TL;DR

DeepTech is raising like infra. CuspAI ($450M / $2.6B), Proxima Fusion (€411M / €2.4B), PsiQuantum ($1B / $7B), and Nearfield ($380M / $1.6B) show capital chasing materials, energy, quantum, and chip tooling that sit under the AI stack.

If you only watch SaaS and chatbot rounds, you will miss where the real checks are landing. The last stretch of capital into DeepTech looks less like science grants and more like infrastructure finance: nine-figure rounds, multi-billion valuations, strategics in the cap table.

Materials discovery. Cambridge-based CuspAI closed a $450M Series B at a $2.6B valuation, co-led by Kleiner Perkins and NEA with significant participation from Bezos Expeditions (and the UK Sovereign AI Fund among others). The pitch is blunt: a search engine for materials that can screen on the order of 300 trillion molecular structures in months rather than years, for carbon capture, semiconductors, and clean energy. For AI and Software & SaaS operators, this is not a side story. New materials feed chips, batteries, and the physical layer your models eventually run on.

Fusion energy. Munich’s Proxima Fusion raised €411M at a €2.4B valuation, reported as Europe’s largest private fusion round, with Google and RWE among the backers. They are industrializing the stellarator path toward clean baseload power. AI data centers and advanced manufacturing do not scale on hope. They scale on electrons. Energy DeepTech is becoming a corp-dev and procurement signal, not only a climate narrative.

Quantum computing. PsiQuantum raised $1B in a Series E led with BlackRock (alongside Temasek and Baillie Gifford), valuing the company at $7B, to push fault-tolerant, million-qubit-scale photonic quantum machines. That round landed in September 2025 and still frames the 2026 DeepTech mood: late-stage capital is willing to underwrite decade-long physics bets when the compute thesis is clear.

Semiconductors. Dutch toolmaker Nearfield Instruments raised $380M in a Series D at a $1.6B valuation (led by Fidelity, with Temasek and others) to scale 3D metrology and inspection tools for advanced chip manufacturing. As chips get smaller, the tech to check them becomes as valuable as the tech to make them. AI demand is a wafer problem before it is a model problem.

Why this matters for Agentivo’s focus areas (AI, Software & SaaS, and B2B operators in North America): these rounds sit under your stack. Materials and inspection shape chip supply. Fusion shapes long-run power cost for compute. Quantum is the adjacent compute platform investors keep pricing as optionality. Put simply: capital is moving from bits back to atoms. If you buy AI products, diligence vendors, or advise founders, watch the DeepTech layer.

My View on This

Do not treat DeepTech as distant from AI advisory. Ask vendors and founders who depend on scarce chips or power what their supply assumptions are. For corp-dev, watch materials AI, semiconductor tooling, and energy as strategic adjacency to software platforms. The winners in application AI will still need someone to invent the material, inspect the wafer, and keep the lights on.

Sources