The winner will take it all in brave new world of AI

As the major tech giants battle it out to launch the latest chatbot, spending billions in the process, only one company will succeed

Another week, another major artificial intelligence (AI) launch. Meta has just released Llama 3, along with lots of hype about how it will blow all the other competitors out of the water. It comes hot on the heels of new offers from Google, Amazon and Chat GPT. Meanwhile, Microsoft has just invested another $1.5bn (£1.2bn) on the UAE’s G42 while the much-hyped French start-up Mistral is reported to be raising more money at a valuation of $5bn, an extraordinary sum for a company less than two years old.

There are so many different AI systems to choose from you probably need a super-smart chatbot just to keep track of them all.

Here’s the problem, however. For all the billions being spent right now, only one of them will survive. AI is going to be a “winner takes all” industry, and if there is any money, which looks far from certain right now, it will be made by just one player – and most of the cash poured into the technology will end up completely wasted.

If you have any money invested in the markets, the chances are a chunk of it will be going into one AI system or another.

Only this week, Meta, the company that controls Facebook, Instagram and WhatsApp, launched Llama 3 onto the market. According to the company, it has “vastly improved capabilities” that will take it a step closer to human intelligence. According to its head of global affairs, the former deputy prime minister Sir Nick Clegg, it will also be a little less “sanctimonious”, which is, to put it politely, something of a surprise from someone who used to lead the Liberal Democrats.

The field is getting very, very crowded. Over the past few weeks, we have seen Gemini from Google; GPT 4 Turbo, which sounds more like a Formula One car than a piece of software, from ChatGPT; and 8x22B, hardly a snappy name, from France’s Mistral. And Elon Musk has released Grok to paying subscribers to X (yup, all seven of them). We can expect even more to be launched over the next few months, especially as some of the major Chinese competitors wade into the market. At a cost of a few hundred million dollars a time, if not a billion or more, the world’s major tech corporations, or else the big venture capital houses, are each placing huge bets on a technology they believe will be dominating the global economy within just a few years.

There is a catch, however. Unlike, say, automobiles, where there are a dozen or more major players; or banking, where there are a hundred or more; or even oil, where there are five or six global giants, AI looks poised to become a classic “winner takes all” industry.

The reason is simple. AI is very similar to the search market, which it may well replace.

There were lots of search engines around before Google, from Yahoo to AltaVista to AskJeeves. They are all forgotten now. There were two big forces at work.

We all gravitate towards the standard one that everyone else uses, so that we are all on the same page, and once that process is under way, the market leader can generate more revenue, invest more in the product and steadily pull away from the rest of the pack. At a certain point, its momentum becomes unstoppable. The winner doesn’t just take 30pc, 40pc or 50pc of the market. It takes the whole cake, and leaves nothing on the table for anyone else.

In AI, that may be even more true than it was for old-style internet search engines. The sums of money involved are huge.

The large language models that are the basis of all the AI systems flooding the market at the moment use computing power and even electricity on an epic scale. There is a reason why Nvidia, which makes most of the microchips to run chatbots, now has a market value of more than $2trillion, and its shares are up by 78pc this year alone. It is selling a ton of highly priced chips. Even the electricity bills will be daunting. One estimate last year was that by 2027, AI servers would use as much electricity as the whole of Sweden. Someone will have to pay for all that power.

Meanwhile, as competition becomes more intense, salary costs are going through the roof. Programmers can ask for whatever they want. On top of that, many of the systems have to pay for copyrighted material to “train” the software. Put it all together, and chatbots are expensive to run.

It is not even certain that there will be a profitable business at the end of it.

Some of the main models have tried charging for access, but it remains to be seen whether they can make that stick. With so many rivals on the market, including Open Source models such as Mistral’s, there will be lots of pressure to make the product free.

Advertising may well not ride to rescue. We don’t mind a few sponsored messages on search engine results too much, but we won’t put up with a smart chatbot casually throwing in a suggestion for a new car, or switching bank, into every conversation that we try to start.

With everyone gravitating towards one model, with soaring costs, and with very little in the way of revenues, there is not going to be room for many runners in this race for long.

It is impossible to say which one it will be, although Anthrophic’s Claude AI backed by Google and Amazon may well prove to be the smart bet. Within the next year, only one winner will emerge to scoop the prize, such as it is – and all the others will have simply burnt their way through billions of dollars chasing a market that has been locked up by one of their rivals.

‘With so many rivals, there is not going to be room for many in this race for long’