In partnership with
Good morning, {{first_name | startup founders}}.
CuspAI just landed one of the biggest AI funding rounds of the summer — a $450M Series B that values the materials-discovery startup at $2.6 billion, with Jeff Bezos personally writing a check alongside Kleiner Perkins and NEA. Nvidia is throwing in compute power too, betting that AI can find the next generation of chip and battery materials faster than any lab.
With the round pushing CuspAI’s valuation up roughly 5x in under a year, is materials science becoming the next frontier where AI labs and venture capital collide — or just another sign that any startup adjacent to semiconductors can command nosebleed prices?
In today’s Startup News AI:
CuspAI raises $450M to hunt for new materials
Skyfall AI emerges from stealth to run a company solo
Infinity raises $15M to build a CUDA alternative
Introducing The First Agentic CRM
Get revenue agents, workflows, and automations across every stage of your motion. Access customer data in real time through Attio's web app, MCP, API, and SDK.
Then Ask Attio anything about your business and get instant answers.
It's the CRM that runs the work behind every win.
What’s new? CuspAI raised a $450M Series B led by Kleiner Perkins and NEA, valuing the British startup at $2.6 billion — up from $520M just nine months ago — with Jeff Bezos, Nvidia, and the UK’s Sovereign AI Venture Fund joining the round.
What matters?
CuspAI’s MIRA platform lets researchers define target material properties and lets generative AI generate, evaluate, and prioritize candidate compounds for chips, batteries, and clean energy.
The round coincides with the launch of CuspAI’s AI Materials Foundry, a 45-plus-member consortium that includes Nvidia’s compute infrastructure and Meta’s Fundamental AI Research team.
Bezos is doubling down on materials AI, having also backed Prometheus, a physical-engineering AI startup that raised $12B at a $41B valuation.
Why it matters?
Materials discovery has long been one of science’s slowest, most expensive bottlenecks, and investors are now pricing AI’s ability to compress it into a defensible, venture-scale business. For founders in hard tech, CuspAI’s valuation jump signals that “AI plus atoms” plays are starting to command software-style multiples.
What’s new? Skyfall AI emerged from stealth, an “Enterprise World Models” startup founded by the ex-Microsoft team behind Maluuba, with backing from Fidelity, Inovia Capital, Touring Capital, M13, NextView Ventures, and Garage Capital.
What matters?
Founders Sam Pasupalak and Kaheer Suleman plan to acquire a small SaaS business for up to $1M and let their AI, not a human executive, run it end-to-end.
Their flagship system, Morpheus, is a continual reinforcement-learning platform that simulates thousands of business decisions across finance, operations, marketing, and HR before acting.
The team’s pedigree includes Maluuba, acquired by Microsoft, and early work on Siri, Alexa, and deep learning research alongside Yoshua Bengio.
Why it matters?
If an AI system can run a real business and grow revenue without a human CEO, it reframes “AI agents” from copilots into a genuine labor substitute at the executive level. Investors backing this bet are underwriting a much bigger claim than most enterprise AI pitches: that judgment itself, not just tasks, can be automated.
SPONSORED BY MINTLIFY
AI agents now read your docs almost as much as humans do
5% of traffic to your docs is now AI agents, not humans. If your documentation isn't structured for machine readability, your product is invisible to Claude, Cursor, and every other coding agent your buyers use daily. Mintlify is built for both audiences.
What’s new? Infinity raised a $15M round at a $100M valuation from Touring Capital, Principal VC, and researchers at OpenAI and Anthropic, to build low-level kernel software that helps AI chips run models more efficiently.
What matters?
Founder Jeremy Nixon, a former Google Brain researcher and creator of AGI House, built an AI research agent called Ignition that automatically generates, tests, and optimizes kernel code across GPUs, phone chips, and Systolic Arrays.
The startup charges a performance-based fee tied to tokens-per-second gains rather than upfront licensing, aligning its revenue directly with the speedups it delivers.
Infinity already counts AI chipmaker D-Matrix as a customer, evidence its kernel-optimization pitch works beyond Nvidia’s own hardware.
Why it matters?
A viable CUDA alternative would loosen one of Nvidia’s stickiest lock-in points — the software layer developers depend on to get performance out of any chip. For founders building inference infrastructure, Infinity’s backing from researchers at both OpenAI and Anthropic is a notable signal that insiders see real technical merit here.
The Shortlist
Databricks raised fresh capital at a $188 billion valuation this week, a 40% jump from its $134B round five months ago, led by existing investor Coatue.
LimX Dynamics closed nearly $200M in pre-IPO funding this week to scale its humanoid robots, pushing the Chinese startup toward a public listing.
Moonshot AI halted new subscriptions for its Kimi K3 model just two days after launch, after demand for the 2.8-trillion-parameter coding model overwhelmed its GPU capacity.