically? What are the driving factors behind it? What plans do they have for the development path? What insights can I bring from the experiences of other companies? The core of this interaction is how to maximize the probability of realizing the founder's vision and help them turn their ambition into reality. If we can increase the probability of success even a little each quarter, the compounding effect will become apparent.

5. Turner Novak: Are there specific actions that can most enhance this probability of success? For example, small things like preparing for board meetings in advance?

Eric Vishria: There are indeed many details, and each company and founder is different. But regardless of whether product-market fit (PMF) is achieved, the core lies in "how to build a sustainable company." Excellent companies must possess durability and resilience.

Sometimes you encounter founders with great personal charisma, but the key question is: do they really want to build a company that "transcends the individual"? At least for now or in the future, the company needs to harness the power of many people to create a synergistic effect. Therefore, founders must have the ability to build a system around the company, which involves what kind of people to hire, what the strengths of team members are, and how to complement the founder's own abilities—I invest a lot of energy in these areas.

Interestingly, I listened to Ben Thompson's podcast interview yesterday. He is an excellent industry commentator, and from the interview, you can sense that Mark Zuckerberg is always paying attention to market dynamics and understanding industry trends. This indicates that excellent founders not only need to be skilled in narrative and team building but also must maintain a keen awareness of the external world, continuously iterating their understanding of the business—this may be one of the key details that enhance the probability of success.

He learned lessons from this. For me, the most interesting point he mentioned is his regret over not being able to control the mobile platform. This also explains some issues, such as why he insists on open-sourcing the Llama model and why he has invested billions of dollars in developing Llama. One reason is: "I don't want another platform controlled by others that I have no influence over."

Clearly, he has many grievances against Apple, and the relationship between the two is quite tense. But once you understand this, you will grasp his psychological logic—he does not want OpenAI, Anthropic, or other companies to control the core models and become the new platforms, while Facebook (Meta) has to rely on these platforms, whether for advertising assets or other businesses. Once you figure this out, everything makes sense.

The key is that as the company scales, much of the founder's work is essentially to keep the company running smoothly and ensure business continuity. Part of this work should allow the founder to "lift their head" and have a longer-term strategic perspective, gaining insights into industry trends and changes—this is one of the core capabilities of a founder. In my view, the philosophy of building a business is complementary to the logic mentioned above.

6. Turner Novak: How do you think this trend will change in the future? There is now a saying (perhaps no longer a joke) that "one person can build a billion-dollar company." How do you view this change in your portfolio or observations?

Eric Vishria: I believe this process will be full of unexpected twists and turns. For example, in Benchmark's portfolio, there are several companies with fewer than 100 employees that went from startup to achieving over $100 million in annual revenue in just 12 to 18 months. This speed is not two or three times that of traditional SaaS companies, but five to ten times. Of course, there are differences: annual revenue may include experimental income, which may not be entirely reliable. But setting those factors aside, whether from human efficiency or growth speed, it is astonishing—and this is largely due to AI technology.

7. Turner Novak: A few months ago, you tweeted about the "growth rules" of the traditional SaaS industry (for example, first achieving $1 million in revenue, then growing at rates of 3x, 3x, 2x, 2x, ultimately reaching $100 million in revenue at the time of IPO), but this set of rules has now been completely overturned, especially in the IPO phase. What do you think is the reason for these companies growing faster? Is it a surge in market demand?

Eric Vishria: The only conclusion we can draw at this point is that customers find the product experience "magical," and therefore are willing to pay for it. But how long can this willingness to pay last? Is the product replaceable? Is there a moat? Is it sustainable? These questions remain unresolved and vary by company and product. But whenever we see a product in a new field growing at such speed, we must acknowledge: its product must possess some kind of "magic" that effectively addresses user pain points, making users willing to "actively pay."

8. Turner Novak: What is the fastest-growing product you have seen? (Assuming it can be publicly discussed)

Eric Vishria: I believe ChatGPT is undoubtedly the fastest-growing product in history. Besides that, in our portfolio, nearly half of the companies have achieved over $100 million in revenue within 18 months from zero—this speed can be described as "light speed." The growth models vary: some rely on one-time $20 subscriptions (like tools such as Cursor), while others depend on large orders of $5 million to $10 million. This diversity and flexibility are astonishing.

9. Turner Novak: If I were an investor or founder hoping to build a company that can last for 10 years, change the world, and achieve billions in revenue, how should I assess the "revenue quality" of the current business?

Eric Vishria: I believe that the best founders often possess two traits: extreme optimism and extreme skepticism—they are full of faith in their mission and business prospects but remain vigilant about everything else. This skepticism drives them to act quickly, and they also possess a high degree of "intellectual honesty": regardless of how they express it externally, they always know where the company's moat is and where the weaknesses lie.

Take many of the rapidly growing AI companies today as an example; their moats may still be quite weak, but they have a "speed moat"—by leveraging faster iteration speeds and slight advantages, they continuously stay ahead of competitors, and this advantage may transform into a real barrier over time.

10. Turner Novak: You mentioned the issue of "limited moats," which raises the question: in an era where many processes can be automated, what constitutes a truly lasting moat? For example, do traditional "switching costs" still exist? Suppose there is a new CRM tool that can automate the copying and pasting of all data and clone all the integration features of Salesforce, with users facing almost no switching costs—would such a business have value?

Eric Vishria: Perhaps we can find answers in the case of Google Search. Why was Google able to win in the early search engine competition?

1. Product superiority: Whether it was the PageRank algorithm, web crawling capabilities, or the relevance of search results, Google significantly outperformed contemporaries like Yahoo and AltaVista.

2. Performance experience wins: The loading speed was extremely fast, which was a decisive advantage in the internet environment at the time.

3. Simplicity and lack of distractions: The page was not cluttered with low-quality display ads, providing a pure user experience.

However, it is worth noting that Google did not achieve profitability in its early days (when it was founded in 1998) and only launched AdWords at the end of 2001. This model was not original (inspired by Bill Gross's Goto.com), but Google turned it into a commercial miracle thanks to its massive search traffic and superior execution capabilities. This illustrates that creating a "magical product" is the starting point, but the more critical aspect is to maintain the product's lead over time—the latter is far more challenging than the former.

Social networks build moats through "network effects," while Google ultimately formed a "advertiser-user" bilateral network effect and solidified its position by controlling ecosystem elements such as browser entry points, operating systems (Android), and hardware (Chromebook). Comparing Google in 2000 (which was merely a "faster, more accurate but unprofitable search engine") with today's AI startups reveals that history is repeating itself: the seemingly thin "product advantage" in the early stages may become the starting point for a moat over the next decade.

Returning to the current AI market, investors' future growth expectations for certain companies may seem aggressive, but considering the overall scale and potential value of the industry, this optimism is not unfounded. After all, the iteration speed of AI products far exceeds that of traditional software, with rapid functional evolution. Of course, there are also voices of skepticism—such as OpenAI's current negative gross margin issue. But just as Google's early lack of profitability did not hinder its construction of long-term value, the key lies in whether the product addresses real and ongoing needs and whether it can establish an irreplaceable ecological niche through iteration.

In this uncertain field, every seemingly crazy vision can find reasons for support or opposition—this is precisely the allure of venture capital. The market is large enough, and changes are rapid enough that the ultimate winners may not be the companies with the "deepest moats" today, but those teams that can continuously create "irreplaceable product magic" and translate speed advantages into ecological barriers.

Perhaps the situation has changed (I certainly haven't seen their financial statements), but overall, I am not too worried about it. At least for now, most marginal costs are concentrated in the inference stage, and inference costs are rapidly declining. It's like betting on "Moore's Law"—history has shown that such bets are usually wise. Therefore, I do not believe that gross margin issues are a core concern worth worrying about.

Of course, there is indeed pricing pressure and a trend toward commoditization in the industry, while the costs of model development continue to rise. But perhaps as the importance of pre-training decreases and the importance of post-training increases, this cost growth will stabilize. So I believe that gross margin challenges are not really an issue at this stage (especially in the inference stage). If founders are willing to endure short-term pressure, time may actually become their friend—because the dividends of technological iteration will gradually become apparent.

11. Turner Novak: Suppose you are running a company with 12 to 36 months of funding reserves after financing; how do you assess the risk range? After all, you cannot accurately predict the speed of technological advancement.

Eric Vishria: Clearly, each company's situation is different, and strategies need to be continuously adjusted. However, if a company has growth momentum or escape velocity, it is easier to secure follow-up financing even as funds approach depletion—after all, the market for AI companies still has ample funding supply, and even with stock market fluctuations, quality projects still attract investment. But if a company lacks growth momentum and a compelling narrative logic, the risks will significantly increase.

12. Turner Novak: Where will the value in the AI field primarily concentrate over the next decade?

Eric Vishria: Interestingly, looking back at the early internet of the 1990s, infrastructure companies (like Cisco and Sun) were among the first winners. At that time, Nvidia faced over 90 GPU competitors but ultimately stood out due to its technological barriers. Similarly, in the first wave of the AI field, Nvidia is clearly the biggest winner—Peter Thiel once said that Nvidia captured 125% of the profits in the AI field (because other companies were still in the red), and this statement may even be conservative.

However, just as the scaling of infrastructure gave rise to consumer giants (like YouTube after the spread of broadband in the U.S. and the explosion of Instagram and Snapchat after the maturity of 4G networks), the AI field will also follow a similar pattern:

1. Infrastructure Layer: Currently, hardware and computing power companies represented by Nvidia dominate, solving the "computing power supply" issue.

2. Application Layer: As computing power costs decrease and model capabilities improve, a large number of consumer and enterprise applications will emerge in the future. For example, enterprise tools may deeply integrate AI in vertical fields (like Glean reconstructing internal enterprise search through LLM), while consumer products may form monopolies around "personalized experiences."

13. Turner Novak: As an investor, which direction are you currently more inclined to bet on?

Eric Vishria: The nature of our work makes it difficult to plan opportunities from the top down—the key is to discover founders who have deep insights into the market, understand technological boundaries, and can combine model capabilities with specific scenarios. However, the product development logic in the AI era is fundamentally different from that of traditional SaaS:

• SaaS Era: Founders start from "customer problems" and use mature technologies like cloud computing to provide better solutions (like Salesforce disrupting traditional CRM with a cloud model).

• AI Era: Founders need to start from "technological capabilities" and think about "how to apply model characteristics to specific fields." For example, the founders of Cursor deeply understand the reasoning boundaries of large language models, enabling them to develop precise programming assistance tools.

This reversal of logic means that technologically inclined founders may have an advantage—they need to be like Andrew from Cerebras, who understands semiconductor technology and can combine it with AI computing power needs.

In contrast, traditional SaaS giants mostly remain at the superficial level of "adding chatbots or autocomplete features to products," with very few cases that truly leverage AI to reconstruct business logic. The reasons behind this are, on one hand, the disruption of product development logic, and on the other hand, established companies always try to "protect existing businesses," when in fact, they should be all-in embracing new technologies.

As one of my partners said: "In the face of AI, you have two choices—either be disrupted twice or actively use it to reconstruct your business." This is not anyone's fault but a natural law of technological iteration. Teams that can break free from the inertia of "customer demand-driven" thinking and truly understand the essence of AI technology are the ones that may define new business rules in the next decade.

It's like the world has undergone a dramatic change before us. As you said, the market has redefined value, and this change has led to a slowdown in growth for traditional companies—new hotspots are emerging, and people's energy and resources are shifting accordingly, putting old businesses at risk of being disrupted.

14. Turner Novak: As a founder, how do you understand these technological waves and decide what to embrace? Looking back over the past five years, we have experienced the rise of AI, Web3, and other fields, but AI is clearly not a fleeting trend; it is a transformative force that will last for decades. Have you ever had a moment when you realized: "AI is fundamentally changing business logic"?

Eric Vishria: For me, the answer lies in observing the speed of technological iteration—the capabilities of models are improving at a visible pace, and more and more people are using AI to create practical value. This trend is exciting because it indicates that AI will permeate various fields such as biosciences, materials science, and mechanical engineering, potentially becoming a foundational enabling technology like the transistor.

Take the transistor as an example; this "switch" component, which emerged in the 1950s, was initially just a replacement for vacuum tubes, but now it exists in every mobile phone, headset, and camera, supporting the entire digital world. AI is showing similar potential: in the future, from smart microphones automatically capturing sound to cameras dynamically adjusting image quality based on light, almost all devices will embed AI capabilities. This "ubiquitous" characteristic makes AI the most transformative technological wave since electricity and the internet.

Speaking of how I personally use AI, I am actually a "simple user." Besides commonly used conversational tools like ChatGPT and Claude, I particularly enjoy voice interaction features—these are simply magical for families with children. For example, my 10-year-old son is obsessed with black holes and often uses AI voice to gain knowledge about them, even asking AI to create songs about his favorite TV shows or about when dad comes home from a business trip. This instant interaction not only satisfies the child's curiosity but also showcases the infinite possibilities of AI in education and creativity.

15. Turner Novak: Looking back since you entered the venture capital industry in 2014, what do you think is the biggest change in this industry?

Eric Vishria: The biggest change has been the intensification of competition and the explosive growth of capital supply. Today, the financing scale of startups far exceeds that of the past, driven by the expansion of market size and technological dividends, as well as the monetary environment. On the other hand, the "return ceiling" for investments is also rising—the new technological waves like AI are giving rise to more potential trillion-dollar companies.

From an industry perspective, over 70% of the top ten companies by market capitalization globally have received venture capital (such as Apple, Tesla, Microsoft, etc.), and this proportion may be even higher among the top 100 companies. This confirms the core role of venture capital in the process of commercializing technology: we are not just providers of funds but also bridges that drive scientific innovation and commercial implementation.

Looking ahead, each generation of venture capital firms is deeply tied to specific technological waves: Sequoia rose during the semiconductor era, Benchmark was established in the early internet days, and a16z emerged during the mobile internet era. Now, AI is shaping a new generation of firms—those investors who can understand the essence of technology and explore "technology-scenario" fit alongside founders will define the business landscape of the next decade in this transformation.

Just as transistors eventually became hidden yet supported everything, AI will also reconstruct every industry in a "subtle and pervasive" manner. As venture capital practitioners, our mission is to discover those ideas that can turn the "impossible" into the "inevitable" at the intersection of technology and business, and to nurture them with capital and resources to grow into the infrastructure of the next era.

The rise of these venture capital firms is often closely linked to specific technological waves of the era—they are the "new entrants" of the times and have established their industry positions by riding the wave. Of course, institutions like Sequoia, Benchmark, and a16z being able to maintain their influence also proves the "self-renewal" characteristic of this industry. Venture capital is essentially a "hustler's business"—we are often jokingly referred to as "high-end headhunters" or "money sellers" because persuading others to believe in an unproven vision is no easy task. This industry is full of competition, but it is precisely because of this that it always maintains vitality and innovation.

Regarding Benchmark's investment stage, the outside world often classifies us as "Series A investors," but the definition of "early stage" has long since become blurred. When some companies can go from founding to achieving $100 million in revenue in just a year, the traditional distinctions between "seed," "Series A," and "Series B" have lost their meaning. For us, the core logic has always been to "invest in the best companies as early as possible"—what we call "Series A" is more like the point at which "the first board-level partner joins," which occupies 80% of our investment scenarios. As for the specific stage names, they are not really important.

In the face of the current rapid market changes, our strategy is to "broaden our vision and remain flexible." Rising valuations and larger individual investment sizes are objective facts, but we never mechanically follow quantitative rules. The reason is simple: discovering an outstanding company is already difficult enough, and once encountered, factors like ownership and price take a back seat—the scarce resource is always "extraordinary companies," not capital.

Benchmark: The Communist Collective of Capitalists

What sets Benchmark apart is its small core team (usually 4-6 active investors) and fully aligned incentive mechanisms. We do not need to scale up, allowing us to focus on "high conviction" investments, purely seeking those companies that can define an era. In contrast, large institutions, due to complex team hierarchies and diverse individual career aspirations, have to establish rules to prevent mistakes, but this can also lead to missing out on genuine innovation opportunities—after all, writing a check is easy, but judging "which checks are worth writing" is the art.

This reminds me of the growth trajectory of startups: early founders collaborate based on trust, without cumbersome systems; as the scale expands, processes and norms need to be established. The venture capital industry is no different—when institutions expand to a certain extent, policies and processes naturally increase, but Benchmark chooses to go the other way: our investment strategy remains extremely simple—to seek generational companies capable of achieving billions in revenue, regardless of the past, present, or future, and nothing more.

Some may question the sustainability of this "unstructured" model, but in fact, the essence of venture capital has never changed: betting on the most ambitious entrepreneurs and supporting them in realizing seemingly crazy visions. As for the size of the fund (such as Benchmark's last fund of about $600 million), it has never been the key to success or failure—after all, who could have predicted in the 1990s that a "small and elite" institution could continuously capture world-changing companies amid the waves of the internet and AI?

Ultimately, the allure of the venture capital industry lies in its perpetual space for "exceptions"—those individuals and institutions unbound by rules, daring to bet on the "impossible" at the intersection of technology and business, will ultimately define the value coordinates of the next era.

Compared to other large funds, our fund size is relatively small. When faced with large funds that can offer "double the capital and triple the favorable terms," how do we differentiate ourselves? First, it is important to clarify that I inherit the Benchmark brand and the long-accumulated excellent performance—most partners in the current team are beneficiaries of this historical accumulation. Therefore, entrepreneurs usually understand that our investment flexibility is not limited by fund size, which is our unique advantage.

For us, the core competitiveness lies in the small team's focus on "deep partnerships." We invest far fewer times each year than most institutions because we must fully commit to collaborating with a few entrepreneurs, pouring our efforts into helping them succeed. The value of this model is that entrepreneurs can genuinely feel the "spirit of partnership"—we are not just investors but long-term allies fighting side by side.

Of course, some may counter that large institutions have vast team resources (such as recruiting teams, marketing teams, etc.) that can provide "army-style" support for companies. However, entrepreneurs who choose Benchmark often place more importance on whether board members can dedicate significant time to deeply engage in the business. After all, the core capabilities of any successful company (recruiting, engineering, sales, etc.) need to be built internally; the value of venture capital lies not in directly providing resources but in being "insightful listeners"—well-versed in company details while also stepping back from daily affairs to provide strategic perspectives.

For example, when a founder faces a tough decision of 51% versus 49%, generic advice is often ineffective, while our value lies in combining a deep understanding of the company with industry experience to propose targeted thoughts like "Have you considered the XX perspective?" This kind of interaction is hard to quantify but forms an irreplaceable bond of trust in collaboration.

Benchmark's "Equal Partner Model" is another key difference. We implement a completely flat management structure: capital allocation, decision-making power, and economic benefits are equally shared among partners, forming a "communist collective of capitalists." This structure eliminates internal competition and ensures that everyone fully supports every project—while in many institutions, there is a common sense of division over "whose project this is," here, all companies are "our projects."

In investment decision-making, we adopt a "high-trust advocacy model." When a partner is optimistic about a project, they invite other partners to participate in discussions (even in the initial meeting, there may be 2-3 partners present), generating a more comprehensive judgment through cross-domain perspectives. Although the final decision requires a vote, the core is ample dialogue and advocacy, rather than mechanical rules.

Reflecting on my personal experience: I graduated early from high school after completing all math and science courses, moved from Memphis, Tennessee, to California, studied mathematics and computer science at Stanford University, briefly worked in a tech investment bank, and then joined LoudCloud (an early cloud computing company) starting as an assistant, experiencing the transformation from LoudCloud to Opsware (from cloud services to cloud management software). This experience gave me a profound understanding of the ups and downs of technology entrepreneurship.

In 2010, I founded Rockmail (a social browser), which was acquired by Yahoo in 2013. This entrepreneurial experience made me appreciate the challenges faced by founders and laid the groundwork for my later joining Benchmark—Jim Goetz from Sequoia Capital had suggested I consider venture capital in 2008, and six years later, this seed sprouted, aligning perfectly with Benchmark's egalitarian culture.

My first investment was Confluent (a company based on the open-source project Kafka, now publicly traded), followed by Amplitude (a data analytics platform, also publicly traded). The transition from entrepreneur to investor has made me better understand the importance of a "learning mindset"—for example, Amplitude's founder, Spencer Skates, is a practitioner of "first principles," a trait particularly common among entrepreneurs with an MIT background.

Comparing the internet bubble of the 1990s with the current AI wave, I believe the biggest difference is that: the internet was a revolution of "distribution mechanisms," while AI is a disruption of underlying technology, closer to the empowering logic of "transistors." The former solved "connection efficiency," while the latter changes the "essence of creation." Just as transistors were initially undervalued but eventually permeated all electronic devices, AI will also reshape every industry in a more subtle yet profound way.

Take Fireworks, one of our investments, for example. This company primarily provides developers with tools that integrate various models, and they have a reasoning cloud platform. The founder was responsible for scaling the PyTorch engineering team at Facebook in its early days, and after leaving with a group of collaborators, they founded Fireworks. Initially, the company planned to offer PyTorch cloud services, but with the rise of generative AI, they realized they needed to elevate the level of abstraction, so they shifted to providing various open-source models, custom models, and model operation services.

It has proven that running large models is no easy task, and they have excelled in this area, rapidly expanding their business scale and becoming one of the fastest-growing companies in our portfolio, among the five or six companies experiencing explosive growth.

16. Turner Novak: Perhaps this is a simple question, what is the payment model for users of such services? Is it like paying with tokens? I think many pricing models are basically token-based. Do you think this model will change?

Eric Vishria: Of course it will. I believe all these business models will continue to evolve—from paying for computing resources to paying for tokens, and in the future, it may shift to other payment forms. What will the final form of the business model look like? It's hard to say. For many application-layer companies, they may charge consumers based on "results." For example, Sierra Company uses a results-based charging model in the customer service field, charging based on the number of resolved tickets. This model is interesting because it bypasses payments for labor or tokens and directly ties to business outcomes. In the infrastructure layer, the charging model may be more fundamental, but in any case, this model will continue to evolve.

Interestingly, if users clearly understand the value of the service to themselves and are willing to pay for results, this model may lead to faster sales conversions because the risks are lower, and the incentives are more aligned. Have you heard of cases where AWS container operations led to high costs? For instance, Coinbase once disclosed in its financial report a massive bill of $55 million due to data monitoring, which is a typical example of excessive resource consumption.

17. Turner Novak: After becoming an investor, how has your perception of this profession changed?

Eric Vishria: I realize that this is a unique and highly challenging job, entirely different from the role of a founder. When I was out raising funds as an entrepreneur, I thought the difficulty of investing lay in "selecting projects," and I still think so. While I prefer to support entrepreneurs and collaborate with them—this is the part I love—it's undeniable that "selection" remains the most challenging aspect.

The emotional ups and downs of investing are not as intense as those of entrepreneurship; there is no sharp daily pressure like that faced by entrepreneurs, but rather a long-term subtle anxiety. However, if you love learning, this job can be very appealing—every meeting is an opportunity to understand new markets and technologies, and to try to comprehend human nature, filled with wonderful chemical reactions.

18. Turner Novak: Saji from Benchling mentioned that you have insights into building management teams and recruitment. If I were a founder who just raised $20 million and had a small team, how should I start thinking about recruiting a sales director or engineering director and building a management team?

Eric Vishria: First, I firmly believe in the importance of team building and leadership development. What companies lack most is not people to execute tasks, but people to lead. The first step is to clarify the needs, which is easier said than done. For sales, for example, you need to first figure out who your target customers are, what type of sales you are pursuing, what the sales process looks like, and how quickly you think the process can scale.

At the same time, you should consider the company's culture and personality traits; only by clarifying these can you clearly outline the profile of the talent you need. Next, you need to engage with candidates from different backgrounds to gain a comprehensive understanding of the market situation, so you can determine "this is the person we need."

When recruiting, focus on "strengths first," looking for individuals with unique expertise rather than merely those who "have no obvious shortcomings"—the latter often leads to mediocrity, which is the arch-nemesis of startups. The world is indifferent to the existence of startups; a company's survival and growth require will, hard work, and sharpness, so you need to recruit those with outstanding strengths.

How do you find such people? Or rather, how do you avoid hiring mediocre individuals? I usually work with a group of trusted executive recruiters, like Andy Price from Artisal, with whom I've collaborated on many hiring projects—I've probably worked with him around 20 times. Recruiters can help manage the hiring process and assist in moving things forward. The key is to first clarify the hiring standards and then strictly evaluate candidates according to those standards, while also conducting thorough background checks—both positive and negative, personally calling to inquire about specific details, and taking the time to understand deeply. Often, recruiters will handle most of the background checks, but as a founder, you need to be personally involved because recruiters may have a vested interest in facilitating the hire.

19. Turner Novak: I once heard you say, "The best CEOs make all new mistakes." What does that mean?

Eric Vishria: It relates to learning. If someone is constantly learning, they won't repeat the same mistakes but will encounter new challenges and make new mistakes. This is actually a good thing—startups are inherently difficult, and making mistakes is inevitable; we shouldn't be afraid of making mistakes. Every fast-growing company may seem chaotic internally, with various issues, because entrepreneurship itself is hard.

The key is to distinguish between "problems arising from scale or rapid growth" and "strategic dead ends"—if you fall into a strategic dead end, you must find a way out. A strategic dead end refers to a situation where, for some reason, the company finds itself in a position that is difficult to sustain long-term.

20. Turner Novak: You previously mentioned the issue of companies going public; can you share your views on the current IPO environment?

Eric Vishria: Why are many companies choosing not to go public right now? We are in a unique market environment. After many companies went public in 2021, the market experienced a pullback, and economic growth has slowed, leading to a somewhat stagnant current environment. Although companies like CoreWeave have successfully gone public and more companies plan to do so, the overall atmosphere still seems cautious.

I believe going public is a good thing, like entering the major leagues, as it can make a company more mature. Although stock market volatility may bring shocks, the largest tech companies in the world are all publicly traded. The argument against going public often states that it requires facing cumbersome regulatory requirements. However, large private companies should already be conducting audits and other regulatory operations; going public only adds a small number of financial and legal personnel, which doesn't have a significant impact.

Another viewpoint is that public companies need to focus on short-term performance, while private companies can concentrate on long-term development. But in reality, private companies also need to deal with capital fluctuations, and truly excellent companies, whether public or not, can continue to innovate—like Tesla, Google, Microsoft, and Apple, which have continued to expand new businesses even after going public.

20. Turner Novak: As an early-stage investor, what have you learned from public market investors?

Eric Vishria: A lot, with the most profound lesson coming from 2021—that even excellent companies can be overvalued. This has now been deeply ingrained in my mind.

20. Turner Novak: So how do you view valuations?

Eric Vishria: Early-stage valuations are completely different from those at the scaling stage. Early valuations are primarily based on judgments about potential outcomes and the balance of risk and reward, which is very subjective and difficult to quantify with specific numbers. In contrast, the valuations of public companies have more reference indicators and data; although there is still a margin of error, they are more objective. In short, this is a fascinating topic.

End!

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。