Last week, Amazon announced plans to invest $4 billion in Anthropic. Shortly after, rumors began swirling that the company was in talks to raise $2 billion more of funding from investors, including Google, at a $20-30 billion valuation, and expected to hit $200 million in revenue by the end of 2023. In July 2023, Anthropic announced the release of its second large language model, Claude 2, and as of September 2023, the chatbot remains in an open beta.
The latest funding for Anthropic from the likes of Amazon and Google is the latest of a series of large, deeply entrenched technology companies proving that the current generative AI wave is less of a revolution for AI startups, and more an extension of the cloud computing wars. With Microsoft, Google, Meta, and now Amazon all competing in AI, it raises the question: are the advancements in AI most impactful to startups, or is this the next foundational layer of computing that is more relevant to the deeply entrenched technology players?
In the early 2000s, venture capital was dominated by ‘internet investors,’ investors that specifically targeted what they called ‘internet companies.’ While internet companies were a popular investment at the time, two decades later, the term ‘internet investor’ is almost laughable. Today, every company leverages the internet. The same is true of big data; every company leverages data.
Potentially reminiscent of the early 2000s internet investors, in 2023 we are seeing a rise in generative AI investors. Goldman Sachs projects that private investments in AI will increase from approximately $92 billion in 2022 to over $158 billion in 2025, a 72% increase over three years. But skeptics argue that the AI funding frenzy could be forming a bubble, with capital thrown at companies that lack earnings, founders with little expertise, and in some cases, companies that lack a product. For example, when Mistral AI raised a $113 million seed round at a $260 million valuation, the company didn’t have a working product; instead, Mistral AI had a vision to compete against OpenAI in the building, training, and application of generative AI models.
The notorious DotCom bubble led to a generation of “internet” startups failed to turn a profit, and startups abandoned fiscal responsibility and spent up to 90% of their budgets on advertising in an attempt to edge out competition without a proprietary product. But the internet became a foundational piece of every technology company rather than a differentiating factor between competing startups.
The rise of AI has similarities to the rise of internet companies in the early 2000s. The benefits of generative AI have largely been concentrated among the deeply entrenched technology players, with the stock prices of Microsoft, Meta, and Amazon skyrocketing. Shares of Nvidia, for example, have tripled this year largely because its GPUs are highly sought after for AI. Meanwhile, early-stage investors are underwriting startups without proprietary products with inflated valuations using these entrenched players as comparable companies despite many not having a clear path to profitability.
The FTC has taken note of the concentration of benefits of generative AI products, explaining the difficulties that nascent competitors will have against established incumbents. Like the internet, AI could become a foundational layer of technology companies with the most prominent benefits concentrated among the established players like Meta, Facebook, Google, and Amazon. Generative AI may not be a game for startups, but rather a foundational piece of every technology company of the future. In another decade, will we view ‘AI Investors’ in the same light as the ‘internet investors’ of the 1990s and 2000s?
