TECH NEWS – The Redmond tech giant has so far used OpenAI and Anthropic models, and now it is preparing to switch to China’s DeepSeek V4!
Microsoft is considering using a self-hosted version of DeepSeek V4 for Copilot Cowork, because OpenAI and Anthropic appear to be setting prices that are pushing themselves out of the market. A token is the smallest unit of data processed by an AI model, and it is usually around four characters long. A model’s context window, or working memory, is measured entirely in tokens, and this includes both the tokens used to understand the input prompt and those used to generate the relevant output. As model complexity and agentic workloads increase, however, token costs become a critical blocking factor, especially for coding-related tasks, where token consumption can unexpectedly skyrocket.
🚨BREAKING: ANTHROPIC SUED FOR FRAUD
A customer filed a federal class-action lawsuit Monday claiming he hit 15% of weekly allowance in just one 5-hour sprint.
“The actual usage provided by the Max 5x and Max 20x plans is far below the advertised amount.”
They are calling it… pic.twitter.com/342T14ZZxl
– NIK (@ns123abc) June 15, 2026
NEW: Inside the 24-hrs before WH slapped export controls on Anthropic
– Last Thursday, Amazon CEO Andy Jassy raised concerns about Fable jailbreak to Trump admin
– Friday AM, Sean Cairncross, Bessent, Susie etc. held WH call to discuss
– Then White House started reaching out…– Sophia Cai (@SophiaCai99) June 13, 2026
Who could forget that Uber burned through its entire 2026 AI budget in four months after encouraging employees to use AI more heavily? In fact, employees interested in the technology now use the term tokenmaxxing to describe using AI models for menial tasks, packed with long prompts and agent loops, to reach first place on internal AI-usage leaderboards. OpenAI and Anthropic are not only raising the prices of their enterprise plans, but also using creative methods to limit the total number of usable tokens.
Microsoft will likely choose DeepSeek V4 or a similar model, though running on its own infrastructure, to power Copilot Cowork. This service combines Copilot’s enterprise functions with advanced AI models from OpenAI and Anthropic, making agent-like, AI-based workflows relatively simple. The change was prompted by Microsoft’s decision to move Copilot Cowork to a consumption-based architecture, where users pay for the total number of tokens used instead of a flat fee.
🚨BREAKING: DeepSeek just raised $7.4 billion at $50 billion+ valuation
CEO Liang Wenfeng wrote the biggest check himself: $2.8B
>Tencent: $1.4B
>CATL: $700M
>JD, NetEase, IDG Capital: $420M each
>China’s National AI Fund: $140MInvestors put money into an LP managed by CEO… pic.twitter.com/PvqCL8Enfd
– NIK (@ns123abc) June 16, 2026
DeepSeek is going heavy-asset.
On June 9, the company posted an opening for IDC planning engineers, a role explicitly scoped to the design and delivery of MW-to-GW scale infrastructure. It follows April’s hiring of data center O&M engineers in Ulanqab, Inner Mongolia. Taken… pic.twitter.com/1SDBKUow26
– SemiAnalysis (@SemiAnalysis_) June 10, 2026
Even though DeepSeek’s models are based on open-source architecture, such a move is unlikely to receive a positive response in the White House, where Anthropic was recently forced to withdraw its Mythos-class Fable 5 model from all users who are not U.S. citizens after Amazon allegedly disclosed a method for breaking into the model’s advanced cyber capabilities. Some, however, speculate that the move will make life more difficult for China-based AI players, who have elevated the “distillation” of American models almost to an art form.
All of this is happening as DeepSeek has just raised $7.4 billion at a $50 billion valuation and is moving full steam ahead to expand its compute capacity.
Source: WCCFTech, Fortune, Axios, Firstpost
