At TechCrunch Disrupt 2025, renowned solo VC investor Elad Gil commented that AI has proven to be one of the most unpredictable technological surges he has witnessed.
Gil is listed on the capitalization table of almost every successful company in the last ten years, notably several of today’s prominent AI firms.
Nevertheless, he believes that over the past year, specific AI markets seem to be nearly dominated by leading companies. Beyond these specific areas, a large portion of AI is still open to anyone.
“I began investing in generative AI in 2021… [A]t the time, it wasn’t getting a great deal of attention,” Gil stated. However, he had observed the substantial progress in capability between GPT-2, introduced in 2019, and GPT-3, introduced in 2021. “The advancement from 2 to 3 was so considerable that by extrapolating the scaling laws, or the curve, one could infer its future importance,” he explained.
That motivated him to support early-stage startups creating products using large language models. His investments spanned both basic model developers like OpenAI and Mistral, and application companies such as Perplexity, Harvey, Character.ai, Decagon, and Abridge. Yet, throughout 2024 and much of 2025, each new release brought substantial advancements in the capabilities of basic models, transforming AI every few months.
“I often said that AI was the one market where the more I learned, the less I understood. Typically, increased knowledge leads to better understanding and easier prediction of the future, etc. However, AI remained unclear. There was simply too much uncertainty. And I believe there are still AI markets like that,” he added.
Nonetheless, he now identifies markets with distinct leaders. The most obvious instance is with the foundational models themselves. Despite the existence of numerous models, and ongoing efforts by countries like South Korea to develop independent models through local companies, leaders have come to the forefront. “Google, Anthropic, OpenAI, maybe xAI, possibly Meta, perhaps Mistral — there are only a few,” he predicts regarding the eventual winners.
Techcrunch event
San Francisco
|
October 13-15, 2026
Following models, he suggests that AI-assisted coding has definitive frontrunners that will challenge newcomers. Not only have the foundational model creators become involved (Anthropic with Claude Code, OpenAI with Codex) but also leading startups like Anysphere’s Cursor and Cognition’s Devin (which acquired Windsurf) will be difficult to surpass. Additionally, well-funded startups like Magic (which Gil considered a potential “outlier”) and Poolside are close behind.
He believes that medical transcription is becoming consolidated, with Abridge as a leading competitor and others, such as Ambience, holding significant positions.
He mentions customer support — an initial focus for both traditional AI and the new wave of AI agent startups — as having market leaders that are difficult to overtake, like his portfolio company Decagon. (It secured $131 million with a $1.5 billion valuation in June.) Bret Taylor’s startup, Sierra, also competes here. Incumbents like Salesforce, HubSpot, and many others are also incorporating AI into their offerings.
So, which markets appear to be still accessible? Gil points to financial tooling (fintech), accounting, AI security, and “other markets that we inherently find intriguing. We just don’t yet know which company will dominate.”
Ironically, rapid growth is no longer a reliable indicator of a company’s potential for significant success. “The CEOs of every major company are essentially directing their teams to prioritize and create an AI strategy,” Gil noted. “These large enterprises are willing to experiment with strategies they would have dismissed two years ago, solely because of AI.”
Therefore, new AI markets can swiftly secure substantial revenue from prominent enterprise clients, “but that doesn’t guarantee long-term retention,” Gil remarks.
It’s only following a market’s initial boom and trial phase that a startup and its investors can determine the sustainability and growth potential of the revenue. “There are misleading indicators, and then there are elements that are genuinely effective,” Gil observed. He identifies legal AI startup Harvey as a market leader that is “genuinely effective.” It completed three major funding rounds in 2025, with its valuation increasing from $3 billion to $5 billion and then to $8 billion in a matter of months.
