RobinsonLabs/ news

Anthropic files for IPO at staggering $965 billion valuation

The Claude maker submits confidential draft paperwork to the SEC, positioning itself as the most valuable private AI company in the world and surpassing OpenAI in the public-market race.

Anthropic has confidentially submitted a draft Form S-1 to the SEC, laying the groundwork for one of the largest technology IPOs in history. The filing follows a $65 billion Series H round that valued the company at $965 billion, driven by explosive revenue growth — annualized run-rate revenue reportedly surpassed $47 billion by May 2026, up from about $9 billion at the end of 2025. More than 1,000 business customers were each spending over $1 million annually by April.

The move gives Anthropic a procedural lead over rival OpenAI in the race to go public, while the company continues to expand its infrastructure through a deal with SpaceX for access to the Colossus 1 supercomputer featuring 220,000-plus NVIDIA GPUs. Key hires including Andrej Karpathy for Claude pretraining have further strengthened its research team.

China unveils $295 billion state-directed AI data center buildout

Beijing is preparing to spend roughly 2 trillion yuan over five years to build an interconnected nationwide network of data centers, driven largely by state telecoms and domestic chipmakers.

The blueprint, drafted by the National Development and Reform Commission and other agencies, calls for China Mobile and China Telecom to operate the bulk of the data centers, linking scattered regional computing facilities into a cohesive national network by 2028. At least 80 percent of technology, including AI chips, would come from local suppliers such as Huawei Technologies, effectively squeezing out Nvidia and AMD from the core infrastructure buildout.

The plan is funded through sovereign debt mechanisms and supplementary private capital, with total investment potentially reaching $735 billion when integrated with power grid upgrades. Chinese data centers are built at lower cost than U.S. equivalents, and the investment does not even include spending by private giants Alibaba and Tencent, whose AI capital expenditure programs run separately.

Claude Opus 4.8 tops AI model benchmark with record intelligence score

Anthropic's newest flagship model leads the Artificial Analysis Intelligence Index at 61.4 and introduces dynamic parallel subagents for large-scale autonomous coding tasks.

Claude Opus 4.8 delivers significant gains across benchmarks over Opus 4.7, including 69.2 percent on SWE-bench Pro (up from 64.3 percent) and a dramatic jump to 96.7 percent on the USA Mathematical Olympiad (from 69.3 percent). The model also achieved 68.1 percent on the 1M-token subset of GraphWalks, a 27.8-point improvement, signaling major gains in long-context reasoning. It maintains the same standard API pricing as Opus 4.7: $5 per million input tokens and $25 per million output tokens, while introducing a new fast mode at double the price for up to 2.5 times higher output throughput.

The release includes dynamic workflows allowing Claude to run hundreds of parallel subagents for codebase-scale migrations, effort controls for balancing speed and quality, and improved self-calibration making the model four times less likely to let code flaws pass unremarked. It is available across the API, Claude.ai, Bedrock, Vertex AI, and Microsoft Foundry.

Microsoft launches in-house reasoning model family at Build 2026

The company unveiled MAI-Thinking-1, its first advanced reasoning model built from scratch with 35 billion active parameters, signaling a strategic pivot toward full AI independence from OpenAI.

MAI-Thinking-1 was trained on clean data with no distillation from third-party frontier models, targeting multi-step reasoning, mathematics, software engineering, and code generation. It joins a broader family of seven new models announced at Build, including MAI-Code-1-Flash for GitHub Copilot and VS Code integration, MAI-Image 2.5 for text-to-image generation, and MAI-Transcribe-1.5 for speech-to-text tasks running five times faster than competitors. Small "Aion" models are optimized to run locally on Windows PCs.

The launch follows a revised OpenAI partnership that extends to 2032 but is now non-exclusive, allowing Microsoft to independently build and deploy its own models while retaining Azure as a first-party cloud. The company has invested $13 billion in OpenAI and $5 billion in Anthropic, and the MAI models are being embedded across Copilot, GitHub, Windows, Azure, and Microsoft 365 to reduce reliance on external AI suppliers.

Inspira Enterprise Expands to Full-Spectrum ServiceNow Delivery Partner

A global cybersecurity and AI services firm deploys 50+ AI agents on ServiceNow and broadens its partnership to cover AI, security, and identity management.

Inspira Enterprise announced a strategic expansion of its partnership with ServiceNow, evolving into a full-spectrum delivery partner covering AI governance, cybersecurity, and identity management. The company deployed ServiceNow's AI Control Tower (AICT) across its own operations, achieving a 35% increase in productivity, a 40% boost in AI adoption, and 60–70% reductions in implementation effort through pre-built accelerators. Inspira now offers over 200 ready-to-deploy use cases across six industries and integrates both ServiceNow's recent acquisitions — Veza for identity governance and Armis for asset intelligence — into its offering.

ServiceNow Lays Off Hundreds, Possibly Thousands, of Employees

The company eliminated roles across sales, solution consulting, and support functions in a sweeping cut that stunned the workforce.

On June 10, 2026, ServiceNow carried out layoffs that affected hundreds of employees, with internal estimates ranging from 300 to as many as 2,500. The cuts hit sales, solution consulting, and customer support roles, with reports of employees learning about their termination during routine meetings as HR joined the call. The San Diego Union-Tribune reported that 63 employees were let go at the company's San Diego office alone. The layoffs followed a period during which CEO Bill McDermott had publicly stated that ServiceNow would not pursue widespread layoffs. The cuts represent one of the most significant workforce reductions in the company's recent history.

ServiceNow and IBM Expand Partnership for Enterprise AI Modernization

The two companies are combining their platforms to help enterprises modernize legacy systems and make their data AI-ready at scale.

ServiceNow and IBM announced an expansion of their multi-year collaboration focused on solving two major barriers to enterprise AI: legacy application layers and the AI-ready data problem. The expanded partnership integrates IBM Bob (for scanning and refactoring legacy systems), IBM watsonx.data (for data governance and quality), and a suite of infrastructure tools including Red Hat Ansible, HashiCorp Terraform, and Instana into ServiceNow's IT workflows. The goal is to enable autonomous IT operations that proactively detect and resolve infrastructure issues before they impact business operations. Joint solutions are expected to launch in the second half of 2026, targeting the enterprise modernization market where ServiceNow and IBM compete and complement each other in the IT service management space.