Defense Tech: what are the biggest challenges now?

Why is defense tech booming while militaries still struggle to get new weapons?

Defense tech currently has plenty of money and plenty of promising technology; the real shortage is the ability to turn those technologies into weapons that armies can buy, produce and update fast enough.

Private capital has moved into defense at a scale that would have looked unusual only a few years ago. PitchBook data compiled by Defense News put defense-tech and dual-use venture funding at roughly $49 billion in 2025, up from around $27 billion the previous year. Anduril subsequently raised $5 billion at a $61 billion valuation after reporting about $2.2 billion in 2025 revenue. Helsing raised €1.7 billion at a €12 billion valuation, while Shield AI raised $1 billion at a valuation above $5 billion before later funding pushed its valuation considerably higher.

Money has become a weak explanation for why promising defense products still move slowly.

The contrast with procurement is brutal. In its latest annual assessment of 104 of the Pentagon's most expensive weapon programs, the U.S. Government Accountability Office found that the average time required to provide an initial capability had climbed beyond 12 years. Even the Pentagon's Middle Tier of Acquisition pathway, which is supposed to field or prototype systems within five years, has admitted programs whose underlying technology still needed substantial development.

NATO is reacting to the same problem from another direction. It recently launched the NATO Front Door to make opportunities easier for companies to find and the NATO Engine to connect defense startups with factories that have spare manufacturing capacity. NATO explicitly says production scale remains a major hurdle for newer suppliers.

Defense companies can now raise billions and develop surprisingly capable prototypes. Governments still struggle to buy those systems quickly, manufacture them in large quantities and change them again when the battlefield moves on.

What is moving faster	What is still moving slowly
Private defense funding	Government procurement
AI and autonomous-system development	Testing and certification
Drone design cycles	Large-scale manufacturing
Startup formation	Recurring production contracts
Battlefield adaptation	Upgrades to established weapon programs

If you want more recent data on this point, please see our latest defense tech market report.

Is Pentagon procurement still too slow for modern defense tech?

Pentagon procurement is still painfully slow compared with the speed at which software, drones, electronic warfare and autonomous systems are changing.

The newest GAO assessment makes the gap unusually clear. Major U.S. weapon programs now take more than 12 years on average to reach an initial capability. GAO also found delayed milestones among some of the Pentagon's costliest programs and concluded that recent acquisition reforms have still failed to produce the speed policymakers wanted.

Twelve years is especially difficult to defend in technologies whose useful configurations can change several times in a single year. An electronic-warfare system can face new frequencies. A drone can encounter new jammers. An AI model can be surpassed. A communications architecture can acquire new vulnerabilities.

The Pentagon knows this. Its modern software-acquisition push openly says that traditional hardware-centered procurement has been too slow for software-defined warfare. Faster contracting mechanisms and dedicated software pathways are supposed to shorten those cycles.

Creating a faster pathway does not automatically produce faster weapons, though. GAO found that some supposedly rapid programs entered the process with immature technology and then lost time solving technical problems that should have been addressed earlier.

For software-heavy weapons facing adversaries that iterate every few months, even a five-year “rapid” acquisition cycle can be far too slow.

Can defense startups and Western militaries actually manufacture enough weapons at scale?

Western militaries still cannot manufacture enough of many key systems at the pace a long high-intensity war would demand.

Moving from a few dozen impressive prototypes to thousands of reliable weapons remains one of the industry's hardest problems. Prototype economics are forgiving. Engineers can hand-assemble units, buy expensive components in small quantities and personally solve unexpected problems. A military production line has to deliver the same performance thousands of times while controlling cost, quality, spare parts, component substitutions and delivery schedules.

That transition requires factories, but it also requires people who know how to run them. Defense manufacturing depends on machinists, welders, quality engineers, explosives specialists, production managers, software engineers and technicians working under unusually demanding security and certification rules.

The largest startups have started spending accordingly. Anduril's Arsenal manufacturing strategy is explicitly designed around high-volume autonomous systems. Helsing has expanded from software into physical weapons and drone production. New missile companies are discussing output in thousands per year rather than dozens.

Yet missile and ammunition output cannot be increased simply by adding another shift at the final assembly plant. Production depends on solid rocket motors, energetic materials, seekers, guidance electronics, castings and other components that may come from only a few approved suppliers. Increasing one part of the chain achieves little if another remains fixed.

NATO's newest initiatives show how serious the bottleneck has become. The NATO Engine was created specifically to match companies lacking manufacturing capacity with factories across allied countries. NATO's secretary general has also said no single member has enough industrial capacity to meet growing demand for some critical capabilities, particularly air defense and strike weapons.

Governments are responding with larger and longer-term orders, which give suppliers a reason to invest in new capacity. But a budget can rise much faster than a factory, a rocket-motor line or a trained workforce can be built.

The next generation of serious defense companies will be judged much more on output, lead times and supplier depth than on how impressive their latest demonstration looks.

Why are cheap drones causing such an expensive defense problem?

Cheap military drones are creating a nasty economic problem because destroying the threat can cost tens or hundreds of times more than launching it.

An expensive interceptor can be exactly the right tactical response to a drone heading toward a refinery, radar or ammunition depot. The problem appears when an opponent repeats that attack hundreds of times.

A high-end surface-to-air missile may cost hundreds of thousands of dollars or several million dollars. Small one-way attack drones can sit far below that price range. Even larger systems such as Shahed-type drones remain cheap enough to make repeated missile interception economically uncomfortable.

That has pushed armies toward layered counter-drone defenses. Electronic warfare works when the drone still depends on vulnerable navigation or communications. Guns can produce a better cost exchange at short range. Interceptor drones promise cheaper engagements. Directed-energy weapons could eventually offer very low marginal shot costs where power and atmospheric conditions permit them.

No single approach covers every threat. Jamming loses effectiveness against systems that navigate independently or use fiber-optic control. Guns have limited engagement envelopes. Lasers still face practical constraints. Cheap interceptors need enough accuracy to avoid wasting several defenders on each incoming drone.

The useful metric is cost per successful defensive effect. A technically excellent interceptor can still be economically unsustainable if an attacker can keep sending much cheaper targets.

Counter-drone approach	Why militaries want it	Main problem
High-end missiles	Excellent reach and probability of kill	Expensive, limited magazines
Electronic warfare	Very cheap repeated engagements	New navigation and control methods can bypass it
Guns	Affordable ammunition	Shorter range
Interceptor drones	Better cost exchange against mass attacks	Detection and guidance still need to work reliably
Directed energy	Extremely low potential cost per shot	Power, weather and maturity

If you want more recent data on this point, please see our latest defense tech market report.

Are electronic warfare and rapid countermeasures making military drones obsolete?

Electronic warfare is forcing military drones to evolve extremely quickly, and specific drone configurations can become outdated within weeks or months.

Ukraine gives us the clearest evidence. Radio jamming has repeatedly degraded ordinary remotely piloted drones. GPS interference has made satellite navigation unreliable in heavily contested areas. Yet drone use has continued to expand.

Designers keep working around each countermeasure. Operators move to different frequencies or frequency-hopping radios. Navigation shifts toward inertial systems and computer vision. Fiber-optic drones avoid radio-frequency jamming almost entirely by carrying a physical data cable behind the aircraft. Greater onboard autonomy can reduce the communication a drone needs during its mission.

Russia's Geran family shows the same process at industrial scale. Recent technical analysis by CSIS describes continuing changes to communications, navigation, propulsion, warheads and attack tactics as Ukrainian defenses evolve. Successful modifications can move into production quickly enough that defenders repeatedly encounter new versions of what appears superficially to be the same drone.

That cycle can run on the scale of weeks. A frequency works until the enemy learns to jam it. A flight profile works until defenders understand it. A navigation system performs well until interference increases.

This creates a problem for traditional military qualification. Long testing protects soldiers from unreliable equipment, but freezing a configuration for too long can leave troops with a thoroughly certified system that has already lost much of its battlefield edge.

The companies best suited to this environment will be able to discover why systems are failing, modify them and push the changes into production without waiting years for another major upgrade.

Can military AI actually be trusted in combat today?

Military AI can already handle narrow combat tasks, while general-purpose autonomous systems making messy battlefield decisions still go well beyond what the evidence supports.

There is plenty of real deployment behind the hype. AI is already used for sensor processing, object detection, route planning, navigation, target prioritization and decision support. Shield AI has pushed its Hivemind autonomy software into military aircraft programs, while the U.S. Air Force's Collaborative Combat Aircraft work is testing much deeper autonomy in uncrewed combat aircraft.

The difficult part begins when conditions stop resembling the testing environment.

War produces damaged sensors, deliberate camouflage, spoofing, communications loss, adversarial electronic warfare, new objects and incomplete information. An AI system can therefore face inputs its designers never encountered. The enemy also has an incentive to discover exactly what causes the model to fail.

That makes military testing much harder than proving that an autonomous vehicle completes a prepared demonstration. Armed forces need to know what happens at the edge of the system's competence and whether humans can recognize when that edge has been reached.

There is also an integration question. The U.S. Air Force has tested autonomy software from one supplier on an aircraft from another, hinting at a market where governments want autonomy that can move between platforms rather than staying permanently tied to one manufacturer's aircraft.

For now, bounded autonomy is the convincing part of military AI. Broader battlefield judgment remains a much higher bar.

Can modern weapons still work when GPS, communications and networks are jammed?

Too many modern weapons still lose a large share of their capability when navigation or communications disappear, making electronic resilience one of today's hardest defense-tech tests.

Modern militaries have spent decades connecting everything. Sensors share information. Drones receive commands. Aircraft exchange targeting data. Soldiers use satellite navigation. AI systems pull information from networks. That connectivity produces enormous advantages when the network survives.

An opponent naturally tries to break it.

Ukraine has provided repeated evidence of heavy GPS interference and communications jamming. Systems that depend too heavily on one navigation source or a continuous control link can lose effectiveness quickly in those conditions.

Defense companies are responding with inertial navigation, visual navigation, terrain matching, mesh communications, alternative positioning systems and more onboard computing. Greater local autonomy also lets a drone or vehicle continue part of its mission when communication with an operator disappears.

Cyber resilience belongs in the same discussion. A connected weapon has to assume that an adversary will try to compromise its software, corrupt information, exploit interfaces or interfere with update systems.

The best systems currently aim for graceful degradation: losing GPS or one network node should reduce capability rather than make the weapon useless.

Is China still a dangerous supply-chain dependency for Western defense tech?

Western defense companies remain exposed to Chinese supply chains several layers below the final weapon, especially in critical minerals, magnets, batteries and electronics.

Gallium is an extreme example. The latest U.S. government assessments put China's share of global primary gallium production at roughly 99%. Gallium-based semiconductors are important for high-frequency and high-power electronics used in radar, communications and other military systems.

Rare-earth processing is another concentration point. Permanent magnets made with rare-earth elements go into motors, actuators and many other high-performance systems. China holds a dominant position in several stages of that supply chain.

The awkward part is that “made in America” or “made in Europe” does not necessarily mean the supply chain is Western. A defense company can assemble the final drone in Texas or Bavaria while depending on magnets, processed minerals, battery materials or electronic components originating much farther away.

Commercial technology makes the issue more complicated. Startups often rely on commercial parts because they are cheaper, improve quickly and are available without waiting for a custom military component. That can produce far better economics, but those supply chains were generally optimized for cost and speed rather than wartime independence.

Diversification also takes time because alternative suppliers have to be qualified. A missile manufacturer cannot simply swap in a new semiconductor or material during a crisis and assume performance remains identical.

Western governments are reducing the exposure, but it is still substantial today.

If you want more recent data on this point, please see our latest defense tech market report.

Has defense-industry consolidation left too few companies able to build critical weapons?

Defense-industry consolidation has left parts of the Western industrial base dependent on a surprisingly small number of companies and specialist suppliers.

The U.S. Department of Defense has previously calculated that the number of major aerospace and defense prime contractors fell from 51 in the 1990s to five. That alone does not prove every market is uncompetitive, but it shows the scale of consolidation.

The more worrying concentration can sit deeper in the supply chain. Two rival missile primes can still rely on the same rocket-motor producer. Several aircraft companies can depend on the same casting, microelectronics or energetic-material supplier. Competition at the logo level therefore does not always mean genuine industrial redundancy.

Today's defense startup boom could improve this. Anduril, Shield AI, Saronic, Helsing and a growing group of missile, drone and software companies are giving governments credible alternatives in areas that were previously dominated by a handful of incumbents.

Their survival depends on contracts large enough to build real supplier networks. Procurement rules, however, still favor companies that already possess secure facilities, certified processes, established production lines and decades of contracting experience.

A healthier market would have several viable manufacturers and several qualified suppliers for critical components. Western defense is still well short of that in some strategically important areas.

Why can't NATO countries just buy the same defense technology together?

NATO still operates more like 32 connected defense markets than one truly integrated market.

The alliance can agree on the threat while individual countries still choose different vehicles, communications equipment, missiles, drones and software. National procurement law, domestic industrial policy, export controls and certification requirements all shape those choices.

The result is expensive fragmentation. Ten countries buying ten slightly different versions of a system create smaller production runs, more engineering work, separate spare parts and more complicated training. Companies also have to navigate multiple procurement organizations to reach what should theoretically be one enormous allied market.

NATO has become much more explicit about fixing this. Its latest industry strategy pushes open architectures, digital standards, modularity and mutual recognition of certifications. The alliance has also launched a Front Door for Industry, alongside its first consolidated public demand signal, so companies can see capability needs without navigating as many separate entry points.

Interoperability matters in combat as well. A Polish sensor should ideally be able to send useful targeting data to a German command system, which can then cue an American or European interceptor without awkward translation between incompatible networks.

NATO is moving toward that model, but allied procurement remains far more fragmented than the size of the alliance would suggest.

Fragmented market	More integrated allied market
Small national orders	Larger common orders
Country-specific versions	More shared configurations
Repeated certifications	Greater mutual recognition
Proprietary interfaces	Open architectures
More suppliers to manage per country	Wider competition across the alliance

Why do good defense startups still get stuck after successful military tests?

A successful military test does not guarantee the recurring procurement contract a defense startup needs to become a real supplier.

This gap is usually called the valley of death, and it remains one of the industry's most stubborn problems.

Innovation programs have become quite good at finding interesting technology. Startups can win accelerator places, prototype contracts and military exercises. The dangerous period arrives when the demonstration ends.

Someone inside the military then has to decide that the capability deserves a permanent role, find money in the right budget, define how units will use it, integrate it with existing systems and place an order large enough to support production. Those steps can involve different organizations and different fiscal years. A military user can love the product without controlling the budget that buys it.

NATO's newest innovation policy shows how seriously governments are taking this transition. Its Rapid Adoption Action Plan and Innovation Scale-Up Package focus heavily on clearer demand, contracting pathways and manufacturing support.

Huge venture rounds soften the problem for the best-funded startups because they can survive longer without major procurement revenue. But a company valued at $10 billion, $20 billion or $60 billion eventually needs enormous recurring sales. Prototype contracts worth a few million dollars will not support that kind of business.

The valley of death has shifted upward: more companies can finance the prototype stage, while dependable government demand still decides who becomes a lasting industrial supplier.

If you want more recent data on this point, please see our latest defense tech market report.

Is defense tech becoming too expensive to fight wars at scale?

High-end defense technology is becoming too expensive to use everywhere, and recent wars are exposing how difficult it is to sustain premium systems in very large quantities.

Advanced fighters, satellites, precision missiles and air-defense interceptors can do things cheaper systems simply cannot do. The problem appears when governments need hundreds or thousands of them.

Drone warfare makes the contrast particularly visible. Militaries can lose relatively cheap unmanned systems repeatedly and keep operating. Losing an exquisite aircraft, ship or missile battery carries a completely different financial and strategic cost.

That is pushing forces toward a wider high-low mix. Expensive weapons remain necessary for the hardest missions, while cheaper missiles, drones and sensors handle tasks that do not justify consuming scarce premium inventory.

NATO now uses almost exactly that logic in its industrial policy. Its Innovation Scale-Up Package says allies need to balance high-end systems with less expensive products that can be deployed and manufactured at scale.

“Cheap” can still be misleading. A $20,000 drone that rarely works under jamming can deliver worse economics than a $70,000 drone with much higher mission success. A low-cost missile that requires three shots per target can also be worse value than a more expensive one-shot interceptor.

Cost per useful effect is a better measure than sticker price alone.

Can defense tech adapt fast enough when the enemy copies or counters it?

Defense technology increasingly has to improve continuously while somebody is actively studying how to defeat it.

A successful drone encourages better jammers. A new jammer encourages different frequencies, fiber-optic control or greater autonomy. A better radar encourages lower flight profiles and new signatures. Improved air defense encourages decoys, saturation attacks and different routes.

That feedback loop can move frighteningly fast. Ukraine's drone ecosystem and Russia's continuing Geran modifications show how quickly both sides can convert battlefield experience into design changes. A weapon that looked highly effective a few months earlier can become much less useful once the adversary learns its patterns.

Winning the first version therefore matters less than maintaining the faster learning loop. Companies need battlefield data, engineers who can understand failure modes, factories flexible enough to incorporate changes and procurement rules that allow updated configurations to reach troops quickly.

Software-defined systems, modular payloads and open architectures become especially valuable under those conditions because they make repeated modification easier.

This is where defense startups can have a genuine advantage, provided their smaller size actually turns into a faster path from battlefield failure to a corrected production batch.

What are the biggest defense-tech challenges now?

The biggest defense-tech challenge today is scaling useful technology faster than the battlefield can make it outdated.

Procurement is still painfully slow. GAO's newest assessment puts the average U.S. major weapon program beyond 12 years to initial capability, even after years of acquisition reform. At the opposite extreme, drone and electronic-warfare configurations can change within weeks.

Manufacturing creates another major constraint. NATO has recently launched an entire factory-matching initiative because promising suppliers still struggle to reach the production volumes allied militaries require. Bigger defense budgets help, but factories, rocket motors, energetic materials, skilled workers and qualified suppliers cannot be created instantly.

Affordability comes next. Recent wars have shown that technically excellent weapons can still create bad economics when expensive interceptors are exchanged against cheap drones or when exquisite systems cannot be replaced at the rate combat consumes them.

Supply chains and interoperability broaden the problem further. Western weapons still depend on highly concentrated critical-mineral and electronics supply chains, while allied militaries continue buying systems that do not always connect cleanly across national boundaries.

Military AI deserves serious attention, but insufficient AI capability does not currently look like the main thing holding defense tech back. Private financing certainly does not either. Billions of dollars are already flowing into the strongest companies.

What remains scarce is industrial execution: weapons that work under jamming, connect to existing forces, cost little enough to use in quantity, can be built by the thousand and can change quickly after the enemy learns how to defeat them.

That is the standard defense tech has to meet now.

If you want more recent data on this point, please see our latest defense tech market report.

Who is the author of this content?

NEW MARKET PITCH TEAM

We track new markets so founders and investors can move faster

We build living "market pitch" documents for emerging markets: AI, synthetic biology, new proteins, and more. Instead of outdated PDFs or hallucinated LLM answers, our clients get a clean, visual, always-updated view of what's really happening: key players, deals, regulations, and signals that matter. Learn more about us.