AI Ecosystem – Battle Lines

The reset will start the battle.

• Amazon has upped its investment in Anthropic which virtually guarantees that when the money runs out, Anthropic will be acquired and become the AI department of Amazon, creating a fully vertically integrated player ready to fight for the AI Ecosystem prize.
• This is a confirmation of what has been clear for some time and with Open AI with Microsoft, Meta, Google and Apple on their own and a number of other smaller players, it is already pretty clear how the field will look once the inevitable reset has happened.
• Amazon has added another $4bn to its investment in Anthropic bringing the total to $8bn which cements the virtual certainty that Amazon will end up acquiring Anthropic.
• It is no secret that Amazon has a problem with its AI.
• The Alexa personal assistant remains as witless as ever and Amazon’s algorithm is still spending money on sending me advertisements for products I have already purchased on its site.
• 7 years ago, Meta Platforms was in a similar position but, since that time, it has invested in the right places, hired the right people and now RFM Research rates it as one of the leaders as opposed to the hopeless laggard it was in 2018.
• Foundation models (GPTs, Gemini, Llama, Claude etc) are rapidly becoming the operating systems of AI and, as such, they are likely to be control points making them strategically vital to making money from AI.
• This is why Amazon needs to acquire one of the leading foundation model providers or risk being left with nothing when the current excitement dies down and everyone gets down to the real business of earning revenues and profits.
• Given the $8bn that Amazon has invested in Anthropic and the fact that it invested significantly when the valuation was much lower, I suspect that Amazon has effective control of Anthropic.
• Furthermore, Amazon’s investments (like Microsoft’s) will have come with the condition that Anthropic train its models using Amazon infrastructure and most importantly, its silicon (Trainium and Inferentia).
• This is because the one thing that all of Nvidia’s competitors are struggling with is traction with developers.
• Hence by controlling one of the main foundation models, Amazon can specify that the model is optimised to run most efficiently on its in-house silicon thereby reducing its dependence on Nvidia.
• This is what most of the main foundation model providers are trying to do but so far, it is slow going.
• I suspect that the event that triggers this migration will be a reset of expectations and valuation back to more realistic levels.
• RFM research has long concluded that statistical-based systems are unable to replicate human intelligence and so the dream of super-intelligent machines doing everything remains as far away today as it was 10 years ago.
• This is not what the market is pricing in and so when this reality hits, there will be a big reset as well as a lot less money to invest in bigger models and data centres.
• Less money will mean consolidation with most of the current crop of start-ups going out of business as well as the acquisition of OpenAI, Perplexity, Anthropic, Mistral and so on by larger rivals who have enough money to survive difficult times.
• It is at this point that I suspect the pursuit of super-intelligent machines (in which over $250bn is being invested this year alone) will give way to the practical application of these new technologies of which there are a lot.
• Generative AI is capable of generating meaningful revenues and profits now which is why I do not expect a reset as hard as the Internet Bubble reset of 2000-2004.
• Limited resources will focus investments on where they can be most productive instead of chasing a scientific nirvana for which there is still no concrete evidence.
• It will also crystalise the contenders who will go on to fight for the AI Ecosystem, and at the moment, it looks like it will be the usual suspects.
• This means that Microsoft, Amazon, Apple, Meta Platforms, Nvidia, Google, Baidu and Tencent will be the main contenders with everyone except Apple, Google and Tencent determined not to repeat the mistakes that led to them losing the smartphone ecosystem.
• I think that most of the smaller players will struggle to survive unless they occupy a niche of which an example is G42 and Cerebras who are building models optimised for Arabic speakers and the Global South more generally.
• The shakeout is coming and with it a more rational but more competitive environment.
• However, the killer robots are not.