Through July 2026, venture capital has poured $415 billion into 4,660 equity funding rounds across the United States. A tougher capital market is separating startups with traction from those running on hype. Defense tech has pulled in $12.3 billion in the first half of 2026—nearly double last year’s total—as capital rotates toward geopolitical security and autonomy.

Major deals in early July alone: Together AI raised $800 million in Series C. TwelveLabs raised $100 million in Series B. Quantum Systems, a drone and autonomy developer, closed a $1.2 billion Series D co-led by Blackstone and Airbus. These aren’t outliers—they’re the new median for serious AI infrastructure and applications.

Who’s Winning Now

Proof of traction beats vision. Early-stage capital is drying up unless founders can show real revenue, real users, or real benchmarks. Teams building for enterprise use cases (defense, aerospace, autonomous systems) are winning. Teams building consumer AI apps are fighting for breadcrumbs. The market no longer rewards “we’ll figure out the business model later.”

Hyperscale GPU clouds, autonomous drones, biotech, and enterprise software are eating venture capital. Defense spending is cyclical but massive—geopolitical uncertainty is structurally bullish for autonomy and edge AI. Quantum Systems raised $1.2 billion because drones matter now.

What’s Changed

VCs now ask: “How much capital do you burn per dollar of revenue?” earlier in conversations. Token economics and go-to-market strategy matter more than model architecture. Margins and unit economics are back in fashion after three years of “growth at all costs.”

AI infrastructure plays are hot. GPU clouds, data orchestration, model fine-tuning—if you’re selling picks and shovels to frontier labs, capital flows. But if you’re the 500th LLM wrapper, you’re fundraising in a desert.

FYI (keeping you in the loop)

What’s happened to consumer AI funding?

Dried up. VCs are skeptical of consumer AI without clear monetization. Copilots, AI assistants, and chatbots need obvious business models (subscription, licensing) or enterprise use cases to attract capital now.