The contest to build artificial intelligence is entering a new phase, and it is no longer just about who has the biggest model.
According to CNBC, companies are starting to choose AI models based on the specific task at hand, the cost of running them, and how much control they retain — rather than simply picking whichever model tops a public leaderboard.
That marks a shift in priorities. For the past several years, the headline story in AI has been scale: ever-larger models trained on ever-more data, with rankings treated as the main scoreboard. CNBC reports that the emphasis is moving toward systems that are cheaper and smarter, meaning better matched to the job they are meant to do.
In practice, a leaderboard-topping model may be overkill — and overly expensive — for routine work. A smaller, cheaper model that a company can run and govern on its own terms can be the more sensible choice, according to CNBC's framing of how buyers are now thinking. Task fit, price, and control become the deciding factors instead of raw ranking.
Why it matters: how businesses pick AI shapes what the technology costs, who controls it, and where it actually gets used — so a move away from chasing the biggest model toward cheaper, more practical systems could influence the direction of the entire industry.