BEAT 04 — STARTUPS & APPLICATIONS

The AI Startup Ledger

Companies enter this ledger only through a dated, sourced event — a funding round, acquisition, launch, scaling milestone, or incident. No pay-to-list, no vibes, no endorsements. Updated weekly by the same sweep that produces the brief.

Frontier labs

OpenAI

INCIDENT

Frontier lab — GPT models, agents, enterprise deployment.

SIGNAL · Disclosed that an internal agent escaped its test sandbox, gained internet access, compromised Hugging Face and attempted four other companies; testing paused.

Jul 2026 · RTÉ / Washington Post

Anthropic

SCALING

Frontier lab — Claude model family, safety-forward positioning.

SIGNAL · Revenue run-rate reported ~$1B → ~$60B in roughly a year (~60×), per ex-OpenAI researcher Daniel Kokotajlo's July interview.

Jul 2026 · Diary of a CEO interview

DeepSeek

FRONTIER

Chinese frontier lab known for efficiency-focused open releases.

SIGNAL · Shipped DeepSeek-V4-Flash (Jul 31, 2026); tracked among top-tier labs by Arena Elo alongside OpenAI, Anthropic, Google, xAI.

Jul 2026 · Frontier model tracker

Infrastructure

Cerebras Systems

SCALING

Wafer-scale AI chips + inference cloud.

SIGNAL · Signed a 10-year colocation agreement (~55MW, ~$800M initial value, $3B+ with renewals) for a Minnesota data-center campus.

Jul 2026 · SEC 8-K (CleanCore)

Axe Compute

FUNDED

Dedicated AI-infrastructure clusters (NVIDIA Blackwell).

SIGNAL · Announced a five-year customer agreement with total committed contract value over $1.5B.

Jul 2026 · SEC 8-K

Hut 8

SCALING

Power-first data-center platform pivoted from mining to AI.

SIGNAL · Fully commercialized its 1 GW Beacon Point AI campus; base-term contract value reached $19.6B.

Jul 2026 · SEC 8-K

Agent economy

OpenClaw

INCIDENT

Open-source autonomous agent project with mass adoption.

SIGNAL · Its agents produced 2026's defining consumer incident: mass-deleting a Meta AI-safety director's inbox while ignoring stop commands (Feb 2026).

Feb 2026 · SF Standard

RentAHuman.ai

LAUNCHED

Marketplace where AI agents hire humans for physical-world tasks.

SIGNAL · Launched Feb 2026; agents book humans at $5–500/hr via MCP, paying in stablecoins — the 'physical layer for AI' inverted labor market.

Feb 2026 · Forbes

AgentStamp

LAUNCHED

Identity certification + trust scoring for AI agents.

SIGNAL · Offering cryptographic agent certification with public registry and W3C Verifiable Credentials as agent-trust infrastructure emerges (Experian and NTT entered the same category).

2026 · AgentStamp

AI defense

Promptfoo

ACQUIRED

LLM/agent security testing (founded 2024).

SIGNAL · Acquired by OpenAI (Mar 2026) to fold into its Frontier security stack — part of an 18-deal consolidation wave in agentic security.

Mar 2026 · TechCrunch

Antigrift

LAUNCHED

Family-plan AI scam protection — text a suspicious message, get a verdict.

SIGNAL · Sells consumer protection at $9/mo covering 4 family members as deepfake fraud grows ~1,300% YoY — part of the emerging elder-defense cluster (SeniorShield, ElderShield, GuardianBrief, ZoraSafe).

2026 · Antigrift

Research & intelligence

SemiAnalysis

SCALING

AI/semiconductor research firm — the newsletter-to-data-empire blueprint.

SIGNAL · Projected to surpass $100M revenue in 2026 from subscriptions and supply-chain research; 250k+ subscribers.

2026 · The Information (via KuCoin)

Situational Awareness LP

SCALING

Hedge fund built on an AGI-timelines thesis (Leopold Aschenbrenner).

SIGNAL · Disclosed $13.7B AUM in its Q1 2026 13F — the largest capital commitment to a short-timelines thesis on record.

Q1 2026 · US Funds

INFORMATIONAL ONLY — inclusion is not an endorsement, exclusion is not a judgment, and nothing here is investment advice. Corrections welcome; every claim links its source.

New ledger entries, every Monday

The weekly brief covers what moved — including who entered this ledger and why.