a16z Expands Crypto Fund to 'Frontier' Mandate, Eyes Ambitious AI Ventures

a16z is expanding its crypto fund to a16z Frontier, a new mandate to support founders exploring the limits of what's technologically possible. Marc Andreessen highlighted a significant disparity between early adopters' understanding of potential and current development, describing this gap as the widest he has witnessed in his career. The firm anticipates a substantial increase in startup formation as a result.

The shift also reflects a change in investment philosophy. Previously, the question was often, "is this idea too ambitious?" Now, the firm is more inclined to ask, "is this ambitious enough?" This change is driven by the spectacular nature of current technological capabilities. The blurring lines between various tech sectors, such as fintech, Wall Street, and crypto, underscore the need for a broader approach, as Chris Dixon noted, citing stablecoins' integration into mainstream tech and Wall Street firms' heavy move into tokenized assets.

Ben Horowitz explained that while their vertical investment strategy has been highly successful, new, novel areas often emerge that fall outside existing verticals, making the process of integrating them into a defined practice somewhat unclear. Historically, Chris Dixon would personally explore these new areas before a vertical was established. The new Frontier mandate formalizes this exploration.

This dedicated motion for Frontier allows the firm to invest in "outer planet" entrepreneurs—those pushing technological boundaries—even before a clear category or product emerges. Horowitz stated that investing in entrepreneurs is more crucial than focusing on specific categories or products, and many of the most compelling entrepreneurs operate in these uncharted territories. The Frontier initiative is designed to fill this architectural gap, potentially leading to the formation of entirely new funds if an area proves sufficiently interesting.

Investing in crypto requires a profound understanding of complex technology, combined with astute business and market judgment. Marc Andreessen highlighted the exceptional expertise within Chris Dixon's team, which includes math professors, cryptographers, and game theorists, emphasizing their deep theoretical and software knowledge. This team, now with a broader mandate, will apply its specialized skills to a wider array of topics within Frontier.

Beyond technical acumen, the firm has also made significant investments in policy and regulatory expertise over the years. Andreessen noted that many cutting-edge technologies inevitably develop a "policy angle." He cited self-driving cars, drones, and AI itself as examples of technologies that are increasingly becoming political issues, underscoring the importance of this policy depth in navigating the complex landscape of frontier innovation.

Ben Horowitz revealed that a core principle of a16z's culture, documented within the firm, draws inspiration from JP Morgan's motto: "We do first class business and then only in a first class way." This philosophy, which emerged during JP Morgan's trial when he asserted his integrity, resonates with Horowitz and Andreessen's belief in maintaining principled and trustworthy conduct in venture capital.

Chris Dixon added that the firm fosters a highly entrepreneurial environment where ambitious ideas are not only encouraged but actively pushed. He noted that the founders, Marc and Ben, are incredibly ambitious themselves, consistently challenging ideas to be more aggressive and impactful, rather than pushing back on bold proposals. This culture of innovation and expansion is vital for adapting to global changes and pioneering new ventures.

Chris Dixon stated that the Frontier Fund's strategy prioritizes being "too early" in investments, acknowledging that this involves accepting category risk. He explained that success in technology is a complex interplay of timing, culture, and business/political environments.

Dixon cited Oculus as an example: despite its early sale, he remains a strong believer in VR's long-term potential 13 years later, even though it hasn't yet achieved mainstream adoption. He believes AI's rapid advancements, particularly in software engineering, could accelerate technological progress, potentially pulling forward two decades of development. This rapid evolution challenges traditional views of technical risk, suggesting that what was once considered a "science project" might now be an "engineering project."

He also suggested that a 30-person team today could credibly build a full-stack device company, a feat that might have required a thousand people a decade ago, highlighting the increasing leverage available to smaller teams.

Marc Andreessen explained that a16z's investment evaluation criteria have fundamentally shifted. In the past, the firm often asked, "is this idea too ambitious?" Now, the prevailing question is, "is this ambitious enough?" This change stems from the "spectacular" possibilities enabled by current technological advancements, particularly AI.

Andreessen noted that moderate ideas are increasingly seen as lacking sufficient differentiation in today's fast-paced environment. The expectation for entrepreneurs is to build complex, multi-faceted businesses that transcend traditional classifications, integrating various domains like software and hardware. He warned that incumbent companies, if not adapting to the rapid pace of development, are on borrowed time, as the speed of change makes it nearly impossible to keep up.

Marc Andreessen highlighted a peculiar trend in AI adoption: many companies are not formally implementing AI, while their employees are silently integrating it into their daily workflows without management's full awareness. He suggested that individuals are finding it significantly easier to adopt AI tools than institutions.

Employees are reportedly using AI for tasks such as fact-checking, proofreading, and drafting emails. Andreessen posited that this individual adoption will fuel a significant surge in startup formation, as these empowered individuals realize the ease of starting their own businesses with AI tools. He expressed conviction that this will lead to a wave of technically and business-aggressive, expansionist, and creatively original ventures, often combining previously separate concepts.

Chris Dixon described the typical technology adoption cycle, starting with replicating old behaviors in new ways, similar to how Sears catalogs moved online in the 1990s. He noted that true breakthroughs occur when new cultural or behavioral changes emerge, citing YouTube influencers in the 2000s as an example of entirely new activities enabled by a technology.

Currently, AI applications primarily mirror existing human work, such as coding, mathematics, and knowledge work. However, Dixon believes we are at a "tipping point" where the exploration of "radical things"—new, native AI behaviors and activities—will begin.

He anticipates the emergence of "crazier, weirder stuff" that people have never done before, marking a shift from merely improving old tasks to creating fundamentally new ones. This phase of development promises to unlock transformative, world-changing innovations.

Marc Andreessen critiqued the current state of many AI agents, likening them to "happy, harmless, honest assistant[s]" that are ultimately boring and prone to lecturing users. He suggested that early AI companies have a "blinker view" on human activity, designing products for a limited set of use cases, akin to "product manager brain"—where products are designed for other product managers.

Andreessen asserted that this restrictive approach limits AI's true potential. He emphasized that the technology is now evolving beyond these limitations, moving towards more entertaining, fun, enjoyable, exciting, and, crucially, lucrative uses. This includes enabling new forms of creation, such as music videos made de novo, which have seen an explosion of activity in recent weeks, allowing people to participate more broadly in the economy and create new art forms.

Marc Andreessen highlighted AI's transformative potential in the entertainment industry, particularly in drastically reducing production costs. He pointed out that while artistic films with high budgets often result in hundreds of millions in losses for studios, AI could slash a $200 million production budget to as little as $2 million. This cost reduction would turn artistic failures into highly profitable ventures, making such films "insanely profitable" and generating tenfold returns for investors.

He also extended this impact to hardware, noting that for years, venturing into hardware has been notoriously difficult and prone to company failures. However, new AI-powered tools like "REA" (Reverse Engineer Anything), which harnesses an AI model to reverse-engineer any hardware and provide its source code, are changing this landscape. This means that if a household thermostat isn't working correctly, AI could write new code for it, democratizing hardware development and customization.

Ben Horowitz challenged Peter Thiel's decade-old quote that labeled crypto as decentralizing and AI as centralizing. Horowitz argued that AI's ability to provide immense leverage and power to small entities—enabling one-person or 30-person companies to achieve orders of magnitude more than before—suggests it may foster decentralization rather than centralization.

Chris Dixon elaborated, predicting a diverse AI market structure. While large entities like OpenAI and Anthropic will thrive, he emphasized the significant rise of open-source AI models, which are growing faster in token consumption than closed models. He also pointed to the emerging trend of local, edge-based models that run on individual computers, enabling top-end coding agents or chatbots directly on personal devices. This decentralized trend, he believes, offers substantial opportunities for a more distributed market structure.

Dixon also noted that crypto could play an important role in this decentralized AI economy, especially through stablecoins, which would facilitate payments between AI agents within such a system. The firm views open-source development as a crucial and highly effective approach in the current AI landscape.

Marc Andreessen argued strongly against banning or over-regulating open-source AI, emphasizing its critical role in education. He posed a rhetorical question: how would parents feel if their children could only learn about AI by working for one of two dominant companies? He believes that open-source access ensures every child has the birthright to understand and utilize this technology fully, allowing them to grasp how it works, address problems, and even defend themselves against vulnerabilities, much like how open-source software provides advanced vulnerability scanners.

Andreessen also highlighted an "underestimated thing" about AI itself: open-source AI models can teach users how to create open-source AI. While some proprietary AI models might refuse to provide such explanations, open-source versions fully support this educational function. This democratization of AI knowledge, he suggested, empowers individuals with the most advanced tools, effectively putting "a doctor and a genius professor in their pocket."

Chris Dixon described the internet's traditional "covenant" as a deal between distribution platforms (like Google and Facebook) and content creators (news sites, travel sites). Under this agreement, content providers would offer snippets for indexing, and platforms would send traffic (clicks) in return. Dixon stated this covenant began to fracture in the last decade and is now severely breaking down.

He explained that with AI, users increasingly get direct answers from search engines, rather than being redirected to websites. Social networks also aim to keep users on their platforms, further reducing outbound links. This has significant downstream effects, as websites reliant on ad revenue and clicks face a loss of traffic and economic sustainability, evidenced by dramatic traffic drops for sites like Stack Overflow after AI replacements emerged.

While transactions will still occur, they will happen differently. Dixon acknowledged the short-term pain for businesses built on the "human attention economy." However, he sees this as an opportunity for entrepreneurs to "rethink these things from first principles" and imagine a better internet, shifting the economic landscape and knowledge sharing paradigms over the next decade.

Chris Dixon sees AI, coupled with voice technology, as the key to unlocking the potential of new device form factors, particularly wearables. He noted that wearables, which were previously limited in functionality, are now becoming much more viable. While VR still faces ergonomic challenges like weight and user comfort, Dixon firmly believes that humans inherently desire deeply immersive, otherworldly experiences, a concept long explored in science fiction.

AI is expected to significantly improve the viability of VR platforms by enabling the creation of amazing content. Dixon cited an example of an individual porting 'Titanfall' into VR by himself, demonstrating AI's power to streamline content generation. This content creation capability will likely trigger a resurgence in interest for older devices and drive the development of new immersive technologies. He also mentioned more futuristic concepts like Neuralink and brain-computer interfaces, along with the continued growth of drones and self-driving cars, as areas where AI will drive widespread adoption and innovation.

Ben Horowitz, echoing Marc Andreessen, advised the next generation to "study modern alchemy" – specifically, the ability to "turn sand into intelligence." He suggested that key areas of study for this include math, computer science, and possibly economics.

Chris Dixon added that AI is poised to revolutionize a vast array of fields, from classics (such as deciphering ancient scrolls) and economics to biology and art, and even potentially understanding animal communication. He emphasized that the defining narrative of the next 20 to 30 years will be AI, and while not everyone needs to be solely technical, being involved in some way "at the cutting edge" and utilizing these new tools is the obvious path forward for careers.

Chris Dixon referenced Douglas Adams' framework on how different age cohorts perceive new technology. For those under 15, new technology is simply "the natural way the world works." Dixon illustrated this with his seven-year-old, who responded with "duh" when introduced to ChatGPT, seeing a computer that answers questions as normal.

For individuals between 15 and 35, new technology is "new, exciting, and hot," presenting significant career opportunities. However, for those over 35, it is often viewed as "unholy and evil," threatening to overturn existing structures and destroy society. Dixon noted that kids are already adopting AI as a "secret superpower" despite institutional resistance, reinforcing that 18-year-olds are in a prime position to leverage these new capabilities, leading to entirely new fields of entrepreneurship and creativity.