The primary business model shift driven by AI is from per-seat software (SaaS) to selling units of automated labor or work product, fundamentally changing how value is created and captured [82, 56].
The best defensive strategy for companies in the fast-moving AI market is not a data moat, but building a multi-product bundle that becomes deeply integrated into a customer's workflow [15, 106].
The future of large-scale AI development is dictated by physical world constraints, primarily the availability of cheap, dense energy and the specialized hardware supply chain, which will concentrate power geographically [21, 45, 88].
The current AI market is so open and fast-moving that companies not experiencing explosive, near-immediate growth likely have a fundamental flaw in their product or strategy [113].
Government actions, such as heightened M&A scrutiny and state-level regulations, pose a significant threat to the technology ecosystem by limiting startup exit opportunities and curtailing innovation [24, 25].
2016-2017
Began investing heavily in the cryptocurrency sector, making an early investment in Coinbase based on the thesis that it was an index on the entire market [11, 92].
Circa 2017
Made a contrarian investment in defense technology company Anduril, founded by Palmer Luckey, at a time when it was unpopular for Silicon Valley firms to back weapons companies [2, 9].
2021-2022
Pivoted his investment focus to the generative AI sector, making early-stage investments in companies like Perplexity, Harvey, Decagon, and Abridge as foundation models became more accessible [8, 79].
Circa 2023-2024
Co-founded Branko with Jared Kushner and Eric Wu, an enterprise AI company aimed at large corporations and governments, and co-led its $30 million Series A round [4, 5].
Present
Actively advises founders in the AI space, warning of a potential valuation peak and recommending they consider selling within the next 12-18 months. He also focuses heavily on the physical constraints of AI, such as energy and hardware supply chains, as the next major bottleneck [90, 21, 88].
▶The Physical Constraints of a Digital Revolution
Gil repeatedly emphasizes that the primary bottlenecks for AI progress are shifting from software to physical infrastructure. He highlights current constraints in the supply chain for specialized memory [74, 88] and predicts future limitations will be energy availability, power grid capacity, and the speed of data center construction [21, 116]. This leads him to conclude that Europe is becoming uncompetitive for AI training due to high energy costs [45].
This theme suggests that the next phase of the AI race will be won not just by superior algorithms, but by securing control over the physical supply chain, creating significant investment opportunities in energy, hardware manufacturing, and real estate suitable for data centers.
▶AI Market Structure: Hyper-Growth and Oligopoly
Gil characterizes the current AI market by its unprecedented growth rates, with companies reaching billion-dollar revenues in a single year—a pace far exceeding previous tech giants [32, 75]. He believes this market is consolidating into an oligopoly dominated by a few foundation model players aligned with major cloud providers [119], with 91% of private AI market cap concentrated in the Bay Area [76].
For investors, this points to a power-law market where returns will be highly concentrated in a few dominant platforms, making early identification of and access to these potential winners a critical factor for success.
▶The Great Business Model Shift: From SaaS to AI Labor
According to Gil, the fundamental business model disruption from generative AI is a shift away from selling per-seat software (SaaS) to selling automated units of labor or work product [82]. He cites companies like Sierra and Gagana, which offer utilization-based AI agents for customer support, as prime examples of this transition [56]. He advises companies to build multi-product bundles to create defensibility in this rapidly changing landscape [15].
This analysis signals a significant threat to incumbent SaaS companies whose value is based on seat-based licensing, while creating an opening for new startups that can price their services based on tangible outcomes and automation.
▶Founder and Investor Strategy in a Frothy Market
Gil's commentary provides a playbook for navigating the current AI boom. He advocates for a consensus investment approach, arguing that companies not showing explosive growth likely have a fundamental flaw [112, 113]. For successful founders, he gives the stark advice to seriously consider selling within the next 12 to 18 months, suggesting the market may be at a peak valuation opportunity [90].
Gil's perspective suggests that traditional 'wait-and-see' or long-term growth strategies may be ill-suited for the current AI cycle; speed, rapid scaling, and opportunistic exits are paramount.