Microsoft Claims Copilot Beats Claude and ChatGPT on Price

Reading Time : 3 minutes

Microsoft is making cost the centerpiece of its case against Anthropic and OpenAI. At its Ignite conference, Microsoft commercial chief Judson Althoff told customers that anything they can do in Claude or ChatGPT, they can do better with Copilot and for less. The statement arrives as companies increasingly scrutinize AI spending, especially as usage-based billing can turn a modest subscription into a much larger monthly bill. Microsoft’s message is simple: its assistant is not merely competitive on capability, but materially cheaper to operate at enterprise scale.

 

 

The evidence is a study Microsoft commissioned and funded with GigaOm, based on hands-on testing in October. Microsoft ran the testing and cost analysis, while GigaOm helped design the research, validate the methodology, and co-develop the findings. The study modeled 100 monthly tasks for an everyday user and 140 for an advanced user, including 20 agentic workflows, across chat, Word, Excel, PowerPoint, and agent tools. It concluded that Copilot cost $30 per month for everyday use, versus $81 for ChatGPT Enterprise and $205 for Claude Enterprise. For advanced users, the figures were $96, $149, and $299 respectively.

 

 

Those numbers are striking, but they are modeled estimates rather than audited invoices. Microsoft says the calculations use October list prices, exclude negotiated discounts, and reflect a defined set of tasks and configurations. The company also acknowledges that results varied by task and should be read as directional rather than universal. The cost advantage largely comes from Copilot’s subscription covering everyday chat and Microsoft 365 usage, while Auto selects models and effort levels to manage the cost of each request. In other words, Microsoft is selling predictability: a flat subscription for routine work, with usage charges reserved for heavier agentic tasks.

 

 

The dispute is not settled. Anthropic says its non-coding Claude Enterprise users spend under $75 a month on average, far below the $205 figure in Microsoft’s modeled scenario. That does not necessarily disprove Microsoft’s result, because the two figures measure different things: one is an average across existing customers, while the other is a constructed workload with repeated prompts and specific product settings. Still, the gap is large enough to raise an obvious question about whether Microsoft’s comparison reflects typical enterprise use or a workload that favors its own pricing model. For buyers, the real test is a controlled pilot using their own tasks, volumes, security requirements, and negotiated rates. Until independent evidence emerges, Microsoft’s price claim is credible in structure but still unproven as a universal business outcome.

 

Bénédicte Lin – Brussels, Paris, London, Beijing, Seoul, Bangkok, Tokyo, New York, Taipei, Hong Kong
Bénédicte Lin – Brussels, Paris, London, Beijing, Seoul, Bangkok, Tokyo, New York, Taipei, Hong Kong

 

#Microsoft #Copilot #Claude #ChatGPT #AI #EnterpriseAI #AICosts #Anthropic #OpenAI #MicrosoftIgnite #GenerativeAI #AIBusiness