Anthropic's Enterprise Push

Anthropic is moving Claude into the core of big consulting firms just as some developers say the company’s coding tools have become harder to use. (anthropic.com) On December 9, 2025, Anthropic and Accenture said they were forming an Accenture Anthropic Business Group and training about 30,000 Accenture professionals on Claude. The companies also said tens of thousands of Accenture developers would get Claude Code. (anthropic.com) On October 6, 2025, Anthropic said Deloitte would make Claude available to more than 470,000 people across its global network and train and certify 15,000 professionals on the system. On November 4, 2025, Anthropic said Cognizant would deploy Claude to up to 350,000 employees globally. (anthropic.com 1) (anthropic.com 2) Those deals put Claude inside the firms that design technology programs for large companies, banks, hospitals, and government agencies. Accenture, Deloitte, and Cognizant all said the work would focus in part on regulated industries and on moving customers from pilot projects into production systems. (accenture.com) (anthropic.com) Anthropic is also tying Claude to coding work, not just chatbots. Accenture said the partnership included a joint offering for chief information officers to scale artificial intelligence-powered software development, and Cognizant said it would use Claude Code for coding, testing, documentation, and DevOps workflows. (accenture.com) (anthropic.com) At the same time, developers using Claude Code have been arguing that long sessions now burn through paid limits faster. The Register reported on April 13, 2026 that Anthropic cut the prompt-cache time to live for many requests from one hour to five minutes around March 7 after introducing the longer cache around February 1. (theregister.com) Prompt caching is a way to avoid reprocessing the same background material every time a user sends a request. The Register reported that writing to a five-minute cache costs 25 percent more in tokens than base price, writing to a one-hour cache costs 100 percent more, and reading from cache costs about 10 percent of base price. (theregister.com) Anthropic disputed the idea that the cache change was driving higher usage bills. Jarred Sumner of Anthropic told The Register that the shorter cache made Claude Code cheaper for many one-shot requests, while Claude Code creator Boris Cherny said the company was investigating a 400,000-token default context window instead of one million tokens to reduce expensive cache misses. (theregister.com) A separate wave of complaints has focused on model quality, with users alleging that Claude responses have become less capable or less reliable. VentureBeat reported on April 14, 2026 that Anthropic has been tuning settings across user segments and that those changes could affect how people perceive performance even if the underlying model weights have not changed. (venturebeat.com) The split is now visible across Anthropic’s business. The company is selling Claude as a large-scale, compliance-focused system for consulting firms and enterprise clients while answering developer complaints about quotas, cache behavior, and whether the product feels different from week to week. (anthropic.com) (theregister.com)