Most AI Companies Won’t Survive (Tech Investor Explains)

The speaker suggests that founders of successful AI companies should consider exiting in the next 12 to 18 months, as this period might represent a value-maximizing opportunity. This perspective draws parallels to past technology cycles, like the dot-com bust, where a vast majority of companies failed. For instance, during the internet bubble of the 90s, roughly 1,500 to 2,000 companies went public, but only a dozen or two survived. This pattern, where 90-99% of companies in a technology cycle go bust, is not new and has been observed in industries like automotive, SAS, mobile, and crypto. The speaker emphasizes that the AI cycle will likely be no different, with only a handful of companies achieving long-term success. Therefore, AI company founders should assess the durability of their businesses. If a company is among the few destined for lasting importance, it should continue. However, for many others, the current period or the next 12-18 months could be the optimal time to sell before their offerings become commoditized, face intense competition from labs, or become obsolete due to market or technological shifts. Identifying the "handful" of durable companies involves looking for specific characteristics. Core labs like OpenAI, Anthropic, and Google are seen as having a strong, durable position, potentially forming an oligopoly aligned with cloud providers. For application-level companies, three key lenses are suggested: 1. **Improvement with model advancements:** Does the product or service significantly improve for customers as the underlying AI models get better, ensuring continued demand? 2. **Product depth and integration:** Is the company building multiple, integrated products deeply embedded into customer workflows, making them difficult to remove? The challenge often lies in change management rather than AI quality. 3. **Proprietary data:** Is the company capturing, storing, and utilizing proprietary data, particularly in a "system of record" capacity, which can offer a defensible advantage? Regarding exit options, the current market features unprecedented buying power from multi-trillion-dollar market cap companies. Potential acquirers include major labs/hyperscalers (e.g., Apple, Amazon, Google), large tech incumbents (e.g., Oracle, Samsung, Tesla, Snowflake, Stripe), or companies focused on specific verticals (e.g., Thomson Reuters for legal). Another option, often underutilized, is the merger of competitors, especially private companies, to reduce competition and solidify market position, mirroring historical examples like X.com and PayPal. The speaker concludes that while AI as a technology is transformative, only a select few companies will truly endure, making strategic exit timing crucial for many.