Ex-DeepMind founders raise $500M; Wayve $8.5B tender

Three ex‑DeepMind researchers turned their poker‑winning reinforcement‑learning agents into stock and crypto traders, raising an undisclosed Series A led by Creandum that values EquiLibre at $500 million. Their algorithms now trade billions of dollars daily for quant hedge funds and have posted a perfect record of no negative months since launch.

Wayve let employees sell $85M of vested shares back to investors, pricing the company at a fresh $8.5 billion after its $1.2 billion Series D. The move, Wayve’s second tender, mirrors a growing AI‑startup trend that uses liquidity events to lock in talent instead of waiting for an exit.

A study of 15 early‑stage SaaS landing pages found 40% fail the 10‑second clarity test, visitors can’t tell what the product does and sign‑ups stay at zero. The first few seconds decide conversion, so ditch jargon and spell out the customer outcome to stop losing sign‑ups.

AWS announced a new $1 billion Forward‑Deployed Engineer org that embeds engineers inside client firms to build custom AI agents and hand over lasting AI capabilities. The model mirrors recent OpenAI and Anthropic joint ventures but is funded internally, letting Amazon scale rapid, self‑service AI deployments for enterprises.

A founder recounts how his first co‑founder match through Antler collapsed after two weeks, almost derailing the startup. The story highlights the danger of speed‑dating style founder matching and why deep vetting matters before committing.

Meta will limit the on‑device Conversation Focus feature to three hours per month for free users and charge $19.99 a month for a Premium plan that raises the cap to fifteen hours. The move tests whether hardware owners will pay for AI tools that already sit in the device, signaling a broader shift toward subscription‑based monetization of consumer AI.

Anthropic’s new Claude Science app lets researchers run analyses, query 60+ databases, and generate fully traceable figures and code in a single environment. It auto‑manages compute across laptops, HPC clusters, or cloud, promising faster, reproducible work in genomics, proteomics, structural biology and more. The result is a tighter, quicker path from data to publishable insight.

Ramp’s new paper links credit‑card AI spend to workforce data for 21,500 U.S. firms and finds that companies with high AI‑spending intensity grow headcount about 10% within two years, driven largely by a 12% rise in entry‑level hires. Low‑intensity adopters show no significant change, challenging fears of AI‑driven layoffs.