Tech Newsround – Apple & The Magic Money Tree

Apple – Fatal Flaw.

• Apple is reportedly considering launching an AI-powered search service of its own, but without a deep competence in building either an LLM or a search service, I don’t think that this idea is going to go anywhere.
• The idea is that Apple builds its own search service called iAnswers or something similar, deeply integrates it into all of its devices and services and thereby can terminate its arrangements with Google and OpenAI and bring an important part of its ecosystem back in-house.
• Assuming that Apple is capable of building this (big if), then this would result in a substantial increase in the value of the iOS ecosystem, increase Apple’s differentiation over its competitors, and probably widen its margins considerably.
• Consider Google, which is willing to pay Apple $20bn+ or more to be the default search provider on iOS, from which RFM Research has long estimated it derives half of its mobile revenues.
• If this value were to accrue to Apple instead of Google, one can see just how valuable this would be.
• However, there are a lot of issues:
  • First, Search: where there has been no discernible sign of share loss to AI-generated search despite AI having billions of MaUs.
  • Second, AI competence: where Apple has proved through its fumbled launch of Apple Intelligence that it is very weak.
  • Third, Search competence: where Apple also demonstrably has no competence, having outsourced it to Google for years.
• Consequently, while this is obviously a good idea given the upside for Apple that could be generated, the chance of this strategy succeeding is virtually zero, and so I remain pretty unconcerned with the outlook for Google and OpenAI’s ability to generate revenues from iOS devices.

AI – The Magic Money Tree

• Summer is over and the silly season has restarted with investors willing to throw ludicrous amounts of money at companies that should either be making profits by now or have at least some revenue.
• The likely outcome is that these investors will end up being disappointed when the promised super-intelligent machine fails to appear, leading to a correction, although not as bad as the Internet bubble of 25 years ago.
• Anthropic has raised $13bn in a series F at a ludicrous $183bn, while Mistral is raising €2bn at a €12bn valuation, neither of which makes any sense at all.
• Meanwhile, companies like Safe Superintelligence and Thinking Machines command valuations of $32bn and $10bn, respectively, despite having no revenues and no product.
• There is only one scenario in which any of this makes any sense, which is if one of these companies invents an AI that is more intelligent than humans.
• Should this happen, then the inventor is likely to become the most valuable company ever created, which is what all these investors are betting on.
• Unfortunately, all of the empirical evidence suggests that none of them will be successful for the foreseeable future, meaning that all of them are substantially overvalued.
• This does not mean that AI based on LLMs and SLMs is not valuable, and RFM forecasts that large amounts of economic value will be created from new products and services.
• However, without super intelligence, these companies are all working on products that are likely to become commodities, and even today, one can clearly see that the differences between the leading LLMs are far less than they were just one year ago.
• Hence, they are never likely to live up to the valuations that are being attributed to them, and when the Magic Money Tree runs out of bounty, there will be a correction.
• These companies will feel it most, and I suspect that most of them will go out of business and be acquired by someone with very deep pockets (e.g. Apple).
• This is why I have no interest in going near any of them and instead would invest in the picks and shovels (Nvidia, Samsung, SK Hynix, Micron, etc) or adjacencies such as nuclear power and inference at the edge.