Daily AI Brief: Anthropic’s big week — Bun acquisition, Claude Opus 4.5, $1B Claude Code run‑rate and Snowflake tie‑up

Yesterday and today brought a flurry of Anthropic announcements that position Claude at the center of recent AI momentum. Highlights include Anthropic’s acquisition of Bun, the launch of Claude Opus 4.5, Claude Code reaching an approximately $1 billion annualized run‑rate, and a $200 million partnership with Snowflake to integrate agentic AI into enterprise data platforms. Separate research using Claude agents also identified substantial smart‑contract vulnerabilities — reported but not exploited — underscoring both capability and risk.

Key items (Dec 1–3, 2025):

• Anthropic acquires Bun (Dec 3) — Anthropic purchased Bun (a fast JavaScript runtime and tooling project) to accelerate Claude Code infrastructure work. Bun remains open source under the MIT license. (See Anthropic announcement and coverage.)
• Claude Opus 4.5 (Dec 3) — Anthropic released Opus 4.5, an incremental model update optimized for coding, agents and efficiency.
• Claude Code $1B run‑rate (Dec 3) — Anthropic announced Claude Code has reached an estimated $1 billion annualized revenue run‑rate shortly after public availability.
• Snowflake partnership — $200M (Dec 3) — Snowflake and Anthropic agreed to a roughly $200 million integration deal to bring agentic AI into Snowflake’s data cloud for enterprise analytics and automation.
• Agent security research (Dec 1) — Claude agents autonomously discovered potential smart‑contract exploits worth up to $4.6M; Anthropic reported the findings and did not exploit them, noting implications for AI in security research and defense.

Other major players (past 48 hours)

No major model releases, product launches, or regulatory actions for OpenAI, Google/DeepMind or Meta were reported in the last 48‑hour window covered by this brief.

Why this matters

• Anthropic’s Bun acquisition and Opus 4.5 suggest a targeted push to improve developer tooling, model efficiency and agent capabilities — all beneficial for coding and production deployments.
• Claude Code’s rapid revenue ramp and the Snowflake enterprise tie‑up indicate strong commercial traction for agentic AI in the enterprise sector.
• The smart‑contract research highlights dual‑use concerns: agentic AIs can accelerate security discovery but also raise risks if misapplied.

Sources & further reading

Primary reporting and announcements from Anthropic and partner coverage: Anthropic announcement, Orrick coverage, and reporting on agent research: The Register. Snowflake partnership details: Anthropic press.

Discussion: Do these moves make Anthropic a stronger contender against the likes of OpenAI and Google for enterprise AI — or do the dual‑use risks from agent research mean regulators should step in faster?