SaaStr - OpenAI Compute Margin Analysis

web

saastr.com·saastr.com/have-ai-gross-margins-really-turned-the-corner...

Industry blog post relevant to understanding AI compute economics and cost trends; useful background for analyzing deployment sustainability and the economic pressures shaping which AI capabilities get developed and deployed commercially.

Metadata

Importance: 32/100blog postanalysis

Summary

Analyzes OpenAI's reported compute margin improvement from 35% to 70% between early 2024 and October 2025, arguing that while foundation model economics are improving, B2B AI startups face a 'treadmill problem' where cost savings on older models are offset by rising costs of frontier models and agentic workflows consuming 10x-100x more tokens.

Key Points

•OpenAI's compute margin doubled from ~35% in early 2024 to ~70% by October 2025, approaching traditional SaaS gross margin territory.
•Inference cost declines are occurring on older models, while frontier models are becoming more expensive due to multi-step reasoning tasks.
•Agentic workflows have caused token consumption per task to jump 10x-100x since December 2023, dramatically increasing per-task costs.
•Anthropic went from negative 94% gross margins in 2024 to projecting 50% in 2025 and 77% by 2028, showing rapid improvement at the foundation layer.
•B2B startups building on top of frontier models may not benefit from headline margin improvements as they remain dependent on expensive, improving-but-costly APIs.

Cited by 1 page

Page	Type	Quality
Anthropic Valuation Analysis	Analysis	72.0

Cached Content Preview

HTTP 200Fetched Mar 20, 202620 KB

# Have AI Gross Margins Really Turned the Corner? The Real Math Behind OpenAI’s 70% Compute Margin — And Why B2B Startups Are Still Running on a Treadmill

by [Jason Lemkin](https://www.saastr.com/author/jasonlkn/ "Posts by Jason Lemkin") \| [Artificial Intelligence (AI)](https://www.saastr.com/category/topics/ai/), [Blog Posts](https://www.saastr.com/category/resource-type/blog-posts/), [SaaStr.Ai](https://www.saastr.com/category/saastr-ai/)

OpenAI (or at least, **[The Information reporting on OpenAI)](https://www.theinformation.com/articles/openai-getting-efficient-running-ai-internal-financials-show)** just dropped a bombshell: their “compute margin” — the share of revenue left after paying for the massive server costs to run ChatGPT — hit 70% in October 2025.

In January 2024, that number was 35%. They’ve essentially doubled their margin efficiency in less than two years.

> OpenAI’s compute margin on paid products is now about 70% as of October which is roughly double early 2024 levels.
>
> Looks like “Compute margin” will be Sam Altman’s favorite way to frame profitability. [pic.twitter.com/4rGCxDVdbk](https://t.co/4rGCxDVdbk)
>
> — Shay Boloor (@StockSavvyShay) [December 22, 2025](https://twitter.com/StockSavvyShay/status/2003076837640151447?ref_src=twsrc%5Etfw)

If you’re a B2B founder or investor, you know exactly why this matters. A 35% gross margin is a services business. A 70% gross margin is starting to look like… software.

So the question everyone’s asking: **Have AI gross margins actually turned the corner? Or is this just creative accounting on a burning pile of cash?**

And the harder question for B2B founders: **Does any of this actually help you?**

The honest answer: probably not as much as you’d hope.

Let’s dig in.

## The Headline Numbers Look Great. The Reality Is More Complicated.

Traditional SaaS is beautiful from a unit economics standpoint. You build the software once, host it cheaply, and the marginal cost of each new customer approaches zero. That’s how you get to 75-80% gross margins that make investors swoon.

AI blew that up.

Every single AI inference — every ChatGPT response, every Copilot code suggestion — burns actual compute. GPUs aren’t cheap. Electricity isn’t cheap. And when you’re processing billions of queries, the math gets ugly fast.

Here’s what we were looking at in late 2023 and early 2024:

**OpenAI:** ~35% compute margins (early 2024) **Anthropic:** Negative 94% to negative 109% gross margins in 2024 (yes, negative — they lost more on infrastructure than they made in revenue) **GitHub Copilot:** Microsoft reportedly losing $20+ per user per month at the $10/month price point

And now? OpenAI’s compute margin jumped from around 35% in January 2024 to roughly 70% by October 2025. Anthropic expects its gross profit margin to reach 50% this year and 77% in 2028, up from negative 94% last year.

So margins are improving at the foundation layer. That’s genuinely good news.

But here’s the catch that 

... (truncated, 20 KB total)

Resource ID: 580b6a23c75d3817 | Stable ID: MTRlYjg2Mz