AI prices are falling. Most firms have no baseline to know what they're saving.

Nobody at most firms can tell you, to within a few hundred dollars, what they spend on AI tools in a given month. Seat licenses are small, API charges are buried in software budgets, and nobody has pushed back on the invoice yet. That comfort gets more complicated now. OpenAI and Anthropic have both cut model prices — a direct response to cheaper Chinese alternatives that have been undercutting US providers on capability per dollar. The reductions flow downstream: every platform your staff runs that sits on these APIs gets cheaper to operate, which gives vendors room to cut their own seat prices, widen free tiers, or add capability without raising the bill. But cheaper access also removes the price objection for firms that were waiting before deploying these tools — the firms that were already running them don't gain as much ground as they assume. Build a spending baseline before the comparison matters.