Will AI Be Bigger Than The Internet?

This episode dives into the hidden economics of GPU scarcity—how AI and crypto are colliding over compute power, and what this means for investors.

AI's Evolving Definition and the AGI Conundrum

• The discussion opens by highlighting the paradox of AI adoption: while ChatGPT boasts 800-900 million weekly active users, a significant portion of account holders don't engage with it regularly. Benedict argues that the term "AI" itself is fluid, often referring to "new stuff," while "AGI" (Artificial General Intelligence)—the hypothetical intelligence capable of understanding, learning, and applying knowledge across a wide range of tasks at a human level—is perceived as "new scary stuff." He notes the ongoing debate among leaders like Sam Altman and Demis Hassabis regarding whether AGI is already here in nascent form or perpetually five years away, underscoring the lack of a clear understanding of the technology's fundamental limits.
• Strategic Implication for Crypto AI: The ambiguity around AGI's definition and timeline creates both speculative opportunities and risks. Investors should monitor research breakthroughs and practical applications that demonstrate genuine multi-domain problem-solving, rather than just improved task-specific performance, to gauge true AGI progress.

"AI Eats the World": A Platform Shift Perspective

• Benedict introduces his "AI Eats the World" thesis, framing AI as a platform shift comparable to the internet or smartphones. He emphasizes that such shifts typically lead to market bubbles, significant internal changes within the tech industry (creating new trillion-dollar companies while rendering others irrelevant), and profound impacts on industries outside of tech. He uses the analogy of automatic elevators, which were once "electronic politeness" but are now simply "lifts," to illustrate how new technologies become normalized and lose their "AI" label over time.
• Strategic Implication for Crypto AI: This perspective suggests that while a bubble is likely, the long-term value creation will be immense. Crypto AI projects that can integrate AI as a fundamental, invisible layer—much like databases or the internet—rather than a standalone "AI feature," are better positioned for enduring success.

AI vs. Previous Platform Shifts: Disruptive vs. Sustaining Innovation

• The conversation explores whether AI will be a "disruptive" force, creating entirely new companies like the internet did (e.g., Google, Facebook), or a "sustaining" one, primarily benefiting existing incumbents like mobile did (e.g., Uber, Snap, but ultimately strengthening Facebook and Google). Benedict cautions against rigid frameworks, noting that mobile fundamentally shifted behavior from web to apps and put a "pocket computer" in everyone's hand, enabling new phenomena like TikTok. He stresses that the ultimate impact of AI—whether it's "just another platform shift" or something more akin to "computing" itself—remains unknown.
• Strategic Implication for Crypto AI: Crypto AI investors should analyze whether projects are truly enabling net-new behaviors and business models (disruptive) or merely enhancing existing crypto infrastructure or applications (sustaining). The potential for AI to be "as big as computing" suggests a foundational shift that could redefine decentralized networks and applications.

The "Physical Limits" of AI and Forecasting Challenges

• A critical point raised is the lack of understanding regarding the "physical limits" of AI technology. Unlike previous tech cycles where roadmaps for modems or DSL speeds were predictable, there's no equivalent model for AI's fundamental capabilities in the next few years. This theoretical gap, coupled with an incomplete understanding of human intelligence, leads to "vibes-based forecasting" among even leading experts. Benedict highlights the "schizophrenia" in the AI discourse, where some predict "PhD-level AI researchers" while others focus on API stacks for software developers, suggesting these two outcomes might be mutually exclusive in the long run.
• Strategic Implication for Crypto AI: The unpredictable scaling and capability of AI models mean that investment in foundational research and flexible, adaptable infrastructure (like decentralized compute networks) is crucial. Crypto AI researchers should focus on developing robust theoretical frameworks for AI capabilities and limitations, which could provide a competitive edge in a highly uncertain landscape.

AI Investment, Bubbles, and Compute Requirements

• Benedict asserts that "very new, very big, very exciting world-changing things tend to lead to bubbles," and current AI investment exhibits "bubbly behavior." He compares forecasting AI's compute requirements to predicting internet bandwidth usage in the late 1990s—a task fraught with uncertainty. Hyperscalers are currently operating under the principle that "the downside of not investing is bigger than the downside of overinvesting," leading to massive capital expenditure. However, he questions the sustainability of this, especially if future models achieve similar results with significantly less compute, rendering current overinvestment redundant across the industry.
• Strategic Implication for Crypto AI: The massive capital expenditure by hyperscalers creates opportunities for decentralized compute networks (e.g., those leveraging idle GPUs) to offer more cost-effective and flexible alternatives. Crypto AI investors should be wary of overvalued projects built on assumptions of ever-increasing compute demand and instead look for those that can adapt to potential efficiency gains or offer unique value propositions beyond raw compute.

AI Deployment: Obvious vs. Unclear Use Cases and the "Productization" Challenge

• The discussion bifurcates AI deployment into two categories:
• Obvious Use Cases: Software development, marketing, specific enterprise solutions, and flexible jobs (e.g., Silicon Valley professionals optimizing their workflows).
• Unclear Use Cases: The vast majority of users who have ChatGPT accounts but don't use it weekly. Benedict questions why five times more people know how to use it but can't find a daily application. He uses the analogy of Excel for accountants (transformative) versus lawyers (less immediately applicable) to illustrate that AI needs to be "wrapped in a product and a workflow" to achieve widespread adoption beyond niche power users.
• Strategic Implication for Crypto AI: The challenge lies in "productizing" AI for the masses. Crypto AI projects need to move beyond raw model access to build intuitive, workflow-integrated applications that solve specific, tangible problems for a broad user base. Projects focusing on clear, validated use cases with strong UX/UI will likely outperform those offering general-purpose AI tools.

The Future of AI: New Behaviors and the "Stack" Question

• Benedict emphasizes that new platforms often enable entirely new behaviors and industries, citing mobile's role in ride-sharing (Uber/Lyft) and online dating. He questions how far "up the stack" AI models will go, comparing it to the debate over whether Windows apps were just "thin Win32 wrappers" or truly distinct products. He argues that people buy "solutions, not technologies," meaning dedicated UIs and curated workflows will remain crucial. The raw chatbot, which "asks you literally everything," contrasts sharply with well-designed software that guides users through specific choices based on institutional knowledge.
• Strategic Implication for Crypto AI: Crypto AI projects should focus on identifying and enabling "net new behaviors" that were previously impossible. This involves building specialized applications that abstract away the underlying AI complexity, offering clear solutions rather than just access to models. The "stack" question implies that while foundational models are critical, the real value will be captured by companies building user-centric applications on top.

Competitive Landscape: Hyperscalers and Strategic Questions

• The conversation delves into the strategic positions of hyperscalers. Google, with its existing cash flow, can invest heavily, optimizing search and ads, and potentially creating or copying an "iPhone of AI." Meta faces bigger questions regarding AI's impact on content and social, making proprietary models imperative. Amazon could leverage AI for better recommendations and discovery, moving beyond its "pure commodity retailing model." Apple, despite its compelling Siri demo, struggles with real-world implementation, raising the question of whether AI is a fundamental computing shift (problematic for Apple) or just another service (less so). Benedict notes Microsoft's historical loss of the platform war (dev environment) but gain in device sales (Windows PCs), suggesting complex outcomes.
• Strategic Implication for Crypto AI: The competitive dynamics among hyperscalers highlight the importance of owning infrastructure and distribution. Crypto AI projects building decentralized alternatives to hyperscaler services (e.g., decentralized compute, storage, or model hosting) could carve out significant niches. Furthermore, projects that can integrate AI into existing, widely adopted crypto platforms (like DeFi or NFTs) could gain a strategic advantage.

Evolving Questions and the "Job to Be Done"

• Benedict reflects on how the key questions in AI have evolved. While early 2023 focused on open source, China, Nvidia, and scaling, new "product strategy questions" are emerging due to consumer adoption. He outlines a three-step evolution for industries:
• Step One: Absorb AI as a feature, doing obvious automation.
• Step Two: Build new things with AI.
• Step Three: Redefine the entire industry.
• He emphasizes the need to understand the "true job to be done" for users, citing the newspaper industry's failure to recognize itself as a "light manufacturing and trucking company" before the internet disrupted it. AI will similarly expose hidden vulnerabilities and opportunities in various sectors.
• Strategic Implication for Crypto AI: Crypto AI projects must move beyond simple automation (Step One) to create novel applications (Step Two) and ultimately redefine how decentralized systems operate (Step Three). Investors should seek projects that deeply understand the "job to be done" within specific crypto verticals and leverage AI to fundamentally transform those processes, rather than just incrementally improving them.

Defining "Bigger Than the Internet" and AGI

• Concluding the discussion, Benedict reiterates the immense scale of previous platform shifts like the iPhone and the internet, which are often underestimated. For AI to be "bigger than the internet," it would require a profound shift in our perception of its capabilities—moving beyond "people-like things really well sometimes" to genuinely replacing an "actual person" in complex, unconstrained tasks. He acknowledges the difficulty in providing a falsifiable answer to when AGI would truly arrive, echoing the sentiment that "AI is whatever doesn't work yet."
• Strategic Implication for Crypto AI: The ultimate impact of AI hinges on its ability to achieve AGI-like capabilities. Crypto AI investors and researchers should closely monitor advancements in model reasoning, generalization, and autonomous agency. Projects that contribute to or leverage these breakthroughs in a decentralized, verifiable manner will be at the forefront of a potentially transformative era, but the path remains highly uncertain and speculative.

The conversation highlights AI's profound, yet uncertain, platform shift potential, demanding strategic foresight from Crypto AI investors and researchers. Success hinges on identifying projects that move beyond basic automation to create novel, productized solutions and redefine industry paradigms, rather than merely sustaining incumbents.