What Happened
DeepSeek has reportedly reached a $1 billion annualized revenue run rate, up from under $500 million a few months earlier. The jump followed API price increases of between 2.3 and 4.5 times the previous levels. Reports say demand stayed strong and the customer base remained stable, with no churn.
Most of the company's computing capacity still goes toward training new models. A low-cost provider has shown it can charge much more without losing customers, which tells every other AI buyer that today's prices are not a promise.
Three Takeaways for Businesses
- Do not budget on today's AI price. If a provider known for low prices can raise them 2.3x to 4.5x, any AI line item in your plan can move. Build a margin for price changes.
- Avoid single-vendor lock-in. Keep prompts, data, and workflows portable so you can switch models when pricing or quality shifts.
- Spend AI on work that earns revenue. If prices rise, tasks tied directly to leads, sales, and customer service are the ones worth paying for.
Customers stayed even after prices rose several times over. That suggests buyers value a working AI system more than a cheap one, so reliability and fit for your process matter more than the lowest rate card.
What You Can Do This Week
List every place your business calls an AI model, note the monthly cost, and ask what happens if it doubles. Then pick the one workflow where switching providers would be hardest and make it portable first.
