According to TechCrunch, venture investor Elad Gil stated at TechCrunch Disrupt that AI has been one of the least predictable tech booms he’s ever seen. Gil, who invested in generative AI as early as 2021 after observing the massive capability leap between GPT-2 and GPT-3, now sees certain markets nearly sewn up by leaders while others remain wide open. He identified foundational models, AI-assisted coding, medical transcription, and customer support as markets with established leaders, pointing to companies like OpenAI, Anthropic, Abridge, and his portfolio company Decagon, which raised $131 million at a $1.5 billion valuation in June. Meanwhile, he sees financial tooling, accounting, AI security, and other enterprise-focused markets as still being anyone’s game, noting that rapid enterprise adoption doesn’t guarantee long-term success.
The Unusual Consolidation Patterns of AI Markets
What makes Gil’s observations particularly insightful is how quickly certain AI markets are consolidating compared to previous technology waves. Unlike social media or mobile apps, where multiple winners could coexist for years, AI markets appear to be winner-take-much-faster due to the massive computational requirements and data advantages. The foundational model space demonstrates this perfectly – while hundreds of models exist, only a handful with extraordinary resources can compete at the cutting edge. This creates a structural barrier that previous technology revolutions didn’t face to the same degree. The capital intensity of training frontier models means new entrants need billions, not millions, to compete effectively.
The Enterprise Adoption Double-Edged Sword
Gil’s observation about enterprises rapidly experimenting with AI solutions reveals a critical dynamic that many startups misunderstand. When every major company has an AI mandate, early revenue becomes a deceptive metric. Enterprises are willing to pilot multiple solutions simultaneously, creating what Gil calls “false signals” of product-market fit. The real test comes when these companies standardize on fewer solutions and commit to long-term contracts. This explains why companies like Harvey, despite rapid valuation increases, might represent genuine market leadership – they’ve moved beyond experimentation to becoming essential infrastructure for their clients.
Geographic and Regulatory Implications
The mention of sovereign AI initiatives in countries like South Korea highlights an emerging fragmentation that could create regional opportunities even in seemingly consolidated markets. While Gil identifies foundational models as having clear leaders, national security concerns and data sovereignty regulations are creating parallel markets where local champions can thrive. This represents a significant departure from previous technology waves where American companies dominated globally. The regulatory environment around AI is developing differently across regions, potentially creating protected markets where new entrants can establish leadership positions despite global consolidation.
The Surprising Incumbent Advantage
What makes AI particularly challenging for startups is how effectively traditional software incumbents can integrate AI capabilities into existing products. Gil mentions Salesforce and HubSpot adding AI to their customer support offerings – this represents a formidable barrier for pure-play AI startups. Unlike previous technology shifts where incumbents struggled to adapt, many established software companies have the data, customer relationships, and distribution to rapidly incorporate AI features. This means that in some markets, the winners might not be AI-native startups but rather traditional software companies that successfully augment their offerings with intelligence.
The Developer Experience Revolution
The consolidation in AI-assisted coding tools represents a fundamental shift in how software gets built. When Gil mentions companies like Cursor and Devin as potential winners, he’s pointing to tools that are changing the very nature of programming. Unlike previous developer tools that augmented human capabilities, these AI coding assistants are beginning to automate significant portions of the development process. This creates incredibly sticky products – once development teams standardize on a particular AI coding workflow, switching costs become enormous due to training time, integration complexity, and workflow dependencies. The massive funding rounds in this space reflect investor recognition that these could become the new foundational platforms for software development.
Long-Term Market Structure Implications
The most significant insight from Gil’s analysis is that AI markets may settle into a layered structure similar to previous technology platforms. At the bottom are the foundational model providers – a small group of well-capitalized companies. Above them are application companies that build on these models, some of which will achieve durable leadership in specific verticals. What makes this different from previous platform shifts is the speed at which these layers are forming and the extraordinary capital requirements at each level. For investors and entrepreneurs, the key is identifying which layer offers the right balance of opportunity versus competition based on their specific resources and expertise.
