After media-for-equity and advisory-for-equity, GPU-for-equity is the new kid on the VC block. As VCs increasingly face commoditization of capital (the ironically million-dollar question "how is your money different from that VC's money?") in the hot AI infra/applications startup space, this appears to be the newest lever for differentiation among investors. Marquee global VCs such as Andreessen Horowitz, Index Ventures, Conviction VC et al. are now renting out thousands of GPUs to their portfolio's AI companies in lieu of equity, or even as a complimentary platform service. The VCs are buying or renting them from large chipmakers. As the demand for AI chips continues to soar - and even capacity powerhouses such as NVIDIA grapple to keep up with it - this enables access for their portfolio companies to an expensive and increasingly scarce resource. This new standard of the 'entrepreneur service' model also of course helps a16z one-up competing VCs in winning hot AI deals. While a clutch of AI companies including OpenAI, Mistral AI, xAI are part of a16z's portfolio, Luma AI's CEO categorically said that the VC's GPU supply made them pick a16z over other investors. While countless sectors just can't seem to escape the long-drawn funding winter, clearly investors are vying to embellish 'capital' as much as possible to woo the hottest contenders in several sunrise sub-sectors within deep tech. The convergence of technology, equity, and strategic resources underscores the evolving nature of VC engagements in the quest for market differentiation and value creation.
Good to know!
Strategy and Investments | Ex-VC at Times Bridge (Times Group) | Ex-Unilever | IIM Indore
4moAn interesting follow up infographic I came across today which shows that Andreessen Horowitz is perhaps among the top 4 hoarders of NVIDIA's chips. Letting it sink in: