Where is enterprise AI heading over the next three years?
Short answer: unit prices keep falling, total bills keep rising, models keep converging, and EU rules phase in through 2027. None of it changes the play: own the data, the evals and the workflow; rent the model. Run your own numbers in the calculator below.
As token prices halved, usage grew about 450 percent (Bain). Gartner expects over 40 percent of agentic AI projects canceled by 2027. Falling unit prices and rising bills are the same forecast.
Unit prices fall, total bills rise, models converge, rules tighten. None of it changes the strategy: own the workflow, rent the model.
// the three-year forecast
your assumptions · open math
Set your own numbers. The point is the shape, not the decimals: the licence line stays small while usage growth and running costs decide the real bill.
3-year total, your assumptions
€85,500
All licences. Cheap to start, nothing owned at the end: no data loop, no integration, no moat. If one workflow is genuinely yours, price the build scenario too.
// the deep dive
Predicting model capability is guesswork, so this page does not try; the LIVE strip below tracks the fast-moving part. What can be forecast is structure. One: unit prices keep falling while agent workloads multiply token use 10 to 100 times per task, so budgets follow usage, not price lists; run your own numbers in the calculator above. Two: frontier models keep converging, which means whatever advantage you rent, your competitor can rent next quarter; only the data, evals and integration you own compound. Three: regulation arrives on a schedule, not a vibe. The EU AI Act phases in obligations through 2027, and the EU-US data framework is again being tested in court, so where your systems run stops being an IT detail and becomes board-level risk. Four: the winner gap widens. Adoption is already near universal, and the roughly 5 percent who capture value at scale are compounding while everyone else pilots. Waiting for the model that makes it easy is not a strategy; the bottleneck is your data and your workflow, and those do not improve on their own.
// chapter faq
Will AI get cheaper for companies?
Per token, yes, relentlessly. Per year, usually no: Bain found usage grew about 450 percent as token prices halved, and agent workloads multiply token use per task. Budgets follow usage, not price lists.
Should we wait for better AI models before starting?
Waiting for models is not a strategy, because your bottleneck is your data and your workflow, and those do not improve on their own. A scoped wait is legitimate when the data is not ready; an open-ended wait just hands the compounding to someone else.
What does the EU AI Act mean for timing?
Obligations phase in through 2027, so governance built early is cheap and governance retrofitted late is expensive. It is one more reason the boring layer (access control, audit, data residency) should precede the exciting use-case.
Every figure in this chapter is sourced. The full source list lives on the main map. Open the map
This is one chapter of ten. The whole course is free.
The full map has the interactive tools, the 8 minute audio edition, the live layer and every source. And if you want it run against your own reality, that call is free too.