Matt Grimm on Defense Procurement, Anduril and the Cost of Autonomous Warfare

In this November 2024 interview, Anduril co-founder and COO Matt Grimm discusses defense procurement, autonomous systems and company building with Harry Stebbings. Grimm is an executive of a defense contractor and his policy arguments also reflect Anduril’s commercial interests.

Grimm says he is a lifelong Democrat who donated to Kamala Harris and other Democrats, and personally wished the election had gone differently. Nonetheless he sees possible opportunities for defense innovation under the incoming Trump administration.

He describes Anduril internally as mission-driven rather than partisan. He expects the company to operate under either party and hopes new appointees will bring urgency, more fixed-price contracting and investment in the industrial base. These are hopes about a future administration, not established outcomes.

On China and Taiwan, Grimm says it is too early to predict specific policy changes. He wants the US and its allies to reduce supply-chain dependence on China and argues that deterrence could prevent an invasion; he does not claim to know what will happen.

He says China sanctioned him and other Anduril executives over sales of missiles to Taiwan. He describes learning of it from a news report and interprets the action as evidence of China’s ambitions, an assessment from a participant with a direct commercial interest.

Grimm is optimistic about US semiconductor manufacturing, citing early reports of yields at TSMC’s Arizona facility while acknowledging that the longer-term outcome remains to be seen. His comments are not an independent assessment of the plant.

He explains supply-chain tiers through a phone: components contain chips, which in turn require materials and processing from further down the chain. He argues that reliance on Chinese mining, refining and manufacturing is difficult to remove entirely, and proposes alternative supply or stockpiles.

Grimm lists possible escalation scenarios, including Taiwan, North Korea, Russian expansion and disrupted shipping in the South China Sea. Anduril’s stated objective, he says, is deterrence by equipping the US and allies with capabilities that make conflict more costly.

On Russia’s war in Ukraine, he explicitly labels his suggestion of a negotiated settlement in the following year as conjecture and says he has no inside information. He imagines one possible outcome involving Russian control of territory near the then-current lines; he acknowledges resolution would be difficult.

Discussing Israel and Gaza, Grimm describes the conflict as tragic, supports Israel’s response, and predicts extensive Israeli control of Gaza. These are his political views and forecasts, not an independent assessment of the conflict or its humanitarian and legal questions.

Asked whether war is good for business, he says Anduril’s goal is to deter war and describes an unused stockpile as the ideal. His company nevertheless sells defense equipment and has a commercial stake in procurement.

He argues that software and AI can affect production, maintenance and battlefield awareness. Anduril’s Lattice system, he says, aims to combine information from many sensors into a shared picture for soldiers and commanders; these are company descriptions, not independent performance results.

The company considers existing capabilities and the military services’ published planning documents when identifying gaps. Stebbings asks whether following government plans risks building only a faster version of what already exists.

Grimm replies that military officers and civil servants provide continuity even as political appointees change. Plans may lag fast-moving technology, however, so Anduril must both understand customer requirements and propose new capabilities.

Venture funding gives Anduril room to build and demonstrate prototypes before asking the government to buy, he says. A working system that interfaces with existing equipment can make a more persuasive case than a presentation seeking funding to begin development.

Stebbings asks whether government buyers understand new technology. Grimm rejects the idea that procurement officers are unintelligent: unlike venture investors, they must reliably supply troops and cannot tolerate the same rate of failed bets.

He still wants procurement incentives to change. His advice to a supplier is to understand the buyer’s responsibilities and demonstrate how a technology would be fielded and supported, recognizing the difficult gap between a promise and implementation.

His proposed reforms begin with greater urgency about geopolitical risks, more competition and less cost-plus contracting. Under his simplified description of cost-plus, a contractor recovers costs and receives a margin; he argues this can reward spending more rather than making equipment cheaper.

Grimm cites figures for shareholder returns and says Lockheed Martin spent $6 billion on buybacks and $1.5 billion on internal research and development in 2023. These are his figures and do not independently establish the full economics of defense contracts or all research spending.

He argues that traditional procurement can deliver equipment based on requirements written many years earlier. He estimates some programs take 15 to 20 years from requirements to full deployment, while modular hardware and software updates could shorten cycles; fielding, training and maintenance still take time.

Grimm does not expect defense to be a winner-take-all market. He cites large 2024 defense budgets across the US, UK, EU and Australia, while acknowledging that salaries, healthcare and other expenditure mean the total is not Anduril’s addressable market.

He wants Anduril to become another major prime contractor alongside existing companies, rather than displace them all. He also advocates greater and more efficient UK and European defense spending; these are an executive’s policy preferences.

Governments already favor domestic providers, he says. Anduril’s approach is to adapt core technology to local standards and invest in local teams, manufacturing and support so customer countries can retain industrial capability.

Asked about research spending, Grimm roughly estimates a very high share of Anduril’s budget goes to R&D, without giving a precise audited figure. He says the growth-stage company reinvests revenue and remains EBITDA-negative at the time of the interview.

He says incumbents have strengths worth learning from: government relations, long-term maintenance and logistics, and recruiting people motivated to serve. He concedes that fielding and supporting hardware around the world is harder than updating software in the cloud.

One regret is not taking a firmer position on remote work earlier. Anduril has an office-first approach, he says, because hardware requires physical production, although the pandemic made it difficult to bring dispersed employees back together.

Another mistake was trying to constrain the growth of support functions through automation. Grimm says underinvestment in business systems and personnel hampered growth, and that the company is now investing in an internal operating system called Arsenal OS.

Scaling production requires unglamorous systems for procurement, supply planning, customer pipelines and financial reporting. With hindsight, Grimm says he should have invested in those people and tools earlier.

He distinguishes reflective regret from being immobilized by it. He considers himself a harsh critic, but says leaders can celebrate a successful demonstration and still identify what the team should improve next time.

On fast decisions, Grimm says waiting for genuinely useful information tomorrow can be sensible. His objection is to months of analysis and consensus-building where a quicker decision would allow a team to learn and adjust.

He estimates Anduril had raised about $4 billion excluding debt and around $5 billion including debt. Early fundraising was led by Trey, with Grimm handling the mechanics; more recent rounds were led by the CFO. These are his contemporaneous estimates.

Some R&D efforts failed, he says, but he does not believe the company could have achieved the same result on half the capital. In hindsight, he would have stopped some projects earlier and reused useful components elsewhere.

Grimm supports giving employees a choice to sell some shares when that would improve their family’s financial security. He says roughly 30% of shares eligible in structured tenders were actually sold, a share-based figure rather than the percentage of employees participating.

He criticizes opaque secondary-market SPVs, claiming some brokers use inaccurate information, layered fees and misleading descriptions of access. He says Anduril can control direct access to its cap table and information but cannot control every outside representation; the specific alleged misconduct is not independently verified in the interview.

Asked what made Anduril successful, Grimm credits its ability to attract both defense-oriented technical talent and investment. He stresses the people doing the work; Stebbings adds that the founders brought complementary product, execution, fundraising and government-sales skills.

Grimm says he wants employees of all backgrounds to feel accepted, motivated and rewarded, while opposing hiring to hit a numerical target. He estimates women accounted for roughly 18–20% of employees at the time; the figure is his estimate.

He worries some newly funded defense startups will discover that execution is harder than it looks. Beyond a promising product, winning and delivering a long-term government program requires budget knowledge, compliance, accreditation and support.

Grimm jokes that investors should buy Anduril shares, a recommendation in which he has an obvious interest. He cautions that investing in defense requires understanding procurement and that a diversified portfolio may be more appropriate than treating defense as an entire asset class.

He says Anduril’s acquisition strategy combines promising technology with its sales, contracting and operational infrastructure. He cites Dive Technologies, an autonomous-submarine company, and describes local hiring in Australia to support a Royal Australian Navy development program.

Grimm attributes his execution style to fast decisions, clear communication, continual learning and hiring people who care about quality. He describes an early workout and a long day at the office, followed by bedtime with his children.

He rejects the idea that one can have everything without sacrifice. If he had more time, he says, he would spend it with his children while they are young.

For marriage, he emphasizes choosing a partner carefully and communicating expectations, responsibilities and goals. He disagrees with the rule never to sleep angry: a pause can help people return to a disagreement thoughtfully, provided they actually address it later.

Grimm urges Stebbings to be near the concentration of technology talent in California, while acknowledging exceptions elsewhere. He argues that venture capital had spent too much attention on quick trends and hopes it will return to more durable technological innovation.

Stebbings worries that investors trained on software may struggle with capital-intensive businesses, recalling the poor returns of some clean-tech investments. Grimm expects a mixed outcome but believes investors can learn new models.

If raising again from one investor, Grimm chooses Founders Fund, praising its support when defense investing was controversial. The firm is an existing Anduril investor, so his endorsement reflects that relationship.

He returns to the importance of integrated business systems for manufacturing. A technician’s daily task depends on the coordination of inventory, planning and software behind the scenes, an area he says he had previously underestimated.

From Peter Thiel, Grimm says he learned that there is no investment advantage in merely repeating consensus. He distinguishes useful contrarian thinking from being disagreeable for its own sake and cautions that some early defense valuations may make returns difficult.

He says Anduril’s last round valued it at $14 billion post-money and considers it underpriced. Reaching a hypothetical $100 billion valuation would require production at scale and strong execution, he argues; this is his forecast, not a valuation assurance.

Grimm is skeptical about armed robot dogs specifically, but expects more robot-on-robot warfare. He challenges the notion that advanced Western systems will necessarily dominate low-cost opponents.

He compares his rough estimates for inexpensive Iranian Shahed drones with substantially more costly interception missiles and warns of an unfavorable cost and production curve. The figures are illustrative estimates from a defense executive and are not independently verified here.

Anduril’s thesis, he says, is to lower the cost and increase production of defensive systems and munitions. He argues Western production would struggle in a prolonged major conflict; both the assessment and proposed remedy reflect the company’s market position.

Grimm calls China the West’s most difficult long-term adversary, citing human rights, production capacity and supply chains. He alleges genocide against Uyghurs and condemns restrictions on expression and religion; these are his claims and political judgments, not findings established by this interview.

He favors a US TikTok ban, arguing that a strategic rival should not control a channel shaping young people’s information. This is his policy position; the interview does not establish the extent or effects of any particular influence operation.

On free speech, Stebbings raises the possibility that a joke could incite violence. Grimm opposes prior state control of opinion while accepting that speech may carry reputational or legal consequences; the two disagree on where the boundary should sit.

Grimm sees structural problems and a lack of ambition in parts of Europe but rejects a wholly bleak view. From his American perspective, he says Europe still has a future worth building and defending.

Asked about formative childhood experiences, Grimm recalls financial instability and arguments between his parents in suburban Philadelphia. At 14, he says, he decided he wanted a different financial life for his own family.

He believes money offers freedom and options but does not itself create happiness. Buying his childhood dream BMW M3 brought excitement but not lasting fulfillment; he places more weight on relationships, friends and family.

If failure were impossible, Grimm says he would start a company to bring semiconductor fabrication back to the US. He thinks Anduril may be his last company and hopes eventually to build it into a successful public business, a future ambition rather than an announced outcome.