AI spend stalls as price war creates buyer’s market

AI spending across venture‑backed tech firms has stalled, even as token usage climbed, according to a new Ramp report cited by TechCrunch. Token spend grew only 1.5% in August‑September after a 42% jump in July, while price cuts from OpenAI and Anthropic have turned the market into a buyer’s market.

What happened

• Ramp’s lead economist Ara Kharazian told TechCrunch that token spend was flat across August and September, rising just 1.5% versus a 42% rise in July.
• The top 1% of AI‑heavy firms slipped below the $10,000 per‑employee‑per‑month peak they hit in summer, while median firms now spend $12‑$15 per employee per month and the top 10% spend $600‑$1,000.
• October data showed an 8.9% token‑spend increase from September 23 to October 23, but overall growth remains constrained.
• OpenAI halved API prices for GPT‑6 Sol and GPT‑6 Luna; Anthropic cut pricing for Fable 5.1 and Opus 5.5, a move that a filing says shaved about $600 million from its revenue growth.
• Bloomberg reported Anthropic’s valuation fell to $1.2‑$1.3 trillion from an early‑October $2 trillion estimate, and OpenAI’s valuation slipped to $1.25 trillion from $1.4 trillion.
• Kharazian attributes the slowdown to fierce price competition and smarter customer budgeting rather than a demand shortfall.

Why it matters for growth teams

For small B2B growth teams, AI tools are a core part of lead generation, prospect enrichment, and email automation. The data shows you can run more token‑intensive workflows without a proportional cost increase, but only if you shift to cheaper models or enforce usage limits. Ignoring the buyer’s‑market dynamics could leave you paying premium rates for tasks that a $12‑$15 per‑employee solution—or even a lightweight open‑source model—could handle just as well. Understanding the pricing landscape lets you stretch limited budgets while still leveraging AI’s productivity boost.

What to do this week

1. Pull your AI‑tool invoices and calculate spend per employee; compare it to the $12‑$15 median benchmark and flag any outliers.
2. Pilot a cheaper model or a simple Yes/No service (e.g., Jev‑style outputs) for low‑complexity use cases like triaging leads or basic sentiment checks.
3. Set hard usage caps or budget alerts in your AI platforms and negotiate enterprise discounts based on actual volume rather than projected spend.

Sources

• Ramp data shows AI spending stalling just as Anthropic nears its IPO — TechCrunch - AI