The Moat Myth
The relentless pursuit of artificial intelligence dominance has sparked what many call the “model wars,” a high-stakes competition where tech giants vie for supremacy. However, according to seasoned tech analyst Benedict Evans, the playing field is surprisingly level. In a recent address at Fortune’s Brainstorm AI conference in London, Evans posited a thought-provoking idea: the primary differentiator between leading AI labs isn’t groundbreaking technology or proprietary algorithms, but rather, their virtually unlimited access to capital.
Evans’ assertion challenges the conventional wisdom that AI innovation is solely driven by intellectual prowess and algorithmic breakthroughs. He argues that foundational models, such as OpenAI’s GPT or Google’s Gemini, are rapidly becoming commoditized. This means that these models are increasingly interchangeable and readily available, diminishing the competitive advantage of any single company.
The concept of an economic “moat,” popularized by Warren Buffett, refers to a company’s sustainable competitive advantages that shield its long-term profits and market share from rivals. In the context of AI, many initially believed that proprietary algorithms, unique datasets, or specialized talent would create such a moat. However, Evans contends that this hasn’t materialized.
After two years of intense competition among Big Tech companies, there still appears to be no fundamental moat in the AI landscape. There are no significant barriers to entry, no strong network effects, and no clear winner-takes-all dynamic. Instead, the primary driver of progress has been a massive influx of capital investment.
Last year, the big four cloud companies collectively spent over $200 billion on building infrastructure to support AI development. This year, that figure is expected to exceed $300 billion. This exponential increase in spending highlights the capital-intensive nature of the current AI race.
“This has become very, very capital intensive, at least at the moment, very, very quickly,” Evans observed. He further noted that a significant portion of this capital is ultimately flowing to Nvidia, the leading manufacturer of GPUs, which are essential for training AI models.
The result of this massive expenditure is a proliferation of AI models, which are becoming increasingly accessible. This, in turn, creates an environment where anyone with substantial financial resources can build a foundational model that rivals those developed by top AI companies.
DeepSeek, for example, is an AI company that leveraged existing open-source models and a $1.6 billion investment to create a competitive AI model. This serves as a compelling illustration of how capital can level the playing field and enable new entrants to challenge established players.
The Commodity Conundrum
Evans argues that AI models like OpenAI’s GPT, Anthropic’s Claude, and Google’s Gemini are evolving into “commodities.” These models are becoming readily available, interchangeable services, akin to undifferentiated, low-cost infrastructure.
This commoditization trend has profound implications for the AI industry. It suggests that the ultimate battleground won’t be about who has the best base model, but rather about who can most effectively package, integrate, and govern that model within real-world products and services.
In other words, the competitive edge may lie not in the foundational model itself, but in the layers of applications and services built on top of it. This shift in focus requires a different set of skills and capabilities, emphasizing product development, user experience, and regulatory compliance.
Evans elaborated on this point in a blog post, using OpenAI’s recent launch of its Deep Research tool as an example. He argued that OpenAI and other foundation model labs lack a true moat or defensibility beyond access to capital. They haven’t achieved product-market fit outside of coding and marketing, and their offerings are essentially limited to text boxes and APIs for other developers to build upon. The key is understanding where the defensibility will arise, and it’s appearing to be less and less in the raw foundation model capabilities.
The Shifting Sands of AI Competition
The commoditization of AI models is reshaping the competitive landscape, forcing companies to re-evaluate their strategies and focus on new areas of differentiation. As the underlying technology becomes more accessible, the emphasis is shifting towards application development, integration, and governance. The AI race is evolving beyond pure model performance into a complex ecosystem of applications and services.
Here are some of the key trends emerging in the AI industry:
Application-Specific AI: Companies are increasingly focusing on developing AI solutions tailored to specific industries or use cases. This approach allows them to create more targeted and effective applications that address specific customer needs. Instead of broad, general models, businesses are looking for AI that directly solves their problems. This trend creates opportunities for niche players and specialized AI providers.
AI-Powered Products: The integration of AI into existing products and services is becoming increasingly common. This can enhance functionality, improve user experience, and create new revenue streams. From smart appliances to AI-powered software, the possibilities are vast. The challenge lies in seamlessly integrating AI without sacrificing user experience or raising ethical concerns.
AI Governance and Ethics: As AI becomes more pervasive, concerns about bias, fairness, and accountability are growing. Companies are starting to invest in AI governance frameworks and ethical guidelines to ensure responsible AI development and deployment. This includes addressing issues like data privacy, algorithmic bias, and the potential for misuse. Robust governance is crucial for building trust in AI and ensuring its responsible use.
Edge AI: The deployment of AI models on edge devices, such as smartphones and IoT sensors, is gaining traction. This enables real-time processing of data without relying on cloud connectivity, reducing latency and improving privacy. Edge AI is particularly valuable in applications where low latency and data security are critical, such as autonomous vehicles and industrial automation.
AI-as-a-Service: The emergence of AI-as-a-Service (AIaaS) platforms is making AI more accessible to businesses of all sizes. These platforms provide pre-trained models, development tools, and infrastructure, allowing companies to quickly and easily integrate AI into their operations. AIaaS lowers the barrier to entry for smaller companies, enabling them to leverage the power of AI without significant upfront investment.
Capital’s Enduring Role
While the commoditization of AI models may diminish the importance of proprietary technology, capital will continue to play a crucial role in the AI industry. Access to funding will be essential for companies to: This is not to say that technical expertise is irrelevant, but rather that it’s becoming a necessary but not sufficient condition for success.
Train and fine-tune AI models: Training large AI models requires significant computational resources and expertise. Companies with access to capital can afford to train larger models on more data, potentially achieving better performance. Even with commoditized base models, fine-tuning them for specific tasks requires significant investment.
Develop and deploy AI applications: Building and deploying AI applications requires investment in software development, infrastructure, and talent. Companies with access to capital can invest in these areas to create compelling AI-powered products and services. The ability to rapidly prototype and iterate on AI applications is crucial for staying ahead of the curve.
Acquire AI talent: The demand for AI talent is high, and skilled AI engineers and researchers command premium salaries. Companies with access to capital can attract and retain top talent, giving them a competitive edge. Building a strong AI team is essential for driving innovation and developing cutting-edge solutions.
Conduct research and development: Continuous innovation is essential in the rapidly evolving AI landscape. Companies with access to capital can invest in research and development to explore new AI techniques and applications. This includes exploring new architectures, algorithms, and data sources.
Navigate regulatory hurdles: As AI becomes more regulated, companies will need to invest in compliance and legal expertise. Companies with access to capital can afford to navigate these regulatory hurdles effectively. Understanding and complying with AI regulations is becoming increasingly important for maintaining a competitive advantage and avoiding legal risks.
The Future of AI Competition
The AI industry is undergoing a period of rapid transformation. The commoditization of AI models is leveling the playing field, but capital will remain a critical determinant of success. Companies that can effectively leverage capital to develop compelling AI applications, attract top talent, and navigate the evolving regulatory landscape will be best positioned to thrive in the long run. The future belongs to those who can translate raw AI power into tangible value.
The future of AI competition will likely be characterized by:
Increased specialization: Companies will focus on developing AI solutions for specific industries or use cases, rather than trying to build general-purpose AI models. This allows for greater focus and efficiency in development and deployment. Vertical AI solutions are likely to outperform horizontal ones in many industries.
Greater emphasis on application development: The focus will shift from building base models to creating compelling AI-powered applications that solve real-world problems. The key is to identify pain points and develop AI solutions that address them effectively. User experience and product design will be critical factors in success.
Growing importance of AI governance: Companies will prioritize ethical and responsible AI development and deployment, ensuring that AI is used for good. This includes implementing robust data privacy policies, addressing algorithmic bias, and ensuring transparency in AI decision-making.
Continued innovation in AI hardware: The demand for more powerful and efficient AI hardware will continue to drive innovation in areas such as GPUs, TPUs, and neuromorphic computing. Specialized hardware will be crucial for supporting the growing demands of AI applications.
Collaboration and open source: Collaboration and open-source initiatives will play an increasingly important role in the AI ecosystem, accelerating innovation and democratizing access to AI technology. Open-source models and tools will enable smaller companies and researchers to participate in the AI revolution.
In conclusion, while access to capital may be the primary differentiator in the current AI landscape, the long-term success of AI companies will depend on their ability to innovate, adapt, and build compelling AI-powered solutions that create value for customers and society as a whole. This requires a holistic approach that encompasses technical expertise, product development skills, ethical considerations, and a deep understanding of customer needs. The AI arms race is far from over, and the rules of engagement are constantly evolving. Companies that can master these challenges will be best positioned to lead the way. The commoditization of foundation models is not the end of AI innovation, but rather a new beginning, one where creativity and business acumen will be as important as technical prowess.