Unlock Editor’s Digest for free
FT editor Roula Khalaf has chosen her favorite stories in this weekly newsletter.
Although there was nothing new about artificial intelligence in 2024 that could match the pure wow factor of using ChatGPT for the first time, rapid improvements in the underlying technology still keep the field vibrant. It was dripping. As for 2025, I see things going well.
Will AI development hit a wall?
By 2025, that momentum will have slowed down. Even some of the tech industry’s most optimists believe that simply throwing more data and computing power into training ever-larger AI models, which have traditionally been a reliable source of improvement, is not enough. It has acknowledged in recent weeks that profits have begun to decline. In the long run, this deprives AI of a reliable source of improvement. But other advances should more than make up for the delay, at least within the next 12 months.
The most promising developments appear to come from models that perform a series of steps before returning a response. This allows you to query and adjust the initial response to provide more “reasonable” results. Whether this truly rivals human reasoning is up for debate, but systems like OpenAI’s o3 look like the most interesting advancement since the advent of AI chatbots.
Google, which regained its AI momentum at the end of the year after spending two years catching up with OpenAI, announced how AI’s new agent-like capabilities, such as tracking users’ browser behavior and delivering services, will It also showed how it can make life easier. Complete tasks for you. All these demos and prototypes still need to be turned into useful products, but at least they show there’s more than enough in the lab to keep the AI hype going.
Will an AI “killer app” emerge?
For most people, the rise of generative AI means they’re constantly being prompted to finish a sentence or edit a photo in a way they never thought possible. This is an unsolicited and sometimes useful tool, but it is far from life-changing.
Next year is likely to see the first demonstrations of apps that can intervene more directly. This means it will be able to absorb all digital information and learn from its actions, acting as a virtual memory bank and taking over every aspect of your life. But fearing the technology’s unreliability, tech companies will be wary of using them in large quantities too quickly, and most users will be wary of trusting them as well. Dew.
This means that instead of the real killer apps of AI, we will be left with the “AI in everything” world that technology users are already accustomed to. Sometimes it’s annoying, sometimes it’s useful, but it still doesn’t provide a truly new experience. The age of AI has officially arrived.
Will Nvidia’s GPUs still rule the technology world?
The chipmaker’s huge profits have made it a target of the most powerful tech companies, most of which are now designing their own AI chips. But NVIDIA is moving too fast for its rivals, and the company’s Blackwell product cycle should comfortably get it through the year ahead, although a major product transition period could see a rocky quarter or two. .
That doesn’t mean other companies won’t come in. Three major tech companies plan to use their chip designs in supercomputing “clusters” each containing 1 million chips in 2027, according to chipmaker Broadcom. This is 10 times the size of Elon Musk’s Colossus system, which is considered the largest cluster. Number of AI chips currently in use.
But even as market share begins to erode, Nvidia’s software remains a significant moat for the company’s business, and it should start another significant new product cycle by the end of the year.
Will the stock market AI boom continue?
Big tech companies are in the midst of an AI race that leaders believe will determine the future shape of the industry, and will remain one of the main forces behind the AI capital investment boom. And while some companies begin to claim that applying technology to their business will yield great results, many others, even if unproven, are reluctant to use AI productively. You’ll feel the need to keep spending, even if you haven’t found a way to use it yet.
Whether that’s enough for investors to keep pouring money into AI is another question. It will also depend on other factors, including stock market confidence in the new Trump administration’s intentions to deregulate and cut taxes, and the Federal Reserve’s readiness to continue accommodative monetary policy.
All of this points to a very volatile year with some major adjustments along the way. But with enough liquidity, Wall Street could give in to the AI hype for a while yet.
richard.waters@ft.com
