From investors’ hands to Nvidia’s pockets.
- xAI has raised $6bn in new funding that will be used to bring products to market and build infrastructure but with an unremarkable product suite so far, the main beneficiary of this new cash looks like it will be Nvidia.
- The AI bubble continues to inflate with another $6bn being raised valuing xAI at $18bn which as far as I can see has no commercial product and no revenues to speak of.
- The one asset that xAI does have is distribution as the X platform continues to have good engagement and plenty of users even if it is not as good at generating revenue as it was a few years ago.
- This means that the xAI can integrate its products with the X service and drive engagement, but this is still no guarantee of revenues.
- However, the real problem with xAI is that it is unremarkable by today’s standards which leads me to wonder why anyone would want to use it outside of X where they would have no choice.
- What we are seeing in 2024 is the beginning of the battle for the AI ecosystem where everyone is extolling the performance of their models and pitching developers to come and develop generative AI using their foundation model as a starting point.
- This is key because foundation models are quite difficult and expensive to create and, as such, are something that small developers and hobbyists can’t really do on their own.
- This is why foundation models are becoming the platform of the generative AI boom just like iOS and Android before them were to smartphones.
- The field is getting crowded and the last 12 months have seen the gap between the laggards and the leaders close significantly meaning that becoming the go-to platform of generative AI is going to be hotly contested.
- However, for the moment this does not seem to matter as capital continues to pour into the sector at increasingly ludicrous valuations.
- This is always a really bad sign because it means that projects that would never get funded in an orderly market receive capital which creates oversupply and brutal competition which in turn causes the bubble to burst and a period of depression and reflection.
- The extent of the bubble is on full display in the $18bn valuation of xAI and a study by The Economist which estimates that the AI Boom has added $3tn to the valuation of Amazon, Google and Microsoft while incremental revenues in 2024 will be just $20bn.
- This equates to the market valuing revenues from AI at 150x for 2024 which is clearly unsustainable and a symptom of too much cash chasing too little paper.
- Every single time that this circumstance occurs, the supply of paper rapidly increases to soak up the available cash which is what we are in the process of observing now.
- Generative AI undoubtedly has novel capabilities that enable a range of lucrative and valuable use cases, but it is not the dawn of superintelligent machines that the market is currently pricing in.
- The turn will come when this reality becomes clear at which point there will be a significant reset.
- The big question is not when this happens but who will survive it and here, I expect that it will be the big digital ecosystems that survive as they also have highly lucrative businesses that will enable them to absorb the losses.
- xAI is in a far more tenuous position as is OpenAI, Anthropic, Mistral and so on as they have nothing to fall back on when the shakeout begins.
- Hence, I think that OpenAI will end up as part of Microsoft, Anthropic as part of Amazon and Mistral as part of someone else.
- This will leave the platform battle to be fought between the big players who have the resources to ride out the correction.
- The one exception here is Nvidia which is currently earning around $100bn a year in revenues from AI chips from which it earns 78% gross margins.
- This is why its valuation is far more reasonable than most of the other generative AI players which means that it is likely to correct less when the market falls out of love with generative AI.
- Hence, if I was forced to make a pure play investment in generative AI, this is where I would go, but I prefer to look more laterally where there is more value to be had.
- One of these lateral arcs is nuclear power which is going to be needed to power all the data centres and another is inference at the edge (which Microsoft has just championed) where Qualcomm and MediaTek are likely to be the big players.
- I have a position in both nuclear power and Qualcomm and remain a happy holder.
