R
Recursive Superintelligence
Automating the AI development pipeline for self-improvement.
About Recursive Superintelligence
Recursive Superintelligence is a UK-based AI startup founded by former engineers from Google DeepMind and OpenAI, including Richard Socher (former Chief Scientist at Salesforce) and Tim Rocktäschel (AI Professor at University College London and former Principal Scientist at Google DeepMind). The roughly 20-person team is developing a new kind of AI that keeps improving itself without human involvement — a concept known as recursive self-improvement, widely considered a key pathway to superintelligence. The company raised at least $500 million at a $4 billion pre-money valuation in April 2026, led by GV (formerly Google Ventures) with participation from Nvidia.
Source: Dealroom company dataAt a glance
- Founded
- 2026
- Headquarters
- London, United Kingdom
- Sector
- AI
- Power Law
- Unicorn$1–10B valuation tier
Knowledge graph
News
Ex-Anthropic founders' Mirendil eyes $5B valuation months after $1B seed
Sep 23, 2026
News
DeepMind spinout Emulate nears $700M seed at $3.7B valuation
Sep 17, 2026
Analysis
A quarter of 2026’s new unicorns are less than two years old
Sep 10, 2026
News
Recursive Superintelligence signs $410M multi-year AWS compute deal to scale self-improving AI
Jul 28, 2026
More headlines
Investors
From Dealroom's funding and investor records, grouped by the round each investor first entered.
Seed4 investors entered at this stage
Source: Dealroom Talent Intelligence.
Global footprint
across markets
Market sentiment
An AI-synthesised read of the highest-engagement posts about Recursive Superintelligence on X, ranked by likes and reposts, corporate channels excluded.
Reading the room on X — pulling top posts and synthesizing themes…
Source: X recent search ranked by engagement (likes + retweets) · Synthesis by Claude · Cached for 1 hour
See Recursive Superintelligence on the full Dealroom platform
Live deal flow, founder pedigree, hiring signals, revenue trajectory, and the full cap table.