EXCLUSIVE: Defense tech funding surge signals shift in how security is built

Venture capital is moving into defense and national security at a pace that would have seemed unlikely just a few years ago. By mid-2025, funding in the sector had already reached $28.1 billion, surpassing the previous year’s total. The shift reflects a deeper change in how governments think about technology, speed and risk.

John Larson, president and chief AI officer at Babel Street, sees this as a structural turning point rather than a temporary spike. In his view, the driver is not only geopolitical pressure, but the nature of modern conflict itself.

“First and foremost, this trend is being driven by the fundamental shift of warfare and the growing recognition of the benefits of commercial technologies in this fight. We are operating in an AI-on-AI era now. Threat actors are exploiting publicly available information, flooding the environment with synthetic media and automated deception, and doing it a machine speed. Matching that requires AI capabilities at the same scale.”

That shift is forcing a reassessment inside government. Traditional timelines for procurement and development are increasingly out of sync with the speed at which threats evolve — and with the pace of innovation in the private sector.

“What’s also changed is the investor calculus. Innovation in AI is concentrated among venture and PE-backed firms, and both the Pentagon and the intelligence community have woken up the fact that the gap between adversarial agility and government modernization timelines is a real vulnerability.”

Signals from policymakers suggest that awareness is translating into action. Larson points to a more direct engagement with commercial players, from procurement reform to dedicated innovation units.

“Secretary of War Pete Hegseth has said he intends to supercharge innovation, and that the Pentagon is ‘done running a peacetime science fair while our potential adversaries are running a wartime arms race.’ That kind of candor signals systemic change.”

In practical terms, that means a shift away from a model where governments build everything internally, toward one where private firms act as testing grounds for critical capabilities.

“The surge in non-traditional defense contractor capabilities (PE- and VC-backed) reflects a sharp shift in where investors see both urgency and opportunity converging. Geopolitical instability, AI-driven threats, infrastructure and supply chain vulnerabilities are no longer isolated risks. They compound each other, and they’re moving faster than traditional procurement models were ever built to handle.”

There is precedent for this kind of integration between civilian and military innovation. Larson notes that rival powers have already built systems that rely heavily on commercial technologies.

“There is a precedence for this kind of military-civil fusion, and notably, one of our greatest adversaries has mastered it. China’s People’s Liberation Army has a tremendous track record of harnessing commercial scientific and technological developments to advance its military and economic might.”

For governments, the appeal is clear: faster iteration, broader competition and less direct exposure to the trial-and-error of early-stage technology.

“The government and taxpayers win as PE- and venture-backed companies become the new proving grounds for critical technologies for intelligence, threat detection, and geopolitical risk analysis. Increased market competition yields a larger number of better solutions at a faster pace. The trial-and-error that is inherent in technology development sits with the vendor, not the government.”

At the operational level, the most visible change is how intelligence itself is produced. The volume and speed of data have outgrown traditional analytical methods, particularly in an environment shaped by disinformation.

“In the past, investigators manually scraped, collected, filtered, and made sense of publicly and commercially available data to find the signal in the noise. However, the surface area, speed, scale, and ubiquity of threats are expanding at an unprecedented rate, and risk is compounding faster than humans can keep up.”

Compounding that challenge is the rise of synthetic content and automated deception, which makes verifying information as important as collecting it.

“Additionally, we are now in an environment where AI has industrialized deception; threat actors are exploiting publicly available information and flooding the environment with synthetic media and automated deception where the very veracity of global intelligence must be questioned.”

Larson describes a shift toward what he calls “agentic risk intelligence”, where AI systems continuously process data and surface relevant insights for human operators.

“In the new age of agentic risk intelligence, AI agents continuously ingest massive global data volumes, link entities, detect anomalies, and synthesize evidence for analysts who are increasingly inundated. Human-controlled AI agents then surface relevant signals so operators can move faster to judgment and action, instead of fiddling with filters.”

The emphasis, however, remains on augmentation rather than replacement.

“The key is humans remain in control. AI augments but does not replace operators, analysts, and investigators, it takes on the challenge of sifting through overwhelming volumes of data and disinformation.”

That balance between speed and oversight becomes even more critical as private companies take on a larger role in national security functions. For Larson, the central issue is not capability, but trust.

“As with any use of AI, and especially for homeland security and law enforcement, trust and accountability are critical foundations and data provenance is non-negotiable. Every AI generated lead and insight must be traceable back to source data, so evidence can stand up in court and be verifiable. That’s a high bar, and it should be.”

Taken together, the picture is less about a single trend and more about a reconfiguration. Capital is flowing into defense not just because the sector is growing, but because the underlying model is changing from slow, centralized systems to faster, distributed ones shaped by private innovation.