📰 Key Takeaways

A new Ramp report says enterprise AI token usage keeps climbing, but overall spending has nearly stalled. From August to September 2026, token spending grew just 1.5%, compared with a 42% jump over the prior month in July. Ramp’s chief economist Ara Kharazian says almost all AI spending right now comes from venture-backed, high-growth tech companies. The top 1% most aggressive AI adopters had been spending close to $10,000 per employee per month at their summer peak, and that dipped slightly in August and September. That slowdown among this group alone is enough to drag down overall revenue, even though usage is still increasing. As for typical companies, the median company spends only $12 to $15 per employee per month, and the top 10% spend roughly $600 to $1,000. Both groups grew, but not enough to make up for the gap at the top. Kharazian thinks software and finance have big demand for AI coding agents, but clinics or car dealerships don’t necessarily need to spend the same amount. He also points out that the problem isn’t demand, it’s market competition: OpenAI cut API prices for GPT-6 Sol and GPT-6 Luna in half in September, and Anthropic then sharply cut prices for Fable 5.1 and Opus 5.5 too, saying in recent filings that the move reduced revenue growth by about $600 million. Both companies also offer enterprise discounts to lock in spending, which in practice is pushing revenue down. In October (September 23 to October 23), token spending rebounded 8.9%, but he believes headwinds to growth are still strong. It’s now a buyer’s market, and the era of hyper-growth may be over, which is especially worth watching for Anthropic as it prepares for its IPO.

💬 JudyAI Lab Take

This story matters because AI token usage is still going up, yet overall enterprise spending has nearly stalled. Ramp’s report shows token spending grew only 1.5% from August to September, after a 42% jump over the prior month in July. The growth rhythm of the AI market has clearly changed.

For AI builders, this case points to two signals. First, spending is highly concentrated. Almost all AI spending right now comes from venture-backed, high-growth tech companies, and the top 1% of companies alone can swing the overall numbers. The median company spends only $12 to $15 per employee per month. Second, demand hasn’t disappeared. The pressure is coming from price competition. OpenAI cut API prices for GPT-6 Sol and GPT-6 Luna in half in September, and Anthropic followed by cutting prices for Fable 5.1 and Opus 5.5. Lower unit prices drive more usage, but that usage doesn’t turn into revenue at the same rate. The original article also notes that this is already a buyer’s market, and the era of hyper-growth may be over.

When you plan a product, don’t assume model prices will stay the same. We can go through our own cost structure and ask: if unit prices fall by half again, will our pricing and margins still hold up?

📅 Source Info

🔗 Further Reading