AI News Roundup: Anthropic’s $200B Google Cloud Deal, Cerebras IPO, and Pentagon AI Contracts

Anthropic Commits $200 Billion to Google Cloud in Landmark Five-Year Deal

Anthropic has agreed to spend $200 billion with Google Cloud over the next five years, according to a report from The Information published Monday. The staggering commitment would make Anthropic responsible for more than 40% of the revenue backlog Google disclosed to investors, sending Alphabet shares up roughly 2% in extended trading.

The deal follows an April agreement between Anthropic, Google, and chip partner Broadcom for multiple gigawatts of tensor processing unit capacity expected to come online starting in 2027. Separately, Google is investing up to $40 billion in Anthropic — $10 billion upfront at the company’s latest $380 billion valuation, with the remaining $30 billion contingent on performance milestones.

Anthropic’s run-rate revenue has now surpassed $30 billion, up from roughly $9 billion at the end of 2025. The cloud commitment underscores the enormous compute appetite required to train and serve frontier AI models, and cements Google Cloud’s position as Anthropic’s primary infrastructure partner.

Cerebras Targets $26.6 Billion IPO — Largest Tech Offering of 2026

AI chipmaker Cerebras Systems filed an amended registration statement with the SEC on Sunday, moving closer to a public offering that could raise $3.5 billion and value the company at $26.6 billion. The Sunnyvale-based startup plans to sell 28 million shares at $115 to $125 each.

Investor appetite appears strong: banks are already fielding roughly $10 billion in orders for the $3.5 billion worth of shares on offer. Cerebras reported $510 million in 2025 revenue and has landed deals with both AWS and OpenAI — the latter reportedly worth more than $10 billion. If the IPO prices at or above the top end, it will be the largest tech listing of the year so far.

Sierra Raises $950M at $15.8B Valuation as AI Agent Demand Surges

Bret Taylor’s AI customer-service startup Sierra announced a $950 million Series E led by Tiger Global and Google’s GV, with participation from Benchmark, Sequoia, and Greenoaks. The round values the company at $15.8 billion post-money.

Sierra builds AI agents that handle customer interactions for large enterprises — refinancing mortgages, processing insurance claims, managing returns, and more. The company now counts more than 40% of the Fortune 50 as customers, with agents on its platform handling billions of interactions. Taylor’s resume spans leadership roles at OpenAI, Salesforce, and Facebook, and he served as Twitter’s chairman during Elon Musk’s acquisition.

Pentagon Awards Classified AI Contracts to 8 Tech Giants — Anthropic Still Blacklisted

The Department of Defense has signed AI agreements with eight major technology companies to deploy models on classified IL6 and IL7 networks: Amazon Web Services, Google, Microsoft, NVIDIA, OpenAI, Oracle, SpaceX, and Reflection AI. The systems will focus on data synthesis, warfighter decision-making, and situational awareness.

Notably absent is Anthropic, which the Trump administration blacklisted after the company insisted the Pentagon adopt safety guardrails for AI use in warfare. The Pentagon went further, designating Anthropic a “supply chain risk” — a label previously reserved for companies associated with foreign adversaries. Defense contractors must now certify they do not use Claude in military work.

However, the White House has reportedly reopened discussions with Anthropic in recent weeks. CEO Dario Amodei met with senior Trump administration officials earlier this month in what both sides called a “productive” discussion, and President Trump told CNBC that a deal is “possible.”

Colorado Rewrites Landmark AI Law, Drops Transparency Requirements

Colorado lawmakers are advancing SB 189, a bill that would repeal and replace the state’s pioneering 2024 AI consumer protection law with a narrower, disclosure-focused framework. The original law required companies to assess and disclose detailed information about how their AI systems work; the new bill replaces that with a simple notice that automated decision-making technology is being used.

Consumers who receive adverse decisions can request more detailed information within 30 days, including access to an appeals process with “meaningful human review.” The bill also pushes the start date back to January 2027 and replaces joint liability between AI developers and deployers with a fault-allocation model. The scope is limited to “consequential decisions” in employment, education, housing, lending, insurance, healthcare, and government services.

IREN Acquires Mirantis for $625M, Pivots from Bitcoin Mining to AI Cloud

IREN Limited announced a $625 million all-stock acquisition of cloud infrastructure company Mirantis, signaling the company’s pivot away from pure-play bitcoin mining toward AI cloud services. Mirantis brings Kubernetes-based orchestration, enterprise support, and a customer base of over 1,500 enterprises globally.

Mirantis will operate as a standalone subsidiary, supporting faster deployment of workloads on IREN’s GPU infrastructure while continuing to serve its existing customers. The deal implies roughly four to five times revenue and adds a critical software layer to IREN’s hardware-heavy business.