AI Physical Security

Natural language becomes the dominant UI for physical security

The shift from rule-based alerts to natural language queries is now the defining product architecture across AI physical security. Conntour is building a Google-like NLP search engine for enterprise camera networks, while ARGU's Vision Agents enable live video analysis in natural language with zero model training. The week of July 6 saw $17.5B deployed across just 5 deals, with the capital clustering in platforms that abstract camera complexity behind conversational interfaces. This convergence on language-native video intelligence is compressing the operator learning curve from weeks to minutes.

Nvidia leads all investors with 28 deals in this theme, appearing in a $2.5B Series C (signal [14], alongside Sequoia, Lightspeed, JPMorgan, and B Capital at a $27.5B valuation) and a $200M Seed (signal [18], with Kleiner Perkins and a16z) and an $800M Series C (signal [24], with General Catalyst and Vista Equity). Meta, Google, and Amazon collectively account for 20 additional deals. This strategic concentration means physical security AI is being shaped by platform infrastructure players rather than sector-specialist VCs, aligning product roadmaps with hyperscaler compute ecosystems.

Why it matters · Startups backed by Nvidia or Meta gain preferential access to compute, silicon roadmaps, and distribution — creating a two-tier ecosystem that pure-VC-backed companies will struggle to match.

Series C rounds alone account for $7.6B across 9 deals over the last 90 days, while the 'unknown' category (largely mega-rounds) represents $16.5B. By contrast, seed captures only $2.5B across 7 deals. The July 6 week's $17.5B spike — the largest single week in the 90-day window — was driven by a handful of large structured rounds, not broad early-stage activity. This barbell structure reflects investor preference for de-risked, revenue-generating platforms like Flock Safety (valued at $8.4B) over greenfield bets.

Why it matters · Early-stage physical security AI founders face a more crowded and skeptical seed market; the path to large rounds now requires demonstrated enterprise traction, not just a prototype.

TRULEO's body-worn camera analytics and ShotSpotter's gunshot detection are embedding AI deeper into law enforcement workflows, while Prepared is transforming 911 call centers with AI dispatch assistance. Flock Safety's $8.4B valuation anchors the public safety sub-vertical as the most capital-efficient exit path in the theme. However, mounting regulatory pressure — analogous to signal [13]'s $1.4T Meta youth-safety penalties — signals that surveillance-adjacent products face increasing legislative risk, particularly around civil liberties and data retention.

Why it matters · Operators and investors in public safety AI must price in compliance costs and potential product restrictions that could truncate TAM in key jurisdictions.

Alta Ares, the French AI air defense startup focused on detecting, identifying, and intercepting drones and aircraft, represents a new sub-category attracting sovereign and institutional capital. The broader defense-tech signal is reinforced by Axon Enterprise's presence in the company set and the week of June 22 delivering $5B across 8 deals — the second-largest weekly deployment in the 90-day window. As drone proliferation accelerates globally, AI-native detection platforms are being evaluated as critical infrastructure rather than niche defense contracts.

Why it matters · Investors with defense mandates should watch French and Israeli startups (Alta Ares, ARGU) as NATO-aligned government procurement opens new non-VC revenue channels.