← Back to archive
Crusoe tripling valuation to $30B

Image via 404 Media

Sunday, July 5, 2026

Crusoe tripling valuation to $30B

Crusoe is raising $3 billion at a $30 billion valuation, tripling its value on demand for AI infrastructure. Meta launched Pocket, a new app that generates interactive games and apps using AI.

Top Stories

1
Crusoe raising ~$3B

Bloomberg

Crusoe is raising ~$3 billion at a $30 billion valuation, tripling its worth from $10 billion in October, as investors bet heavily on AI infrastructure providers serving major tech companies like Meta and Oracle.

fundingai-infrastructuredata-centerscrusoe
2
Venice AI unicorn round

Substack

Venice AI, a privacy-centric AI platform with $70M ARR, raises $65M Series A at $1B valuation by offering uncensored access to 200+ models with full user encryption, appealing to those who view AI restrictions as dangerous surveillance.

privacyai-modelsfundingunicorn
3
Meta launches Pocket

Google Play

Meta launches Pocket, an AI-powered app for creating interactive games and apps via prompts, marking the company's latest expansion into mainstream AI creation tools following its Gizmo acquisition.

metaaigaminggenerative-ai
4
Reinforcement Learning with Metacognitive Feedback Elicits Faithful Uncertainty Expression in LLMs

Yale University & Google Research

Yale and Google researchers develop RLMF, a metacognitive training approach that teaches LLMs to self-assess performance accuracy, improving faithful uncertainty calibration by up to 63% over standard RL and enabling smaller models to outperform frontier systems like GPT-5 in expressing reliable confidence.

llmreinforcement-learningcalibrationuncertainty
5
TUA-Bench: A Benchmark for General-Purpose Terminal-Use Agents

Meta AI

TUA-Bench establishes a 120-task benchmark for evaluating terminal-use agents on both routine and expert workflows, revealing that even frontier models achieve only 65.8% success—demonstrating substantial remaining gaps in agent reasoning and execution planning in text-based environments.

agentsllmbenchmarkingterminal-use

Keep Reading

Industry Voices

Enjoyed this issue?

Get daily AI intel delivered to your inbox. No fluff, just the stories that matter.