Anthropic profit forces OpenAI price cuts

About this episode

AI pricing is changing fast. OpenAI, Anthropic, and Microsoft's GitHub are all moving away from flat-rate subscriptions toward usage-based billing, and the shift is going to hit anyone whose business runs heavily on AI tools. Anthropic has already shifted some business customers to actual-usage billing. GitHub launched a new usage-based system that kicks in after monthly allotments run out. OpenAI executives have publicly floated pricing AI more like electricity or water, where heavier users pay more for slide decks, longer agent runs, code debugging, and email drafting.

This episode breaks down the AI pricing shock hitting OpenAI, Anthropic, and Microsoft, what it means for businesses already building on these tools, and which alternatives are starting to look attractive. The driver is straightforward. AI labs are burning cash on chips, data centers, and talent at a rate that flat-rate subscriptions can't support. OpenAI reported a $14 billion projected loss for 2026. Anthropic just filed for IPO at a $965 billion valuation. Microsoft is spending tens of billions on AI infrastructure. The math on a $20-a-month subscription that produces unlimited GPT-5 output doesn't work anymore.

The corporate response is already visible. Walmart capped staff use of its in-house AI agent. Uber is limiting monthly employee spending to $1,500 per AI coding tool. Companies that rolled out generative AI broadly in 2024 and 2025 are now reading the meters, because the same prompt that cost $0.02 in 2024 can cost $2 today on a reasoning model.

The lower-cost alternatives are gaining real attention. Alibaba's Qwen and DeepSeek both run at a fraction of OpenAI and Anthropic pricing, and both have closed the quality gap enough that routing simpler tasks to a cheaper model is a defensible engineering decision. The question for every business spending on AI is which tasks need a frontier model and which can run on a model that costs 90% less for the same output.

What this means for AI strategy in 2026. Flat-rate pricing was a customer acquisition tactic that worked when the labs were trying to win mindshare. Usage-based pricing reflects what AI actually costs to deliver, and it's the model the industry will settle on. For developers, freelancers, and small businesses using ChatGPT, Claude, GitHub Copilot, and Cursor every day, the bill is about to look different. For agencies and consultants billing clients for AI work, the margin model needs a rebuild.

We cover the OpenAI, Anthropic, and GitHub pricing changes in detail, how Walmart and Uber are responding, why Qwen and DeepSeek matter more this quarter than they did last quarter, and what the shift to electricity->

Keywords: AI pricing, OpenAI pricing, Anthropic billing, GitHub Copilot pricing, usage-based AI, token pricing, AI subscription, ChatGPT pricing, Claude pricing, Qwen, DeepSeek, Walmart AI, Uber AI, GPT-5 cost, AI ROI, AI infrastructure cost.