Welcome, Startup enthusiasts!

Good Morning. It’s July 20, 2026.

No image has been created. This issue uses startup developments reported on July 20, 2026, with claims checked against Reuters, TechCrunch, and official product documentation.

🚀 The Startup Khan Newsletter

July 20, 2026 — The Infrastructure Behind Intelligence

🧠 Top Story: CuspAI Raises $450M to Invent Materials That Don’t Exist Yet

British startup CuspAI has raised $450 million in Series B funding at a $2.6 billion valuation.

Kleiner Perkins and NEA led the round, with participation from the UK government’s Sovereign AI Venture Fund, Bezos Expeditions, AMD Ventures, Lux Capital, Invest-NL, and others. The company also launched the AI Materials Foundry, a coalition of more than 45 organizations—including Nvidia and Meta—designed to pool the computing and scientific resources needed to discover new materials. (Reuters)

CuspAI’s platform, MIRA, supports the materials-development process from generative design and simulation through synthesis planning and experimental validation. The target markets are enormous: semiconductors, clean energy, and advanced manufacturing all depend on materials with better performance, lower costs, or improved availability. (Reuters)

Why founders should care:

AI’s next breakout companies may not sell chatbots.

They may use AI to solve expensive, physical-world bottlenecks where successful discoveries create defensible intellectual property and measurable economic value.

👉 Founder takeaway: Look for industries where better prediction can eliminate years of trial and error.

💰 Other Funding Moves

1️⃣ Natural raises $30M to build payments for AI agents

Natural raised $30 million to develop payment infrastructure that lets AI agents complete transactions rather than stopping when human approval or traditional checkout flows are required.

The company is entering a high-stakes market: the financial layer for autonomous software. As agents move from recommending actions to executing them, identity, permissions, fraud prevention, and transaction controls become core product infrastructure. (TechCrunch)

2️⃣ Infinity raises $15M for AI inference infrastructure

AI infrastructure startup Infinity announced a $15 million round at a $100 million valuation.

The financing included Touring Capital, Principal VC, and researchers associated with OpenAI and Anthropic. Infinity is focused on inference—the production stage where trained models process real customer requests. (TechCrunch)

The founder lesson: training gets attention, but inference generates the recurring bill.

3️⃣ Moonshot AI hits a capacity wall

Chinese startup Moonshot AI temporarily paused new subscriptions after demand for its Kimi K3 model strained available capacity. The company is also seeking fresh capital and restructuring ahead of a possible Hong Kong listing, according to Reuters. (Reuters)

Demand is great. Demand without capacity is a warning.

🧰 Tool of the Day: LiteLLM

LiteLLM gives AI teams a unified interface for accessing more than 100 language models.

Its gateway can help founders route requests between providers, monitor spending, assign budgets, and reduce dependence on a single model vendor. It is also compatible with OpenAI-style APIs, making provider changes less disruptive. (LiteLLM)

Best use case:

👉 Put LiteLLM between your application and model providers before switching costs become painful.

📊 Trend Check: New York’s Startup Engine Accelerates

More than 240 New York City startups raised a combined $1.13 billion in seed funding during the first half of 2026, up from $1.06 billion during the same period in 2025.

Average seed rounds reportedly increased from $5.4 million to $6.64 million. Across all stages, New York startups raised approximately $16 billion in H1 2026, approaching the $19.1 billion raised during all of 2025. (TechCrunch)

The signal: capital remains concentrated, but strong startup ecosystems outside Silicon Valley continue gaining depth.

🌟 Founder Spotlight: Chad Edwards and Max Welling

CuspAI co-founders Chad Edwards and Max Welling are building at the intersection of AI, chemistry, and industrial production.

Their playbook is worth studying:

Start with a difficult scientific constraint. Build proprietary technology around it. Then create an ecosystem that brings together capital, compute, researchers, manufacturers, and prospective customers. (Reuters)

That is more than a product strategy.

It is market orchestration.