AI Spending Growth Slows as Competition Drives Down Prices

Spending on artificial intelligence stopped growing in late summer even though companies were using more AI than before.

Data from Ramp, a business spending platform, shows that token spend grew by only 1.5% between August and September. In July, it had jumped 42% from the previous month.

This sharp slowdown surprised many observers who expected AI spending to continue rising fast.

The slowdown matters because almost all AI spending comes from venture-backed technology companies that are growing quickly. If these companies slow down their spending at the same time, the entire AI industry suffers.

Currently, the top 1% of companies have reduced their spending slightly from their summer peaks, and that alone is enough to slow down total revenue across the market, even though companies are using more tokens overall.

Price competition is the main reason for this slowdown. OpenAI cut API prices in half for some of its models in September.

Anthropic responded by cutting prices for Fable and Opus models, a move that reduced its revenue growth by about $600 million, the company reported. Both companies also offered enterprise discounts to lock in major customers, but these discounts hurt their revenue targets.

Companies are also becoming smarter users of AI. As teams gain experience, they learn to use cheaper models instead of expensive ones. They set budgets, control which tasks can use AI, and route simple jobs to lightweight tools.

One example is Jev, a newer company that helps choose between different AI outputs more cheaply than traditional approaches.

Other competitors are adding pressure. Meta and Nvidia have introduced affordable enterprise offerings and open-source models.

Companies from China and the United States are also releasing cheaper alternatives. This abundance of options means it has become a buyer's market, where customers have the power in negotiations.

Some investors hope new use cases will drive growth again. AI coding emerged about 18 months ago and helped Anthropic and OpenAI become trillion-dollar companies.

Personal agents are now being released by OpenAI, Meta, and others. However, industry experts are not confident personal agents will provide the same boost as coding did.

Anthropic is preparing for an initial public offering, but the slowdown in AI spending is affecting its expected value. The company's valuation has fallen from around $2 trillion in early October to $1.2–1.3 trillion.

OpenAI also lowered its valuation from $1.4 trillion to $1.25 trillion and reduced its fundraising plans.

The fundamental question remains: companies can use more AI for less money, which is good for customers but challenging for AI providers who need to grow their revenue.