The Generative AI Platform: Understanding Value Accrual and Key Considerations

The Generative AI Platform: Understanding Value Accrual and Key Considerations

Introduction

The field of generative AI has witnessed significant growth, with various companies playing crucial roles in the development and application of AI models. However, understanding where value accrues within the generative AI platform is essential to identify potential winners and long-term drivers of success. In this article, we will explore the different components of the generative AI stack, the challenges faced by various players, and the potential strategies for building sustainable businesses in this dynamic market.

Components of the Generative AI Stack

The generative AI stack primarily comprises infrastructure vendors, model providers, and application companies. Infrastructure vendors, such as cloud platforms and hardware manufacturers, currently capture the majority of revenue flowing through the stack. They run training and inference workloads for generative AI models, making them a crucial part of the ecosystem.

On the other hand, application companies that integrate generative AI models into user-facing products experience rapid revenue growth. However, they often struggle with retention, product differentiation, and gross margins. Model providers, responsible for training generative AI models, are yet to achieve significant commercial scale despite being instrumental in creating the market.

Differentiation and Defensibility in the Stack

Identifying differentiated and defensible components within the generative AI stack is vital for understanding market structure and long-term value drivers. While traditional moats exist for incumbents, finding structural defensibility beyond these moats is challenging.

End-User Applications vs. Vertical Integration

In previous technology cycles, owning the end-customer was considered essential for building a large, independent company. However, in the generative AI space, the assumption that the biggest companies will be end-user applications is not necessarily true. B2B and B2C apps in generative AI often lack technical differentiation and struggle to build strong network effects or unique data/workflows that are hard for competitors to replicate.

Vertical integration, where app developers combine AI models with their products, offers advantages in driving differentiation. Consuming AI models as a service allows for quick iteration and model provider flexibility. However, training models from scratch can provide defensibility through continual re-training on proprietary product data. Yet, it comes with higher capital requirements and a less agile product team.

Building Features vs. Building Apps

Generative AI products take various forms, including desktop and mobile apps, plugins, extensions, and bots. The question arises as to which of these will become standalone companies and which will be absorbed by incumbents already incorporating AI into their product lines. The differentiation and scalability of these features or apps will play a significant role in their future trajectory.

Managing Through the Hype Cycle

While generative AI products generate buzz and usage, it is unclear whether churn is inherent in the current market or a temporary artifact of an early market. The revenue associated with these companies is relatively small compared to their usage. OpenAI, for example, has the potential to become a massive business, particularly if their integration into Microsoft's product portfolio is successful. However, the sustainability of closed-source models and the potential for open-source models to compete strongly remain to be seen.

Actionable Advice:

1. Focus on Vertical Integration: Consider tightly coupling your user-facing app with a home-grown model to differentiate your product and create defensibility.

2. Explore Unique Applications: Look for opportunities to build features or apps that stand out in the generative AI market, keeping in mind the potential for absorption by incumbents.

3. Embrace Hosting Services and APIs: Commercialization in generative AI is often tied to hosting. Explore opportunities to offer proprietary APIs or leverage hosting services for open-source models to facilitate integration and enhance network effects.

The Role of Infrastructure and Cloud Providers

Infrastructure vendors, including cloud platforms and hardware manufacturers, play a crucial role in the generative AI stack. They are currently the biggest winners, capturing a significant portion of revenue flowing through the platform. App companies spend a considerable portion of their revenue on inference and fine-tuning, resulting in a significant share going to cloud providers. Nvidia, with its strong moats built around GPU architecture and a robust software ecosystem, has emerged as a major player in the market.

Commoditization and Graduation Risk

As AI models continue to evolve, there is a belief that AI models' performance will converge over time, potentially leading to commoditization. The risk of customers switching to in-house AI development poses a challenge for app companies relying on model providers. Graduation risk is an important consideration when evaluating long-term partnerships and dependencies on external model providers.

The Potential for Challenger Clouds

While the big three cloud providers (AWS, GCP, and Azure) currently dominate the generative AI market, there is an opportunity for vertical clouds with specialized offerings to gain market share. These niche players can differentiate themselves by offering unique services tailored to specific industry needs or use cases.

Conclusion

The generative AI platform offers immense potential, but understanding value accrual and key considerations is crucial for building sustainable businesses. While infrastructure vendors currently capture the majority of revenue, differentiating and defending components within the stack is challenging. Vertical integration, unique applications, and strategic partnerships can play a significant role in shaping the future of the generative AI market. By staying agile, leveraging hosting services and APIs, and exploring specialized offerings, companies can navigate this dynamic landscape and seize opportunities for growth.

Actionable Advice:

1. Focus on Vertical Integration
2. Explore Unique Applications
3. Embrace Hosting Services and APIs

In summary, the generative AI market presents both challenges and opportunities for various players. By understanding the dynamics of the stack, the importance of differentiation, and the potential for various business models, companies can position themselves for success in this rapidly evolving field.