Claude Report — 2026-05-10

🚀 Anthropic Official

Anthropic signs $1.8B Akamai compute deal

Locks a seven-year capacity contract with Akamai on May 8 to power Claude inference, Akamai's largest deal ever, sending the stock up 27%. Revenue starts Q4 2026 and stacks on top of recent SpaceX, Google/Broadcom, and AWS commitments.

Claude Managed Agents adds Dreaming, Outcomes, multiagent

Ships three Managed Agents capabilities on May 7: Dreaming (research preview that mines past sessions for self-improvement patterns), Outcomes (rubric-based grader with revision loop and webhooks), and multiagent orchestration (lead agent delegates to specialists on shared filesystems, already used by Netflix). All under the managed-agents-2026-04-01 beta header.

Amodei confirms 80x annualized growth, $30B run rate

Reveals on May 8 that Q1 2026 grew 80x annualized vs an internal 10x plan, calling it "too hard to handle." Annualized revenue hit $30B (3x year-over-year), driving the SpaceX/Akamai capacity scramble.

Dragos: Claude misused against Mexican water utility

Discloses on May 8 that an unknown threat group used Claude during a Dec 2025–Feb 2026 campaign against nine Mexican government agencies, breaching a water utility's IT (OT not compromised). Claude identified a vNode industrial gateway, generated targeted credential lists, and ran password sprays without prior ICS knowledge.

CNBC: Mythos triggers cybersecurity "hysteria"

Reports on May 8 that banks, insurers, and regulators are scrambling over Mythos disclosures, while researchers argue the same vulnerabilities are reproducible with public Claude and GPT models via orchestration, undercutting the panic narrative.

🔌 Claude Code Plugins

sap-fiori-mcp-server

Lands in the official marketplace on May 9 (PR #1777) bringing SAP Fiori tooling as an MCP server entry.

clickhouse

Added to marketplace on May 7 (PR #1683) for ClickHouse database operations from Claude Code.

zilliz

Added to marketplace on May 7 (PR #1686), exposing Zilliz Cloud vector storage workflows inside Claude Code.

qdrant-skills

Added to marketplace on May 7 (PR #1685) packaging Qdrant vector-search skills.

dash0

Added to marketplace on May 7 (PR #1641) for Dash0 observability integration.

outputai

Added to marketplace on May 7 (PR #1709) bringing OutputAI workflows to Claude Code.

pigment

Added to marketplace on May 7 (PR #1684).

Marketplace policy scan tightened

Tightens hook-scope, telemetry, and disclosure validation on May 7 (PR #1771), making policy violations blocking for new plugin submissions.

Unchanged in window: frontend-design, superpowers, context7, code-review, code-simplifier, github, skill-creator, playwright, feature-dev, claude-md-management.

🛠️ Skills

No relevant news today.

Unchanged in window: superpowers:brainstorming, superpowers:tdd, superpowers:systematic-debugging, superpowers:writing-plans, superpowers:executing-plans.

🤖 Agents & Subagents

No relevant news today.

Unchanged in window: superpowers:subagent-driven-development, claude-router:orchestrate, ralph-loop, claude-router:fast-executor, claude-mem:do.

🔗 MCPs & Integrations

claude-agent-sdk-python 0.1.77

Replaces generic "Command failed with exit code 1" errors with actionable messages carrying the actual result text on May 8, and deprecates "Skill" in allowed_tools in favor of the new skills option on ClaudeAgentOptions.

claude-agent-sdk-python 0.1.78–0.1.80

Bumps bundled Claude CLI through 2.1.136, 2.1.137, and 2.1.138 across May 8–9, keeping Python users in sync with the latest CLI hotfixes.

Unchanged in window: github-mcp-server, modelcontextprotocol/servers, filesystem MCP, claude-mem, context7.

💡 Community — Workflows & Ideas

EPAM commits to 10K Claude-certified architects

Announces multi-year partnership with Anthropic on May 7 covering 10,000 architect certifications and 250 forward-deployed "Black Belts," signaling a large enterprise services pipeline for Claude Code and Agent SDK.

OT vendors push for Mythos access

Reports on May 8 that operational-technology vendors are pressing Anthropic for inclusion in the Mythos preview, arguing the current rollout omits a critical-infrastructure segment that's a frequent attack target.

📰 Quick Mentions

• Pentagon CTO calls Mythos a separate "cyber moment" — DoD keeps Anthropic blacklisted but treats Mythos as its own national-security question.
• Plugin marketplace pins all GitHub Actions to commit SHAs — Supply-chain hardening lands across the official marketplace repo on May 7.
• EPAM issues Q2 guidance citing Anthropic AI-native revenue — May 9 filing flags Anthropic partnership as a growth driver despite trimmed FY outlook.