As large-scale model capabilities continue to improve, the AI application layer is facing a common anxiety: if companies like OpenAI and Anthropic not only control the underlying models but also have distribution channels and brand advantages, what can startups do at the application layer?
This is precisely the question a16z partner Joe Schmidt attempts to answer. Drawing on the metaphor of the “Yellow Brick Road” from “The Wizard of Oz,” he divides AI application opportunities into two categories: the main road that large model companies are entering, such as code and image generation, and the “other parts of Oz,” which are vertical scenarios relying on complex workflows, data accumulation, and system integration. In his view, the real opportunity for startups lies in the latter.
From sales to insurance, Joe Schmidt emphasizes that enterprises are not paying for a smarter chat window, but for a system that can be accountable for business outcomes. This involves understanding messy data, managing multi-person approvals, and handling compliance. The underlying models will become stronger and more replaceable, but data, processes, and operational memory solidified around specific industries remain irreplaceable.
Founders often ask if OpenAI and Anthropic will kill everything. While large model labs are indeed entering a vast swath of the application layer, the “application layer” is not a homogenized set of opportunities. The “Yellow Brick Road” describes paths where labs invest significant resources, such as code generation and writing, because these improve as raw model capabilities increase. However, in other parts of Oz, value comes from the scaffolding around the model—making output trustworthy, compliant, and integrated into business processes.
If you are starting a company, the Yellow Brick Road is the most conspicuous but also the most dangerous path. It involves connecting high-performance models to off-the-shelf tools like Slack or Salesforce. The problem is that large model labs are doing exactly this through products like Cowork and Codex, and they possess better margins, control, and distribution power. Startups following this same playbook are likely on a path to nothingness.
For startups, the situation is not entirely bleak. Beyond the yellow brick road, companies are building agent-centric experiences where models are woven into complex software. These companies focus on multi-step, multi-stakeholder workflows and tackle problems that are hard for horizontal platforms to reach. Even as large model labs advance, companies in other parts of Oz have defenses: data and learning flywheels, the ability to manage model volatility through cross-vendor routing, cost optimization by matching specific tasks to appropriate model tiers, and the ability to serve as a governance control plane.
Ultimately, the next generation of enterprise software will be built beyond the yellow brick road. The performance of these companies is judged by their customers’ P&L rather than generic benchmark scores. While both paths will see winners, the underlying models are replaceable, but the working systems built in vertical domains are not.
[BlockBeats]
OpenAI’s Dominance and the Future of AI Applications: Implications for Crypto Investors
The recent analysis from a16z’s Joe Schmidt presents a critical framework for understanding the AI application landscape, which carries significant implications for crypto investors navigating the intersection of artificial intelligence and blockchain technology. Schmidt’s “Yellow Brick Road” metaphor effectively delineates between the commoditized application layer increasingly dominated by large model providers like OpenAI and Anthropic, and the specialized vertical applications where true innovation and value creation will likely emerge.
Market Analysis: The AI Application Layer Dichotomy
The AI market is currently experiencing a bifurcation similar to what we witnessed in the early days of cloud computing. On one hand, general-purpose applications like chat interfaces, code generation, and content creation follow the “Yellow Brick Road” – these are precisely the areas where OpenAI, Anthropic, and similar players have inherent advantages through control of underlying models, distribution channels, and brand recognition.
For crypto investors, this creates a cautionary tale about chasing horizontal applications built solely on top of centralized AI infrastructure. These applications face existential threats as the underlying providers vertically integrate, offering increasingly sophisticated out-of-the-box solutions. The parallels to blockchain are striking: just as we’ve seen numerous DeFi protocols lose moats to improved user experience and integrated solutions, AI applications built merely on top of foundational models without additional value are likely to face similar commoditization.
Token Price Implications
The a16z analysis suggests several nuanced implications for token valuations in the crypto space:
-
Infrastructure Tokens: Tokens enabling specialized compute, particularly those supporting complex multi-agent workflows, may outperform generic AI infrastructure tokens. The need for specialized computing resources in vertical AI applications could create sustained demand.
-
Data DAO Tokens: Projects enabling collective data ownership and governance in specific verticals are positioned to capture significant value. As Schmidt notes, “data, processes, and operational memory solidified around specific industries remain irreplaceable” – tokens representing fractional ownership or governance of such data ecosystems could see substantial appreciation.
-
Agent Framework Tokens: The analysis emphasizes “agent-centric experiences where models are woven into complex software.” Crypto projects enabling the creation, deployment, and governance of autonomous agents with specialized capabilities may emerge as major value capture mechanisms.
Strategic Risks for Crypto Investors
Several risks emerge from this landscape:
-
Centralization Risk: As large AI providers continue to vertically integrate, they may build their own blockchain solutions or acquire promising projects, potentially displacing decentralized alternatives.
-
Overhype Cycles: The AI narrative has already driven significant token price movements, with many projects lacking concrete use cases beyond basic model integration. The coming shakeout could be severe for projects without clear vertical specialization.
-
Regulatory Arbitrage: As AI applications increasingly handle sensitive industry data, particularly in regulated sectors, decentralized solutions may face different regulatory challenges than centralized counterparts.
Investment Opportunities Beyond the Yellow Brick Road
The most promising opportunities for crypto investors lie at the intersection of blockchain and specialized AI applications:
-
Vertical AI Marketplaces: Decentralized platforms enabling industry-specific AI solutions with strong data moats. These could leverage token incentives to collect valuable domain-specific datasets that large horizontal providers cannot easily replicate.
-
Hybrid Oracles: Projects that combine AI predictions with blockchain verification for high-stakes applications in insurance, healthcare, or supply chain management where accountability is paramount.
-
Composable AI Agents: Frameworks allowing developers to create specialized agents that can coordinate across different blockchains and AI systems, enabling complex multi-step workflows that transcend individual platforms.
-
Decentralized Compute Networks: Specialized infrastructure providers that offer not just raw compute but also the specific industry knowledge and optimization required for vertical AI applications.
The a16z analysis correctly identifies that enterprise customers “are not paying for a smarter chat window, but for a system that can be accountable for business outcomes.” In crypto terms, this suggests that projects enabling auditable, verifiable AI systems with clear business outcomes – particularly those leveraging blockchain’s inherent transparency and composability – will likely outperform purely speculative AI tokens.
Conclusion
The AI application layer is indeed facing consolidation at the horizontal level, but this creates opportunities for vertical specialization that crypto projects are uniquely positioned to capture. The most promising investments will likely combine specialized domain knowledge with the unique advantages of blockchain technology – particularly in areas requiring trust, transparency, and multi-stakeholder coordination. As Schmidt notes, “the underlying models are replaceable, but the working systems built in vertical domains are not.” For crypto investors, the lesson is clear: focus on the “other parts of Oz” where blockchain can create defensible value beyond what even the largest AI providers can replicate.