New Fronts in the AI Trade

Dear Colleagues: Today’s letter includes my monthly update of Ramp AI Index, our flagship research using spend data from Ramp to track how American businesses are using AI. My last series of posts, Cracks in the AI Thesis, highlighted ways in which assumptions of rapid growth in AI spend are starting to break down. In this post, I describe new fronts in the AI trade.

Extras from Ramp data:

• I’m in Lisbon for Web Summit! I’ll be on stage with Luke Vargas of WSJ to discuss these results live (podcast to come) EDIT: this got cancelled because of Storm Charles, I'm still in Lisbon but we're recording remotely...
• Plain English with Derek Thompson: I joined Derek Thompson’s podcast for the episode: The AI Story Wall Street Doesn’t Want to Hear. Listen on YouTube, Apple Podcasts, and Spotify.

AI spend up 1.65% in October, and headwinds persist.

AI spending rose to a new peak in October, rising 1.65% from September. Spend also increased among the top 1% of spenders. The top 1% of firms now spend a median $10.2K per employee per month on AI, up 9%. Firms in the top 10% spent $801, roughly flat. The median firm spends $16.32, up $0.93.

The risks I identified in the Cracks series remain.

1. AI spend is highly concentrated. 1% of businesses drive 80% of the revenues to OpenAI and Anthropic. That top 1% is also highly correlated, largely made up of VC-backed companies in tech, finance, and technical services.

2. AI spend is failing to proliferate outside of those sectors. Growth in the median and top 10% groups is relatively sluggish.

3. Prices for AI models continue to fall. The average price per million tokens fell 21% over the last month. New entrants will accelerate price declines. Token volume growth continues to outpace token spend as AI gets cheaper and cheaper.

4. Most importantly, spend by engineering teams, which remains the largest enterprise AI spend category, stutters along, despite rising usage.

Today, I see new fronts for the AI trade: Some areas for growth and also some areas with greater competition. I will focus on two of them.

OpenAI vs. Anthropic October data: Single-digit growth in AI spend by businesses

In October, AI token spend grew 2.1% month-over-month for OpenAI and 0.5% for Anthropic.

New Front #1: AI goes beyond coding, new enterprise use-cases power reacceleration in AI spend. Will it be enough?

Today, we introduce a new dataset to Ramp AI Index. We can now track AI spend by department. This new metric allows us to track which teams are driving net-new spend. We expect it will help us catch new use-cases as AI proliferates outside of technical work.

In our first result, we see one of the clearest challenges facing AI companies: driving additional spend and ROI beyond the coding agent use-case that has made Engineering teams the dominant customer for AI products and services. In October, Engineering teams drove 72% of all classified spend, but spend has fallen from highs we observed earlier this summer.

So what is driving modest increases in spend? Since this summer, it looks like customer success and support operations, possibly deploying models for customer service chatbots, are driving a modest reacceleration in AI spend, now driving 13.3% of all classified spend, up from 0.8% in June. We’ve also seen modest growth by executive leadership teams using AI.

At the same time, spending by customer service departments has not reached the kind of hypergrowth we saw with engineering early this year. If it gets there, it’s not clear whether it, alone, can drive outsized gains to OpenAI and Anthropic. These use cases are a good candidate to rely on cheaper and faster models, rather than the highest-margin and highest-performance models (Opus, Fable, Sol, Astra).

Fast and cost-effective models will see greater demand as AI proliferates outside of the engineering use case. Consider it a new front in the AI trade.

One area of reprieve: frontier models’ share of spend is up slightly. Opus, Fable, Sol, and Astra took 69.5% of spend in October, up from 67.7%.

Department-level data will soon be available for weekly tracking on the Ramp AI Index data dashboard.

New Front #2: 3-way race as SpaceXAI posts rapid gains in enterprise adoption.

In October, SpaceXAI led breakout growth in business AI adoption. Among Ramp businesses, 6.3% paid SpaceXAI, up 2.2 percentage points from last month. Anthropic adoption rose 1.1 percentage points to 44.8%, while OpenAI adoption held flat at 40.6%. Overall adoption reached 56.8%, increasing 1 percentage point, slightly faster than September’s 0.66 point gain.

Markets recently punished SpaceXAI following the Democratic sweep last week, given Elon’s relationship with the Trump administration. I think this may have been an overshoot. We have no evidence that business adoption of SpaceXAI’s products and services will flag in a post-Trump political environment. In fact, we recently shared that SpaceXAI is the fastest growing software company in Ramp’s entire dataset.

I argued in my post that SpaceXAI could be the next Anthropic and rapidly rise the charts in business adoption. It’s already following a similar path. Strong execution on the coding use-case (Grok is to Claude Code) and a pivot to consumer with a popular agent that drives mindshare and enterprise adoption (Grokbot is to Claude Cowork). It also has an immense compute advantage.

Expect this two-way race to quickly turn into a 3-way race that puts greater price pressure on OpenAI and Anthropic.

For more analysis of Ramp spend data, follow me on X and LinkedIn, and check out Ramp Economics Lab.