A hypersonic missile developer and an AI chip startup secured the two largest venture rounds of the week, signaling an accelerating investor appetite for defense technology and artificial intelligence infrastructure. Castelion, a Torrance, California-based defense tech company, raised $800 million in Series C equity plus $250 million in debt, while San Jose-based Etched collected $700 million for its AI inference clusters. These deals, along with a $400 million round for Higgsfield, underscore a clear trend: defense tech, AI tools and infrastructure lead the way in current venture capital activity.
Top Deals at a Glance
The week's biggest rounds featured a mix of hardware, software and defense. Check the key deals below that shaped the funding landscape.
Why This Matters
The concentration of capital in defense and AI infrastructure reflects a structural shift in venture priorities. Governments are racing to field hypersonic weapons, making Castelion a strategic bet. AI inference, the compute layer that runs trained models, is becoming a bottleneck for enterprises, which explains the high valuations for Etched and its custom silicon. Higgsfield's round signals that video-generation tools are entering a commercial phase, competing with established players. For startups in adjacent spaces, this funding wave means more competition for talent and partnerships, but also clearer signals on where the market is heading. The winners in this cycle will be those that can deliver real-world performance, not just promise.
A Broader Trend in Venture Capital
This week's rounds are not isolated events. They are part of a multiyear pattern where deep tech and applied AI attract the largest checks. Investors are moving away from speculative consumer apps toward capital-intensive, defensible businesses. Castelion, Etched and Higgsfield share a focus on proprietary hardware or large-scale infrastructure, which creates high barriers to entry. Even voice-to-text provider Wispr Flow, which raised $280 million, relies on custom models rather than generic APIs. The data suggests that the next wave of unicorns will be built on hard engineering and manufacturing, not just software distribution.



