In this episode, Anand Iyer, a venture capitalist focused on the AI and crypto intersection, breaks down why blockchain is not just a sideshow but a critical solution to AI’s biggest challenges, from the compute arms race to the looming threat of centralized control.
The Compute Arms Race & The Niche Model Counter-Attack
- “What's coming down the pike... are more niche, more custom, smaller language models... that don't have the same kind of compute requirements because they're not trying to boil the ocean.”
The AI industry is locked in a compute "arms race." Hyperscalers like OpenAI and Google are amassing enormous power to train foundational models, with data centers projected to consume over half of the new power demand by 2026. However, this centralized approach is creating a strategic opening for a different kind of AI. The future isn't just bigger models; it's also a wave of smaller, specialized models (SLMs) trained for specific domains. These SLMs require far less compute, making them perfect candidates for decentralized networks built on distributed, heterogeneous hardware—a core strength of blockchain infrastructure.
The Ownership Dilemma: Decentralizing AI Control
- “Imagine for a second… you have that quadrant of hallucination gone awry and there's some mal-intent behind it. Now suddenly… teenagers who are using this for relationship advice are getting influenced by a large corporation.”
As users form increasingly intimate bonds with AI—using it as a therapist or relationship coach—the question of who controls these systems becomes critical. The concentration of AI in the hands of a few "technocrats" poses a dystopian risk of mass influence and manipulation. Blockchain offers a powerful antidote: a path toward "user-owned AI." By leveraging crypto’s core principles of privacy, sovereignty, and verifiability, projects are building an alternative stack where users can have a direct stake and say in the AI they rely on, preventing a future where our digital confidants are puppets for corporate interests.
The On-Chain IPO: A New Playbook for AI Startups
- “We are entering a regime where the product is the raise and the market is the funder… The next billion-dollar project might be 'vibe-coded' by a single dev or two… and sustained by fees.”
The traditional venture capital path for AI is becoming a bottleneck, demanding trillion-dollar outcomes and leaving many innovative, profitable companies in a "no-man's land." Tokenization is emerging as a powerful new playbook. By launching a token, AI startups can achieve liquidity and raise capital directly from their communities, bypassing the rigid VC cycle. This meets a clear and growing retail demand for exposure to AI, a sector that has dramatically outperformed the broader crypto market over the past year. This is not just a theory; as Bill Gurley noted, crypto is becoming one of the only viable paths to liquidity in venture today.
Key Takeaways:
- The collision of AI and crypto is creating the third major arc in blockchain's history, moving beyond store-of-value and DeFi to drive real-world utility. This convergence offers a critical alternative to the centralized, capital-intensive path of traditional AI, focusing instead on decentralization, user ownership, and new models for capital formation.
- AI's Power Problem is Crypto's Opportunity: The insatiable energy demand of large, centralized AI models creates a strategic opening for more efficient, specialized AIs built on decentralized compute networks.
- Decentralize or Be Manipulated: The fight is on to prevent a handful of corporations from controlling the "super-intelligent beings" we interact with daily. User-owned AI built on blockchain primitives is the primary defense.
- The AI Tokenization Wave is Coming: Profitable AI startups that don't fit the traditional VC mold will increasingly turn to "on-chain IPOs," creating a new, high-demand asset class that offers investors direct exposure to AI's growth.
For further insights and detailed discussions, watch the full episode: Link
This episode reveals how the insatiable demand for AI compute is creating a new investment frontier where crypto’s core principles of decentralization and ownership offer a powerful alternative to the centralized, capital-intensive race dominated by tech giants.
The Core Thesis: Infrastructure in a Post-AGI World
- Anand Iyer, a venture capitalist with deep roots in both DeFi and AI, frames his investment thesis around a single, forward-looking question: what does infrastructure look like in a post-AGI world? He argues that the current trajectory toward Artificial General Intelligence (AGI)—AI systems with human-like cognitive abilities—demands a fundamental rethinking of the systems we build today.
- Anand’s approach is to "live in the future" and work backward, anticipating the second and third-order effects of AI's exponential growth.
- He sees the collision of AI and crypto not as a niche but as the third major arc of blockchain technology, following store of value (Bitcoin) and decentralized finance (DeFi).
- This new arc, he believes, will be defined by AI driving real-world utility on-chain.
The Inevitable Acceleration of AI
- The conversation touches on the rapid, almost disorienting pace of AI development, referencing Sam Altman's observation that the world isn't as "weird" as he expected, despite reaching a form of superintelligence. Anand agrees, noting that while the growth is parabolic, societal adoption takes time.
- He uses the gradual public acceptance of autonomous vehicles like Waymo as a mental model for how society will slowly internalize profound technological shifts.
- Anand highlights the personal and intimate ways AI is already being adopted, such as people using ChatGPT as a personal therapist, which underscores the technology's deep integration into daily life.
- "The only thing that's a variable here is what is that amount of time that it'll take for us to realize that change? Is it going to be months, days, years? I don't know what that is, but it will take time."
The Compute Arms Race and the Rise of Specialized Models
- The Data Center Boom: Data centers are now projected to consume three times the power of hospitals, previously the largest users on the grid. This is driven by the need to train and run massive AI models.
- The Limits of Large Models: Anand points to a leaked Google memo suggesting that large models have "no moat." While companies like OpenAI, Anthropic, and Google are amassing vast compute resources to build broad, "boil the ocean" foundation models, he questions the long-term viability of this approach.
- Strategic Insight: Anand argues the next wave of innovation will come from smaller, more niche, and custom models (SLMs, or Small Language Models). These models are hyper-focused on specific domains, require less compute, and are better suited for decentralized infrastructure.
- Projects like Hyperbolic are cited as examples, building infrastructure on distributed, non-homogeneous machines, showcasing the power of decentralized systems.
Second and Third-Order Effects of Compute Demand
- Energy Consumption: Power consumption is forecast to double between 2023 and 2026, with over 50% of that increase driven by AI needs. This is forcing radical solutions, such as Meta exploring co-locating nuclear reactors with its data centers.
- The Venture Lens: Anand contrasts the two investment landscapes:
- Centralized AI: Has attracted approximately $300 billion in investment since 2022, creating a market now valued at roughly $1.5 trillion. This is a high-stakes game where success is measured in trillion-dollar outcomes.
- Decentralized AI: Has received only about $1 billion in the same period. Anand sees this as a contrarian opportunity to "fish where no one else is fishing."
- Actionable Insight: For investors, the massive capital flow into centralized AI makes it a crowded space. The real asymmetric opportunity may lie in the underfunded but philosophically critical decentralized AI sector, which is just beginning its growth curve.
Why Blockchain is AI's Missing Piece
- User-Owned AI: The ultimate goal is to create a future where individuals can own, control, and even stake their share in AGI. This starts with building on core tenets like privacy, sovereignty, and verifiability.
- The Dystopian Risk: The intimate use of AI (e.g., teenagers seeking relationship advice from ChatGPT) highlights the danger of centralized control. Malicious intent or even subtle biases embedded in these models could have widespread societal influence, a problem decentralized systems are designed to mitigate.
- "If we want to work towards this concept of ownership of AI, then that needs to start now, and it starts with all the tenets that you just mentioned: privacy preservation and sovereignty."
The Challenges and Path Forward for Crypto AI
- The Resource Gap: The 300:1 funding disparity between centralized and decentralized AI means projects are operating with fewer resources and the "luxury of time."
- Technical Hurdles: While decentralized training was once considered impossible, projects like Jensen have proven it can be done. The next challenge is scaling these systems and moving beyond transformer-based models, potentially with "blockchain-first principles."
- The Call to Action: The space needs more top-tier talent (like the PhD founders of Jensen and Hyperbolic) and more capital to close the gap and build a viable alternative to centralized AI.
The On-Chain IPO: A New Liquidity Path for AI Startups
- The Venture Squeeze: The traditional VC model is becoming increasingly difficult for companies that aren't targeting "decacorn" or trillion-dollar outcomes. Many profitable, niche AI companies may not fit this mold.
- Tokenization as a Solution: Crypto offers a "beautiful capital coordination tool." By launching a token, these companies can achieve liquidity, reward their community, and fund their runway without being beholden to the rigid expectations of venture capital. This aligns with Bill Gurley's observation that crypto is one of the only viable paths to liquidity in venture today.
- Retail Demand: There is immense, untapped retail demand for AI investments. A Charles Schwab survey revealed that 47% of millennials want to invest in AI, but their options are limited to stocks like NVIDIA. Tokenized AI companies could directly meet this demand.
The Future: Interoperable Capital Markets
- The Product is the Raise: In this new paradigm, a project with a working product and fee-based traction (MVP + fees) is more valuable than a pitch deck. Agent fees can fund the runway, and token liquidity can replace traditional funding rounds like a Series B.
- The Billion-Dollar Solo Dev: Anand predicts we will see a billion-dollar project coded by a single developer, launched on a platform like Olas (formerly Autonolas), and sustained by on-chain fees and community incentives.
- This vision culminates in a world of "interoperable capital markets" where value creation and capture are fluid, decentralized, and accessible to all.
Conclusion
This episode highlights a critical divergence: while centralized AI chases massive scale with immense capital, a new frontier is opening for decentralized AI. For investors and researchers, the key takeaway is that crypto offers not just a technological solution for AI's trust problem but also a revolutionary financial model for its future development.
