Business AI: Will 2026 be a battle over tokens?

Logos of different cryptocurrencies are displayed during the Token2049 conference in Dubai in April – Copyright AFP/File Sai Aung MAIN

How will business-related artificial intelligence evolve in 2026? Next year could well see a battle over AI tokens (the fundamental units of data that AI models), plus AI’s falling cost, and the rise of small AI. To gain an insight, Digital Journal canvassed the views of Mike Finley, Co-Founder, StellarIQ. 

A token is the unit of AI processing (so-called input tokens) or generation (output tokens, roughly four times the price of inputs). A text token can be as short as a single character or as long as a word, depending on the language model and how it was trained.  Measuring token usage instead of revenue potentially provides a clearer picture of AI demand.

Foundation Model providers compete fiercely for token share

According to Finley growth is set to give way to a struggle for market share among leading providers: “Major model providers have amassed massive funding rounds and achieved lofty valuations. The demand for their services has grown exponentially so far but 2026 will see providers catch up and begin competing for share rather than growing freely into green-fields as they struggle to meet committed targets.”

The battle for survival will be with those who can offer – or at least show they can offer – a different kind of offering. Here Finley  finds: “As many open models increase in capability, the foundation model providers are locked in a struggle to differentiate their offerings while simultaneously making it easier to build AI into everything, and fend off competitors seeking the same work.  We’re already seeing them invest across the board in consumer applications, enterprise platforms, development tools, partnerships, marketing, commitment-based pricing and more.”

Finding a unique selling point is key. Finley observes: “One area that remains wide open is non-text modalities like audio and video, specifically when needed in real-time such as for conversational interaction.  This is a massive opportunity that, so far, is owned by OpenAI.  Unlike text, these modalities replace human time second-for-second, making ROI instantly clear when it works. You can skim a 10-page document in a moment but you have to talk to a voice representative in real time. Another way they’ll do it is to make sure the tokens are used in more applications.”

Expanding on this, Finley says: “OpenAI simplified the process of integrating their platform into many software “surfaces”. They provided more of the plumbing needed to deliver their tokens so that adoption will rise. Microsoft is doing this also, as are a number of third parties.”

AI Gets Small

For something growing big it may seem odd to describe it as also something getting smaller. This is Finley’s reference to the physical medium: “Next year, we’ll see booming use of AI on smaller devices, demanding far lower power than AI on larger form factors. Low latency is essential for these smaller AI applications, which generally can’t wait to send data back and forth to the cloud for analysis. This will move processing to the edge.  Ultimately, this trend will drive a tidal wave of device replacement over the next 5 years – everything from TVs to thermostats. Models are extremely close to being able to run on existing edge hardware now, but they’re not quite there.”

Keeping things small should matter, notes Finley: “We’ve all heard of the billions or trillions of parameters that models require, but what are those?  In a data centre, they are typically hairy numbers with seven digit precision each, and users notice immediately when that precision is removed.  But experts have been able to “quantize” those values for use on severely constrained hardware if we’re willing to forego some of the capabilities.”
 

Closing out his predictions, Finley expects: “What will it look like? Cars coming out in 2027 will be able to talk to EMS when they’re in a wreck. Washing machines will analyse what’s in the load and set itself.  Ring doorbell will chat with visitors to determine who they are and what they’re up to.  It should be noted that this will not reduce the cost of creating the models: there will continue to be extensive efforts to build powerful AI using very intense resources, which will then be “distilled” for edge devices.”