AI is the newest, shiniest thing we have seen since, well, perhaps ever, at least in the speed with which it has overtaken consciousness.
ChatGPT was released to the ‘wild’ in November 2022. In commercial terms, yesterday.
In that time, it has overtaken discussion, business planning, capability questions, and profoundly changed the face of stock markets.
An amazing outcome for a technology without a business model.
The committed AI infrastructure spending over the next year by the big 5 LLM builders, OpenAI, Amazon, Microsoft, Amazon and Google is over $200 billion. Depending on your sources, this might vary a bit, but may even be on the low side. It does not count the billions being spent by everybody else, largely on setting about leveraging the ‘infrastructure’ delivered by the LLM’s.
Again, depending on your sources, the revenues being generated over the next year by AI suppliers, both of the infrastructure and tools rapidly becoming available is probably $20 billion.
Nowhere in history has there been a tsunami of investment of this size and speed in the absence of a solid business model. There is no clear way forward to generating a return on that investment.
This is the equivalent to a goldrush, except, in a gold rush if you were the lucky one to find those elusive nuggets, you had some idea what they were worth, and an established way of monetising the metal.
Nothing of the sort exists with AI.
I have done plenty of capital proposals in my time, some with forecasts that bordered on the wildly optimistic because I believed a change of some sport would be generated by the object of that Capex. In my wildest dreams, I have never proposed anything like the ratio of capex to current revenue exhibited by this investment.
There is confusion around the term ‘Trillion’. Historically, the US and UK definitions differed, the UK version being 10^18, 1,000 times larger than the US trillion which is to the power of 12, or one million million. I explain this for clarity and comparative purposes.
On current stock market valuations (August 2024) Nvidia, a business few had heard of a year ago, is the most valuable company on earth with a valuation of US3.2Trillion. They trade places regularly with Apple for the No. 1 spot. Currently Apple is number 2, also at a rounded 3.2 trillion, but a few tens of millions behind Nvidia. Microsoft is third with 3.1 trillion, followed by Amazon at 1.9, and Meta at 1.3. The comparison I wanted to highlight is with the GDP of Australia, of US1.7 trillion. Australian GDP is just over half the market valuation of Nvidia and Apple, a sobering thought.
An investment of 200 billion against current revenues of 20 billion is simply the biggest financial gamble in the history, by a logarithmic amount.
The people running these massive businesses are not stupid. They see and are betting their companies (and they are ‘their’ companies, as control is in a very few hands) on massive returns, which means in turn that the fabric of everything we see and do must change, very quickly. The business models will change, and they will not be just everybody subscribing for modest monthly amounts to the latest LLM model. There will have to be whole new industries being ‘invented’ with successful business models in place for there to be a return on the capex being deployed.
The windows of opportunity that will open, and close just as quickly, over the next decade are immense.
No wonder there is a gold rush, it is just the location of the gold still in question.
