1% of firms drive 80% of AI spend — and other founder signals

3D rendered ai text on dark digital background

Ramp’s October AI Index contains a number that should make any AI-revenue business uncomfortable: just 1% of firms in Ramp’s data account for 80% of spending on OpenAI and Anthropic. That is an unusually fragile revenue base. Per-employee spending in September came in 3.1% below July’s peak, even as total token volumes rose. Ramp attributes the gap to falling model prices. If your revenue tracks usage, the index recommends watching what share comes from your largest accounts and whether their spending is actually recovering before you forecast growth from rising token volumes.

The Model Price War Is Real, But Benchmarks Mislead

Five frontier AI models now score within five points of each other on an intelligence benchmark, according to analysis from Tom Tunguz. The cost to run those models varies by 3.8 times. Google’s Argon is cheapest at $1.99 per task. The comparison includes billed reasoning tokens and tests every model at maximum reasoning effort. A similar score does not mean a similar bill. The practical advice: run your actual application requests through each candidate at your planned settings before picking a default model.

person wearing suit reading business newspaper

Cybersecurity Pulled $6.8B in Q3 2026

Cybersecurity startups raised $6.8 billion in Q3 2026, a 50% increase from the previous quarter. Most of it went into larger rounds. Keyfactor raised $1 billion and Cyera raised $400 million, both focused on AI for identity and data security. On the M&A side, Visa acquired BioCatch for $2.4 billion.

Creator Spend, Outbound Teams, and a Price Cut That Raised Your Bill

Cal AI paid UFC creators thousands of dollars per post and saw no download spike. The co-founder blames audience mismatch: a UFC viewer doesn’t necessarily want to track calories. The team now checks comments for people actively asking for help the app provides, and caps payment per thousand views at measured revenue per thousand views.

Rippling’s outbound team, built in 2022, reportedly lifted conversion from roughly 1% to between 3% and 7% by replacing automated email sequences with people working the same accounts. Reps are paid for meetings held with decision makers. Account executives don’t prospect. Managers supervise four or five reps each.

On the pricing side: in a test of 191 reading-assistant requests, a 33% cut in Sonnet’s token prices produced a 4% higher bill with thinking off, because a changed tokenizer and longer answers increased billed tokens.

Two Signals Worth Filing

  • Anthropic’s daily merges rose from roughly 2,500 in May to more than 30,000 in September. That made source control the biggest development bottleneck, leading Anthropic to build Scion, a git replacement for its monorepo.
  • In Hampton’s survey of 250 founders and CEOs, 47% paid a functional leader more than their own base salary. None of the 58 venture-backed respondents reported more than $300,000 in compensation beyond salary.
Stay on top of AI & Automation with BizStack Newsletter
BizStack  —  Entrepreneur’s Business Stack
Logo