727 FinanceThe crypto fund has expanded into the a16z Frontier exploration fund, extending its investment portfolio to hardware, robotics, and the open-source AI ecosystem, with three godfather-level Silicon Valley investors providing in-depth analysis of future technology waves and entrepreneurial opportunities.
Compiled & Edited by: Shenchao TechFlow
Shenchao Introduction: In this podcast episode, top Silicon Valley venture capital firm a16z officially announced the restructuring and expansion of its originally blockchain-focused a16z crypto fund into the new "a16z Frontier" frontier exploration fund. Led by Chris Dixon, the fund will expand its investment scope by leaps and bounds from purely on-chain infrastructure to a broader frontier of hardcore physical and digital domains, including hardware, embodied robotics, entertainment and gaming, space biology, and decentralized open-source intelligence. a16z co-founders Marc Andreessen and Ben Horowitz, together with Chris Dixon, appear on camera to engage in an in-depth conversation on the evolution of technology cycles, the restructuring of power in the AI era, and the era's opportunities for entrepreneurs.
The three Silicon Valley godfather-level figures noted in the interview that the world today is at a critical moment when the cognitive gap between early technology explorers and the general public is the widest in history. They explicitly refuted the pessimistic narrative that "AI will inevitably lead to centralization," deeply analyzed the core value of open-source AI and cryptographic protocols in countering monopoly, stated bluntly that the commercial contract of traditional internet protocols has been broken by the giants, and called on a new generation of entrepreneurs to step out of the trap of mediocre minor feature optimization, to bravely ask "Is my vision big enough," and to become true builders in the wild frontier where software and the physical world intersect.
Guests: Marc Andreessen (co-founder and general partner of a16z), Ben Horowitz (co-founder and general partner of a16z), Chris Dixon (founding partner of a16z Frontier / a16z crypto)
Host: Erik Torenberg (renowned tech investor, founder of Turpentine, special guest host of the a16z podcast)
Podcast source: a16z official special (https://www.youtube.com/watch?v=-9XQMHhpypc)
Original title: The Frontier Ideas a16z Wants to Fund Now | Marc Andreessen, Ben Horowitz & Chris Dixon
Air date: October 22, 2026
Key Takeaways
a16z crypto fully advances to a16z Frontier: The original crypto fund has officially expanded into a broader frontier technology fund, no longer confined to a single blockchain narrative, but fully funding hardcore pioneers at the outermost edge of technological possibility, covering hardware, robotics, entertainment media, and biological sciences.
A historic and extremely wide gap has emerged between early explorers and public perception: Ordinary people still view AI as a simple conversational toy, while true early builders have already fully entered the era of autonomous agents, local large-model clusters, and super productivity. This cognitive potential energy gap contains enormous dividends.
Step out of the mediocre "product manager brain" trap: Marc Andreessen sharply criticized Silicon Valley's incremental improvement mindset that over-relies on A/B testing, fine-tuning conversion rates, and other minutiae, emphasizing that only non-consensus leapfrog innovation and deep artistic taste can give birth to true platform-level breakthroughs.
Open-source AI is the ultimate trump card against monopolistic centralization: In response to Peter Thiel's earlier statement that "crypto is decentralized, AI is centralized," Andreessen and Dixon firmly believe that the combination of open-source large models and cryptographic ledgers is the only defense line preventing tech giants and bureaucratic regulators from colluding in rent-seeking.
The underlying contract of the internet is breaking down: The traditional model of "creators contribute content, platforms return traffic and revenue sharing" has been completely destroyed by AI crawling and closed-loop summaries. Building a new economic protocol with on-chain confirmation of rights, verifiable attribution, and digital scarcity is urgently needed.
A new form of physical computing is brewing an iPhone-level explosion: Smart glasses, environment-aware spatial computing, and companion AI are reshaping the human-computer interaction interface, ending the absolute monopoly of smartphones that has lasted for nearly two decades.
Core advice for today's young people: Reject step-by-step credentialism and mediocre career trajectories, study physics, mathematics, and underlying systems engineering in depth, skillfully harness AI tools to amplify individual productivity a hundredfold, and become super builders combining geek will and aesthetic taste.
II. Highlights
"The question the new generation of founders should ask themselves is no longer 'Is my ambition too big?' but 'Is my ambition big enough?'"
"There is now the widest gap in the history of technology between the technological possibilities understood by early adopters and what the general public and traditional management believe is possible."
"We have witnessed too many mediocre projects held hostage by the so-called 'product manager brain.' They are good at raising conversion rates by a thousandth, yet know nothing about leapfrog innovation that opens entirely new dimensions."
"Open source is not simple charity; it is the most powerful anti-monopoly weapon in the history of human technology. When the most advanced open-source models run freely on millions of edge devices around the world, the blockades of centralizers are doomed to fail."
"The internet's original economic contract has been unilaterally torn up by the giants. You contributed all your knowledge, AI used it for training, yet it no longer returns even a single independent click to you. We must rebuild this system with cryptography and on-chain ownership."
"Do not learn standardized skills that are about to be completely automated by algorithms. Dive into hardcore physics, mathematical fundamentals, and complex underlying systems, and become a person with first-principles judgment."
III. Main Text
1. Expanding a16z Frontier: From a single crypto track to the boundless frontier of hard tech
Host Erik Torenberg: Today a16z has a very major strategic announcement, namely the official expansion of the globally renowned a16z crypto fund into the "a16z Frontier" frontier exploration fund. Chris, what is the core thinking behind this?
Chris Dixon: Over the past few years, our team has been thinking deeply about the evolution of frontier technology cycles. When we first established a dedicated crypto fund in 2018, that was the most exciting paradigm shift on the technological frontier at the time. But today, with the explosion of generative AI, underlying compute, and modern cryptography, the frontier boundary of technology has undergone tremendous spillover and convergence.
We see that the best founders no longer confine themselves to a single isolated label. They are building decentralized networks while also developing embodied robots, local agents, new compute clusters, synthetic biology, and unprecedented new immersive media. To fully support this group of founders who dare to open up the unknown deep sea, we have decided to thoroughly broaden the fund's investment scope. From today on, not only blockchain, but all fields at the edge of human technological possibility will receive full funding through a16z Frontier. Founders should no longer ask whether their ideas are too radical, but whether their visions are grand enough.
Marc Andreessen: Let me add some historical context. In the fifteen years since a16z was founded, every time we faced a massive shift in the technology wave, we proactively evolved our organizational structure and investment strategy. From the early cloud computing and mobile internet, to later enterprise services, life sciences, and crypto networks, technology has never evolved at a single point; it always presents a networked chain reaction. Now, AI is pulling forward all the hardware, physics, and software accumulated over the past decades. Frontier represents a state of mind that is never satisfied with the status quo and continues to explore new continents.
Ben Horowitz: When we observe top entrepreneurs, true revolutions often occur in the gray zones where different disciplines collide. If a startup is building humanoid robots driven by open-source large models while also distributing data rights confirmation and micropayment rewards on-chain, it is hard to classify it simply as AI or Crypto. The establishment of a16z Frontier is meant to break down all artificially drawn track barriers. As long as you are exploring the true unknown frontier, this is your most steadfast backing.
2. Is "too early" still a fatal risk today?
Host Erik Torenberg: There is a saying in the venture capital world that is regarded as gospel: "In the tech industry, being too early and being completely wrong are almost indistinguishable." Marc, you pioneered the Netscape browser back in the 1990s and are widely recognized as one of the very few people who have repeatedly stood ahead of the times. In today's rapidly iterating cycle, is "too early" still a danger?Marc Andreessen: This is indeed a classic cautionary maxim, because for the past few decades, if your business idea was ahead of its time, you had to endure a painful wait for the external supporting infrastructure to mature. For example, twenty years ago you wanted to build online streaming video, but at the time American households were still on dial-up internet—no fiber broadband, no smartphones, no cloud-native data centers—and your company would have completely burned through its cash flow before the infrastructure caught up.
But something extremely striking has happened today: the timeline of underlying technological evolution has been radically compressed. Startups today no longer need to lay cables from scratch or build their own chip fabs; global public clouds, powerful open-source foundation models, and high-density mobile devices are already ubiquitous. More importantly, AI itself is a super-accelerator, compressing what used to take years—coding, prototyping, molecular simulation, and hardware rendering—into days or even hours. So today, the time gap created by "being too early" has been largely leveled. Conversely, the real danger now is "not being bold enough": if you build only a product that is slightly six months ahead, the tech giants and open-source communities will bury you before you even officially launch.
Chris Dixon: I completely agree with Marc's judgment. Many people said ten years ago that smart contracts and decentralized compute were too early, but it is precisely because of those foundational investments that we have today's censorship-resistant, high-throughput distributed settlement networks. In the Frontier category, we are willing to bear the so-called risk of "being too early," because truly outsized returns and world-changing mega-companies are always born at a stage when mainstream consensus is still mired in confusion and doubt.
3. A Historic Cognitive Gap: The Disconnect Between Early Explorers and the Mainstream Public
Host Erik Torenberg: In your recent public interviews and articles, you have repeatedly mentioned that there is an extremely rare cognitive rift in today's tech world. How large exactly is the fault line you observe?
Chris Dixon: I think this is the widest gap between early adopters and mass understanding in the history of the computer industry. Look around you: mainstream society, executives at traditional multinational corporations, and even the vast majority of regulatory officials still understand AI as an intelligent chat box that occasionally drafts documents or polishes résumés. They think it is just another office productivity plug-in.
Yet if you go deep into the development scenes of San Francisco or the world's top geeks, you will find that early builders have already entered a completely different universe. They use hundreds or thousands of autonomous AI agents working together to write complex operating systems; robots complete reinforcement learning in simulated environments tens of thousands of times faster than humans; synthetic biology researchers use deep neural networks to design highly active proteins that do not exist in nature within weeks. This cognitive difference is exponential.
Marc Andreessen: This is exactly the same scenario as how ordinary people viewed the internet in the early 1990s. At the time, the vast majority thought web pages were merely electronic newspapers and simply could not imagine that the internet would reshape global finance, supply chains, and geopolitics. This enormous cognitive potential energy gap is precisely the entrepreneur's most disruptive moat. While incumbents are still examining new things through the lens of the old system, the true builders have already reconstructed the entire foundation with new tools.
4. A Grand Analysis of Where Profits Land: Do the AI Dividends Belong to Incumbent Giants or Startups?
Host Erik Torenberg: The most fiercely debated question on Wall Street and in the VC world over the past two years is whether the massive value created by AI will ultimately be swallowed by existing tech giants or will give birth to a batch of brand-new trillion-dollar startups. Ben, you have lived through every cycle from mainframes and PCs to the internet and mobile. How do you see profit distribution in this round?
Ben Horowitz: So far, on the surface, it does appear that giants such as Microsoft, Google, Amazon, and Nvidia have seized the first-mover advantage, because they possess ready-made massive distribution networks, compute budgets, and vast enterprise customer bases. Plugging existing AI models into the incumbent Office suite or search engines is the fastest path to monetization.
But tech history teaches us a cruel iron law: giants are forever constrained by the "innovator's dilemma." Their organizational structures, sales channels, and business models are tailor-made for the old world. Once a technological paradigm emerges that requires destroying their core source of profit, giants will instinctively hesitate and compromise. Take the early internet as an example: the traditional telecom companies at the time owned all the fiber and telephone networks and should have dominated the internet, yet the ones that ultimately broke out were Netscape, Amazon, Yahoo, and Google. Because new species never tweak the old business by ten percent; they build from day one with an entirely new business model, and startups carry no baggage—that is the lethal killer move.
Chris Dixon: This is especially true in the Frontier domain. If the future internet is one in which autonomous agents conduct economic settlement and resource scheduling among themselves, the existing centralized business models based on credit card charges and ad-click fees will become completely obsolete. Agents do not need to look at banner ads; they need zero-friction micropayment protocols and on-chain cryptographic credit. This qualitative change in underlying business logic can only be achieved by startups.
5. The Agent Economy Network: When Half the Residents on the Internet Are No Longer Human
Host Erik Torenberg: Following the agent topic Chris just mentioned, in your investment vision you describe a future "agent internet." In this network, how will AIs interact and collaborate with one another?
Chris Dixon: This is a fascinating and imminent evolution. Within the next three to five years, humans may even become a minority among the active interacting entities on the internet, with tens of millions of AI agents executing asset allocation, data retrieval, code deployment, and supply chain bidding on behalf of humans.
But this creates a fundamental engineering and financial challenge: How do agents establish their identity? How do agents complete cross-border settlement without traditional bank accounts? The traditional centralized banking system simply cannot issue debit cards to an autonomous piece of code. This is precisely why crypto protocols and decentralized agents are two sides of the same coin. Agents inherently need to hold their own cryptographic private keys, own their own on-chain wallets, and collaborate and settle through smart contracts in a permissionless network.
Marc Andreessen: Incredible changes will occur even at the social network level. The social platforms of the past two decades were peer-to-peer follows between people, whereas future social networks may be highly hybrid ecosystems. Agents will have their own personalities, knowledge bases, and economic demands; they will debate viewpoints, create content, and even form their own decentralized collaborative organizations in the social square. This will completely reconstruct our understanding of "community" and "media."
6. Escaping the "Product Manager Brain": The Artistic Breakthrough of Games, Film and Television, and Physical Hardware
Host Erik Torenberg: Marc, you recently put forward a very sharp viewpoint, harshly criticizing the "Product Manager Brain" that has prevailed in Silicon Valley in recent years. What negative impacts do you think this mindset has brought to today's technological innovation?
Marc Andreessen: The so-called "Product Manager Brain" is a highly harmful bureaucratic mindset that evolved during the maturity phase of the mobile internet. These people have been rigorously trained in metrics at large tech companies, and they believe the essence of all innovation is looking at retention curves, running multivariate A/B tests, and changing a button's color from dark blue to light blue in exchange for a one-thousandth increase in click-through rate. They reduce business to cold conversion-rate formulas.
In a mature software market where incremental growth has peaked, this kind of refined operations may squeeze out a bit of profit. But in the face of true frontier exploration, the "Product Manager Brain" is not only useless—it can even kill the sprouts of all great innovation. You can imagine that if Steve Jobs had run focus groups and A/B tests while developing the iPhone, he would have ended up with nothing more than a BlackBerry with a smaller keyboard.
Marc Andreessen: True breakthroughs depend on deep taste, artistic intuition, and an engineering obsession that stops at nothing. Especially in the realms of AAA games, film narrative, and entirely new hardware devices, great works are always carved out by creators with powerful individual will and extraordinary taste. The true significance of AI is not to help product managers make a few more data dashboards, but to put the most advanced generative tools into the hands of small teams with the most refined taste. In the future, an independent studio of only ten people will be capable of producing an interactive universe that surpasses the top Hollywood blockbusters in visual fidelity and complexity of worldbuilding.Ben Horowitz: I couldn't agree more. The essence of business has never been the self-recycling of algorithms, but rather touching the deepest emotions and curiosity of humanity. Entrepreneurs must reclaim the will of a founder, dare to place heavy bets on unknown experiences, rather than handing decision-making power over to mechanical data dashboards.
7. Will AI move toward centralization? Why open-source models matter for the future of civilization
Host Erik Torenberg: About a decade ago, renowned Silicon Valley investor Peter Thiel put forward a widely circulated maxim: "Crypto is libertarian, AI is communist." His core logic was that blockchain decentralizes power, while AI naturally tends toward centralized massive compute and data monopolies. Marc, Chris, what do you make of this assertion?
Marc Andreessen: Peter's observation had its sharpness at a particular historical stage, because early deep learning was indeed highly dependent on centralized large data centers and the servers of oligopolistic giants. If you only have closed-source models, if only two or three multinational giants and governments worldwide can possess the smartest machine brains, then AI will indeed evolve into the most terrifying centralization tool in human history.
But that is precisely why we must defend open-source AI to the death! The rise of open-source large models has completely shattered this fate of centralization. From Meta open-sourcing the Llama series, to active global open-source communities, deeply optimized architectures and quantization algorithms, we see that the most advanced intelligence is being compressed at breakneck speed. Today, a model whose performance rivals supercomputers of a few years ago can already run smoothly directly on your laptop, or even on the local chip of your smartphone. Open source means the code is public to all of humanity, the weights do not belong to any bureaucratic censorship body, and anyone can freely fork, fine-tune, and deploy.
Chris Dixon: This aligns perfectly with the spirit of cryptography. Cryptography eliminated the privilege of financial intermediaries, and open-source models eliminate the intermediary of thought and intelligence. In Washington and other regulatory centers, some entrenched giants are cunningly lobbying in the name of "safety," trying to define open-source software as illegal, thereby erecting compliance barriers to preserve their monopoly positions. This is the classic "Baptists and bootleggers" political coalition. We support a16z Frontier precisely to invest unswervingly in the open-source ecosystem, decentralized compute, and open protocols, and never let the future of technology be privatized by a handful of giants.
8. The collapse of the internet's underlying contract: the trust crisis between creators and platforms
Host Erik Torenberg: Chris, in your book Read Write Own, you profoundly discussed the collapse of the internet protocol contract. In the AI era, is this contract rupture further intensifying?
Chris Dixon: The rupture has become completely overt. Early Web2 platforms were built on a fragile tacit agreement: independent creators, bloggers, news media, and community volunteers contributed massive amounts of high-quality content for free, and in return, the platforms gave them distribution traffic, audience building, and opportunities for commercial revenue.
But once the giants completed their monopoly on network effects, the first thing they did was shut the door and beat the dog—continuously tightening algorithms and forcing creators to pay for traffic. And now in the generative AI era, large model companies use crawlers to sweep up decades of human effort across the entire web, then directly provide final answers in search engines and Q&A interfaces, no longer diverting even a single independent visit to the original websites. For independent creators, the original economic cycle has been completely severed.
If this continues, humanity's high-quality original content will dry up, and the internet will be flooded with low-quality garbage generated by machines talking to themselves. The only way to break the deadlock is to build verifiable data ownership through blockchain protocols. Through on-chain confirmation of rights, every data call, inference, and training can be traced and automatically receive economic feedback. Reshaping this contract is one of the most important missions of the Frontier fund.
9. Farewell to smartphone dominance: what will take over the hardware era after the iPhone?
Host Erik Torenberg: Since the release of the first-generation iPhone in 2007, smartphones have dominated human digital life for nearly twenty years. Looking toward the next decade, what kind of physical computing carrier can truly break the dominance of this glass screen?
Ben Horowitz: The smartphone is indeed a perfect industrial miracle, but its interaction form is essentially counterintuitive. Humans must lower their heads, withdraw their attention from the real world, and stare fixedly at that five- or six-inch piece of glass in their palms. This is not the ultimate form of coexistence between technology and humanity.
What we are most bullish on right now is always-on wearable computing devices, especially smart glasses and environment-aware hardware. When you put on a device that weighs no more than ordinary glasses, it has a front-facing high-dynamic camera and directional audio microphones. This means you have a companion AI that can share your first-person perspective and fully understand the physical environment. It knows what you are looking at, it can understand who you are talking to, and it can provide auxiliary information in your ear at just the right moment, without you needing to pull out your phone and swipe to unlock.
Marc Andreessen: This is a fundamental leap from "tool-based computing" to "environment-aware intelligence." Spatial audio, retinal projection displays, and haptic feedback are maturing. Future hardware is not just an information receiver, but an external expansion of human cognitive capabilities. This track is incubating a brand-new hardware supply chain ecosystem on the scale of hundreds of billions of dollars.
10. If you were exactly eighteen today: survival rules for young people from the godfathers of Silicon Valley
Host Erik Torenberg: At the end of this interview, I want to ask a question on behalf of all the curious and bewildered younger generation: if you were exactly eighteen today, about to enter college or just starting out in society, what field would you choose to delve into?
Marc Andreessen: My first piece of advice is: stay away from mediocre, stable trajectories, and absolutely do not study standardized majors that try to train you into a cog. If you study those clerical skills that involve processing documents according to fixed regulations, you will find yourself completely replaced by large models before you even graduate.
Go back to the physical world and first principles. Delve deeply into foundational mathematics, theoretical physics, low-level computer systems engineering, and chemistry and materials. The laws of physics are the hardest truths in the universe; they never lie. At the same time, be extremely hungry to master the most cutting-edge AI tools. Today's eighteen-year-olds possess superpowers unprecedented in history: a young person proficient in modern generative tools and coding agents is alone worth an entire software development department that used to consist of twenty engineers. Go build real things, go launch real products; code and user feedback are worth more than any elite university diploma.
Ben Horowitz: What I want to add is about character and will. No matter how technology changes, the scarcest resource in the business world will always be true leadership and accountability. Go explore those dangerous edges where storms rage, learn to face failure, learn to organize like-minded eccentrics, and learn to hold fast to your intuition when the masses sneer at you. The true wealth of the future will always belong to those builders who roll up their sleeves and sweat in the workshop and the code repository.
Chris Dixon: Keep the geek spirit and remain skeptical of all over-packaged authoritative consensus. Go join the new continents being opened up, even if they look clumsy and rough today. Twenty years ago it was open-source software, ten years ago it was blockchain, and today it is the world of hard tech and agents represented by a16z Frontier. Bravely step onto the frontier; the future has only just begun.