Gone are the days of bootstrapped startups bartering bitcoins for pizzas and shares for garage space. Capitalization of revenues is around for future generations to have their lores: one that involves listed giants and pressures traditional VC funds.
Take Anthropic’s Claude: Its annualized revenues crossed USD100m around the end of 2023. During the year, it has done five funding rounds totaling USD7.3bn. Apparently, the traditional "Series X" funding rounds with their quaint at-least-annual gaps are passé.
Amazon’s just-announced US$2.75bn investment into Anthropic is not exactly a cash windfall into the latter. It is against Claude using AWS as its provider. In the crudest terms, the cloud providers’ revenues increase while the receivables are capitalized as “investment”. In a bull market, higher revenues/profits fetch higher operating valuations, while the value of the investment book too can keep going up as your invested companies do more payable-through-share-capital deals with other providers at higher valuations later.
https://cnb.cx/43ClQ1T details how the largest VC funds are now listed companies, funding start-ups ad hoc as and when the growing companies appear in front of them for more resources. The article cites how big tech companies’ AI investments jumped over 5x in 2023 to USD24bn+. This is a chunky amount that will make a large part of the companies’ incremental revenues.
Pre-2007, financial market companies found innovative ways to capitalize payables to enhance revenues, valuations, and executive bonuses. The ensuing regulations restored some sanity, and it seems the non-finance world might be headed for a similar reckoning.
It is another week, and another open-source model is announced straight to the top of the performing charts, like perpetually chart-topping new songs in music. In many ways, this is another indicator of the upstream of GenAI, all the way to the LLM developments, developing oversupply, excesses, and vulnerabilities. Applications and revenues must follow soon, and the embodiment of GenAI is the likely next stage before bigger innovation applications take over in a few years.
Disembodied GenAI – whether open-sourced LLMs on platforms to copilots, chatboxes, and agents – are instantaneously copied and feverishly changing. Their immediate embodiment should include all existing hardware being upgraded to include AI and attempts with newer form factors. On current evidence, the new form factors will have to go beyond pins and pendants to interest consumers and the efforts are being made everywhere. The more meaningful embodiment that involves new TAM opportunities will be in Robotics.
So, while the current financial gymnastics surrounding capitalized revenues might raise eyebrows, the future of embodied GenAI is brimming with potential. The key is to navigate the financial fog and focus on the transformative applications that lie ahead.