OpenAI Offers Discounts of Up to 90% to Win Anthropic’s Biggest Customers

OpenAI is stepping up its competition with Anthropic by offering aggressive discounts on AI models, coding tools and enterprise contracts, seeking to attract businesses that have relied heavily on Anthropic’s Claude products. According to reporting by The Information, summarized by The Decoder on October 25, 2026, companies including Rippling, Replit and DataCamp have shifted some of their AI workloads or spending toward OpenAI after receiving more attractive pricing or commercial offers.

The strategy comes as companies increasingly use AI models for software development, internal operations and customer-facing products. For these businesses, the cost of processing millions of requests can significantly affect profit margins. OpenAI’s reported discounts—alongside improvements in its models and infrastructure—are giving enterprise customers more reasons to reconsider which AI providers power their applications. However, the headline figure of 90% refers to reported customer cost reductions in particular circumstances, not a universal 90% discount offered to every customer.

Key Takeaways

• Aggressive pricing: OpenAI is reportedly offering substantial discounts on its models and coding products to win enterprise customers from Anthropic.
• Rippling shifts workloads: The company’s CEO said OpenAI’s share of its coding workload rose from 16% to 52%, while Anthropic’s share fell from 57% to 18%.
• Replit’s reported deal: OpenAI reportedly offered a discount exceeding 30% on a contract valued at more than $100 million annually.
• DataCamp’s savings: Its CEO said pricing changes could reduce costs by 80% to 90% in the relevant workflow.
• Enterprise incentives: OpenAI has reportedly expanded access to ChatGPT for some customers without increasing their existing contract costs.
• Profitability trade-off: Lower prices may help OpenAI gain market share but could also put pressure on margins.
• Competition intensifies: Anthropic, OpenAI and lower-cost Chinese AI providers are competing on model quality, usage costs and enterprise contracts.

Why OpenAI Is Offering Deep Discounts

OpenAI and Anthropic compete for many of the same enterprise customers, including software companies that rely on large language models to power coding assistants, research tools and business applications.

Anthropic gained considerable traction among developers through Claude and Claude Code, particularly for complex programming tasks. As businesses adopted these tools, some began directing a growing share of their AI budgets toward Anthropic.

OpenAI is now trying to win back that spending through a combination of lower prices, improved models, coding tools and more flexible enterprise agreements.

The pricing strategy is not limited to discounts on individual API requests. Reported offers include reduced model prices, expanded access to ChatGPT seats and commercial terms tied to customers using OpenAI’s Codex coding tool.

The approach matters because enterprise AI purchasing decisions increasingly involve total operating costs rather than model performance alone. A business may choose a slightly different model if it can complete its work reliably at a substantially lower cost.

How Much Are OpenAI’s Customers Saving?

The reported deals vary considerably, so it is important to distinguish between a discount on the listed price and a reduction in the customer’s total spending.

Company or customer	Reported development	What it indicates
Rippling	OpenAI’s share of coding workload rose from 16% to 52%; Anthropic’s fell from 57% to 18%	A major shift in the mix of AI models used for coding
Replit	OpenAI reportedly offered more than 30% off a contract worth over $100 million annually	Enterprise pricing can influence large purchasing decisions
DataCamp	CEO Jonathan Cornelissen said costs could fall by 80% to 90%	Lower model prices and changes in the AI workflow can produce substantial savings
A US bank	ChatGPT access reportedly expanded to 500 employees without increasing the contract price	Bundled access can lower the effective cost per employee

Source: The Information, as summarized by The Decoder and Explainx.ai. These figures are attributed reports, not independently audited company-wide results.

The examples show that OpenAI’s commercial strategy takes several forms. Some customers are receiving lower prices, while others are changing which model handles their workload or negotiating additional products within existing contracts.

The distinction matters because an 80% reduction in total costs could result from a combination of cheaper models, better efficiency, lower token consumption and different usage patterns—not simply an 80% reduction in the price of every token.

Rippling Shifts Its AI Coding Workload

Rippling, an HR and business software company, provides one of the clearest examples of how pricing competition can change AI adoption.

According to CEO Parker Conrad, Anthropic models handled 57% of the company’s coding workload in May 2026, compared with 16% for OpenAI. The reported mix later changed to 52% for OpenAI and 18% for Anthropic.

The figures suggest that OpenAI gained a substantially larger role in Rippling’s coding work over the period.

However, workload share is not the same as market share or total spending. The figures do not establish exactly how much money Rippling saved, whether the total volume of coding work changed or how the two providers performed across every task.

For AI providers, the commercial implication is significant: large customers can redirect workloads when they find another model that meets their quality requirements at a lower effective cost.

Replit’s Reported $100 Million Contract Offer

Replit, a software development platform that uses AI to help users build applications, has also been cited in the reporting.

According to The Information, OpenAI offered Replit a discount of more than 30% as part of a contract worth more than $100 million annually. The offer reportedly came after disagreements over Anthropic’s willingness to extend an existing discount without a higher spending commitment.

Replit subsequently moved the majority of its AI spending toward OpenAI, according to the report. The account should be treated cautiously: the reported offer is not the same as independently verifying every element of a finalized contract, and public descriptions of the companies’ relationship have not always been consistent.

Nevertheless, the case illustrates how enterprise AI contracts can become a competitive battleground. At this scale, even a relatively modest percentage discount can represent millions of dollars in annual spending.

Why DataCamp Reported Cost Reductions of Up to 90%

DataCamp, a workforce training and education company, has also been evaluating OpenAI’s pricing.

CEO Jonathan Cornelissen said the company was considering an OpenAI offer that included a 40% discount on models used internally to develop products. The reported terms also tied the discount to using OpenAI’s Codex instead of competing coding tools.

Cornelissen described the potential savings as an 80% to 90% reduction in costs. That figure should not be interpreted as a blanket discount on every OpenAI model. It reflects the company’s reported cost comparison and the specific products and workflows involved.

The episode demonstrates why the headline percentage can be misleading without context. A business that switches to a cheaper model, reduces the number of tokens required for a task or changes how its software uses AI may see its total bill fall far more than the nominal discount alone.

For software companies whose products depend on frequent AI calls, these savings can improve gross margins and make it easier to offer competitive prices to their own customers.

Anthropic’s Discount Policy Creates an Opening

OpenAI’s offers come amid differences in how the two companies reportedly manage enterprise pricing.

A separate The Information report published in September 2026 said Anthropic generally requires customers that exhaust the discounted usage covered by their spending commitments to negotiate new terms or face higher prices.

OpenAI has reportedly taken a more flexible approach, allowing some customers additional time to negotiate after reaching their commitment limits.

For businesses with rapidly growing AI usage, these differences can matter as much as the headline price. A contract that looks inexpensive initially may become less attractive if the customer repeatedly exceeds its included usage.

However, stricter contract terms do not automatically mean Anthropic’s products are more expensive in every situation. The actual cost depends on model choice, volume, negotiated terms, reliability and the amount of work each model completes.

Anthropic also remains a significant competitor. Model quality and developer preferences can change quickly, and a customer that moves workloads toward OpenAI today could shift them again if another provider delivers better results or pricing.

What the Discounts Mean for OpenAI’s Profitability

Aggressive pricing can help OpenAI attract customers, increase usage and build longer-term relationships with enterprise buyers. But discounts can also reduce the revenue earned on each unit of AI consumption.

The trade-off depends partly on whether OpenAI’s newer models and infrastructure are cheaper to operate. If efficiency improvements reduce the cost of serving a request, OpenAI may be able to lower prices without sacrificing the same amount of margin that an equivalent discount would have required previously.

The company’s reported pricing strategy therefore appears to combine two elements: using commercial incentives to win customers and making its offerings more cost-efficient.

Still, public information does not provide a complete picture of the profitability of the reported contracts. Neither the full terms of individual deals nor the underlying cost structures are publicly available for all the examples.

Chinese AI Models Add Further Price Pressure

OpenAI and Anthropic are not competing only with each other. Chinese AI developers, including DeepSeek and Moonshot AI, have also increased competitive pressure by offering models at lower prices.

For companies that can switch between providers with relatively little engineering work, lower-cost alternatives create another negotiating option. Some businesses can route different tasks to different models, using a more expensive model for difficult work and a cheaper one for simpler requests.

This makes the market increasingly sensitive to the cost of completing a task rather than the price of a single token. A model that uses fewer tokens or requires fewer retries can be cheaper in practice even if its published token price is higher.

For OpenAI, aggressive enterprise pricing is therefore part of a wider contest over the economics of AI deployment.

The Bigger Picture

The reported discounts illustrate how the AI industry is moving from a race focused primarily on model capabilities toward a more complex competition involving pricing, distribution, developer tools and enterprise contracts. As AI becomes embedded in everyday software, the provider that offers the right combination of performance, reliability and cost can influence where companies direct large amounts of spending.

The strategy also carries risks. Deep discounts can make customer acquisition more expensive, and businesses may switch providers again when contracts expire or better models emerge. OpenAI’s ability to retain customers will depend on whether the quality and productivity benefits justify continued adoption after introductory incentives end.

FAQs

1. Is OpenAI offering every customer a 90% discount?

No. The reported 80% to 90% figure refers to cost reductions described by DataCamp’s CEO for particular circumstances. Other examples involve different discounts, contract terms and product bundles.

2. Which companies are reportedly shifting AI spending toward OpenAI?

The reporting names Rippling, Replit and DataCamp among companies shifting some workloads or spending toward OpenAI. The extent of each change differs, and not all contract details are publicly confirmed.

3. Why is OpenAI discounting its AI models?

OpenAI is seeking to attract enterprise customers, increase usage of its models and coding tools, and compete with Anthropic. Lower prices can also reflect improvements in model and infrastructure efficiency.

4. Is Anthropic losing its entire enterprise customer base?

No. The reports describe shifts in spending and workload at particular companies, not the disappearance of Anthropic’s customer base. Anthropic remains a major provider, and customers can use models from multiple companies simultaneously.

5. Could the price war make AI cheaper for businesses?

It could. Competition can reduce prices and encourage providers to improve efficiency. However, actual savings depend on model performance, usage volume, contract conditions and the amount of work needed to complete a task.

Looking Ahead

The next stage of competition will depend on whether OpenAI can convert discounted contracts into durable customer relationships. Enterprises will likely continue comparing total costs, model quality, usage limits and support before deciding how much of their AI workload to allocate to each provider. Publicly reported workload shifts offer useful signals, but they do not reveal the complete economics of individual contracts.

For Anthropic, retaining customers will require balancing pricing policies with the strength of its models and developer products. For OpenAI, the challenge is to win more enterprise spending without allowing discounts to undermine profitability. As cheaper models and more efficient AI systems enter the market, businesses are likely to gain more negotiating power—and the cost of delivering useful AI work may become as important as model capability itself.