OpenAI Undercut Claude on Price, Took 52% of Rippling’s Code

OpenAI’s efforts to undercut Anthropic on price are producing some results, The Information reported on 24 October 2026. At Rippling, an HR software firm, Anthropic models powered 57% of the coding work of hundreds of engineers in May and OpenAI models 16%; now OpenAI handles 52% and Anthropic 18%, Rippling CEO Parker Conrad told the publication.

Conrad gave the reason in cost terms: the company sought to cut costs and OpenAI proved cheaper, according to The Information. Only the opening of the article is free to read; the rest is behind a paywall, and this piece uses only what we could read.

Business Pill · SWITCHING COSTS

A one-minute Business Engineer explainer of switching costs: why it can be easy, or hard, to leave a supplier. It teaches the general idea only and says nothing about The Information’s report.

The key insight: As we read it, the number to watch is how fast the share moved: Rippling’s CEO told The Information that Anthropic models went from 57% of his engineers’ coding work in May to 18%, while OpenAI’s share rose from 16% to 52%, as the company sought to cut costs and OpenAI proved cheaper.

What The Information Reported

The article, by Laura Bratton and Stephanie Palazzolo and headlined “Anthropic’s Margin is OpenAI’s Opportunity”, opens with one line: “OpenAI’s efforts to undercut Anthropic on price are producing some results.”

It then gives Rippling as the example. In May, Anthropic models powered 57% of the coding work of hundreds of engineers at the company, while OpenAI models handled 16%. “Since then, the situation has reversed: 52% of the coding work is now handled by OpenAI models and 18% with Anthropic ones, as the company sought to cut costs and OpenAI proved cheaper, CEO Parker Conrad said.”

In plain terms, a company running AI coding tools for hundreds of engineers moved the majority share of that work from one model maker to the other within about five months, and its chief executive tied the move to price.

Share of Rippling’s coding workMay 2026Now (24 Oct 2026)Change
Anthropic models57%18%−39 points
OpenAI models16%52%+36 points

Source: The Information, 24 October 2026, citing Rippling CEO Parker Conrad. The change in points is our arithmetic. The readable text does not say what handles the rest of the work.

Share of the coding work of hundreds of Rippling engineers handled by Anthropic and OpenAI models, in May 2026
Share of the coding work of hundreds of Rippling engineers handled by Anthropic and OpenAI models, in May 2026 and now, as Rippling CEO Parker Conrad told The Information (24 October 2026). The readable text does not say what handles the rest. Source: The Information.

What Was Posted on X

An X post by @choblin29 summarising The Information’s article. Its Replit figures (more than 30% off on a contract above $100 million a year) also appear in Techmeme’s summary of the article; its 40% offer to DataCamp does not appear in the part of Techmeme’s summary we read, and we have not checked it.

Commentator kimmonismus reads the discounts as OpenAI gaining coding workloads at the cost of potentially deeper losses, and argues that Anthropic’s renewed lead on OpenRouter suggests better models can quickly win users back. Both readings are the poster’s.

@shanumathew93 posted on 25 Oct that OpenAI’s enterprise discounting “is moving real coding workloads,” which the poster calls “super important for closed lab P&Ls.” Citing The Information, the post repeats Rippling’s shift from 57% Anthropic and 16% OpenAI in May to 18% and 52%, and says Replit moved most of its AI spending to OpenAI.

Replit, Another Coding Customer

Replit, another coding customer, made a similar move. TechCrunch reported on 16 October that Replit chief executive Amjad Masad had said onstage at TechCrunch Disrupt that his company was no longer working with Anthropic, citing “cost, partnership, and performance”, TechCrunch reported. Masad later clarified that customers can still use Anthropic models on Replit, but that ChatGPT is now the default, according to TechCrunch.

“OpenAI is now our preferred partner for powering Replit Agent, based on performance, economics, and the strength of our partnership,” Masad told TechCrunch. He also said onstage that Replit had been, until recently, a top-five customer of Anthropic’s, TechCrunch wrote. A person familiar with the partnership told TechCrunch it was still in place.

Update, 24 Oct (evening): Techmeme’s summary of The Information’s article now carries the Replit figures. Citing a source, it says Replit “recently moved most AI spend from Anthropic to OpenAI, which gave 30%+ off in a deal worth $100M+/year, after Anthropic discontinued its discount.” We have not read that part of The Information’s text ourselves. A 40% offer to DataCamp appears only in an X post by @choblin29 summarising the article, and we have not checked it.

Update, 25 Oct: A passage of The Information’s article that we have now read adds two details. DataCamp is negotiating to apply the same discount to the OpenAI models behind an application it sells to its customers, which would bring “significant” savings and higher gross margins, The Information reports. “We’re looking at an 80% to 90% decrease in costs,” Jonathan Cornelissen, DataCamp’s co-founder and CEO, told the outlet. “It’s huge.” The article also says OpenAI recently allowed an American bank with more than $7 billion in assets to double its ChatGPT accounts to 500 workers without raising the annual contract cost of around $40,000, according to an executive at the bank. By our arithmetic, $40,000 a year across 500 accounts is about $6.67 per account per month. We read this passage from a screenshot of the article and have not seen the rest of its text.

Connecting the Coverage

Our Replit piece of 16 October reported Masad’s onstage account and the pushback from a person familiar with the partnership; the Rippling figures are the first company-level numbers we have seen on how far such a switch can go.

Our token-price piece reported YipitData’s note that “across our token-based datasets, we’ve observed a ~50% Y/Y decline in token prices,” and our token price paradox piece set that figure against seven Gemini prices that are set to step up.

Our retention piece carried YipitData’s finding that Anthropic’s B2B retention in its panel rises from 87% after month 12 to 95% after month 24, and our Ramp piece reported that 1% of firms drive 80% of OpenAI and Anthropic spend on Ramp.

The Structural Read

As we read it, the Rippling figures put a size on a switch that until now we had only heard described: Replit’s chief executive said cost was one of his reasons, and Rippling’s gave the share of work that moved.

As we read it, defaults matter here: TechCrunch wrote that Replit’s shift could significantly reduce its spend with Anthropic because most customers stick with the default rather than selecting a model themselves.

As we read it, The Information’s headline frames this as a margin question for Anthropic, and the only figures we could read are Rippling’s shares, not prices or discounts.

Rippling CEO Parker Conrad, as reported by The Information, 24 October 2026

“as the company sought to cut costs and OpenAI proved cheaper”

Three Implications

FOR CODING PLATFORMS Replit told TechCrunch that OpenAI is now its preferred partner “based on performance, economics, and the strength of our partnership.”

FOR MODEL MAKERS The Information’s article is headlined “Anthropic’s Margin is OpenAI’s Opportunity”; the free portion we read does not give either company’s prices or margins.

FOR BUYERS At Rippling the majority share of coding work went from Anthropic to OpenAI between May and October, The Information’s figures show.

The Business Engineer Lens

This story maps onto the Business Engineer framework The Agentic Harness War.

The framework puts it this way: “Model โ€” commoditizing relative to what sits above it.”

As we read it, Rippling’s figures fit that description: the majority share of its engineers’ coding work went from one model maker to the other within months. The framework was published in July 2026 and does not discuss Rippling.

What Is Not Established

We read only the free opening of The Information’s article. We do not know what discounts, if any, Rippling received, how The Information measured the share of coding work, or what handles the remaining work. The Replit contract value and the size of its discount come from Techmeme’s summary of The Information’s article and an X summary, not from text we read; the DataCamp offer comes only from the X summary. We did not find statements from OpenAI, Anthropic or Rippling, and we did not contact them.

This Week In The AI Supercycle

Execution Fault Lines, Concentrated Demand, and Geopolitical Firewalls

This week’s issue: the AI build is not short of demand for its parts. It is short of cheap money, and its end revenue rests on a narrow base of large, often venture-funded buyers.

Read the full issue →

Business Engineer Framework

The Agentic Harness War

A Business Engineer framework on why the agentic surface, not the model, is where value pools in the AI stack.

Read the Map of AI →

The Bottom Line

The Information reports that OpenAI’s lower prices are winning coding work from Anthropic, and Rippling’s chief executive gave the numbers: Anthropic’s share of his engineers’ coding work fell from 57% in May to 18%, while OpenAI’s rose from 16% to 52%. As we read it, the switch shows how quickly a large customer can move coding work between model makers when, as Conrad put it, the company is seeking to cut costs.

95,000+ executives read Business Engineer for the AI strategy frameworks cited by ChatGPT, Claude, and Perplexity.

A note on sourcing. This piece rests on the free opening of The Information’s article of 24 October 2026 (the rest is behind a paywall and was not read), TechCrunch’s report of 16-17 October 2026, Techmeme’s summary of The Information’s article (read 24 October 2026), an X post by @choblin29 summarising the article, and our own earlier pieces, all read on 24 October 2026. Figures that appear only in summaries are labelled as such. We did not contact OpenAI, Anthropic or Rippling.

Sources: The Information, 24 Oct 2026 · TechCrunch, 16-17 Oct 2026 · Techmeme, 24 Oct 2026 · @choblin29 on X, 24 Oct 2026

Scroll to Top

Discover more from FourWeekMBA

Subscribe now to keep reading and get access to the full archive.

Continue reading

FourWeekMBA