March 30, 2026

Palantir Technologies’ CTO Shyam Sankar’s Mobilize: How to Reboot the American Industrial Base and Stop World War III is, on its surface, a book about defense procurement, about the slow death of the arsenal system, the consolidation of the defense industrial base into five prime contractors, and the catastrophic gap between the weapons America says it can build and the weapons it actually delivers.

Anyone who has spent a career inside a large institution, any large institution, will recognize something deeper in these pages, something uncomfortably familiar.

The disease Sankar diagnoses is not unique to the Pentagon. The same disease afflicts every organization that was once revolutionary and has since become the thing it originally displaced.

The Architecture of Forgetting

A pattern repeats across industries, across decades, and across continents. An organization is founded to solve a hard problem, solves the problem, grows, and succeeds. Gradually, imperceptibly, the organization begins to optimize not for the problem it was built to solve but for its own continuation.

The metrics shift from what was built to what was reported, from “does it work” to “did we follow the process.” The organization begins to mistake the scaffolding for the building.

Sankar describes the pattern in the context of defense acquisition. The DoD’s procurement system, originally designed to ensure accountability and prevent waste, has become the single greatest impediment to fielding new capability. A system meant to serve the warfighter now serves itself. Programs requiring three years consume fifteen. Technologies cutting-edge at contract signing are obsolete by first delivery. Seventeen layers of review boards, milestone gates, and compliance checkpoints separate the people who understand the problem from the people authorized to solve it.

The forgetting is sometimes literal. Sankar recounts the case of Fogbank, a complex aerogel crucial to activating the second stage of nuclear warheads on Trident submarine-launched ballistic missiles. When the National Nuclear Security Administration attempted to refurbish the warheads in the early 2000s, the program stalled for years because no one remembered how to manufacture the material. The knowledge had departed with the people who possessed it. No document and no process manual could reproduce the experience of having built the thing. Andy Grove warned about this dynamic in his final major essay, arguing that American elites fetishized startups as engines of growth while ignoring that startups serve the function only when they scale into mature companies making real things on American soil. When mass production departs, Grove wrote, the “chain of experience” enabling subsequent innovation breaks irreparably.

The loss accelerates when institutions begin to treat domain expertise as a cost rather than an asset. Experienced workers carrying decades of accumulated knowledge command higher compensation and tend to resist directives conflicting with what the experienced workers know to be true. Younger cohorts cost less and comply more readily. The exchange appears rational on a spreadsheet. After all, senior compensation declines, organizational friction diminishes, and the hierarchy encounters fewer challenges from below. The growing institutional faith that artificial intelligence will compensate for whatever expertise departs with the experienced workforce recapitulates the very logic the Fogbank disaster exposed. Knowledge born from the experience of building cannot be extracted into a system and retrieved on demand. The knowledge lives in the hands and judgment of the people who acquired it through years of practice and the knowledge leaves when the people leave. Somehow the executives always stay.

The defense industrial base did not set out to become rigid. The base became rigid because the processes designed to prevent failure also prevented success and no one with sufficient authority had sufficient incentive to say so.

The dynamic is not a Pentagon story but the story of every legacy institution waking up one morning to discover it can no longer do the thing justifying its existence.

The Heretic’s Position

Inside every such institution, people see the dysfunction. The observers are found in the labs, the factory floor, the field offices, and the technical ranks where the actual work either happens or does not, rarely in the executive suite. The gap between the press release and the product is visible from the technical ranks, as is the distance between the strategy deck and the shipping date, and between what leadership celebrates and what the customer actually receives.

The observers face a choice Sankar’s book illuminates without quite naming it: the heretic’s dilemma.

To stay silent is to be complicit in the decay. To speak is to be branded as the problem. The institution has an immune system exquisitely tuned not to external threats but to internal ones. The greatest danger to a bureaucracy is not a competitor. The greatest danger is a colleague who points out that the emperor’s new architecture has no clothes.

The heretic is the engineer who builds a working prototype on a weekend and is told the prototype cannot be adopted because it did not go through the eighteen-month review cycle. The heretic is the technical leader who demonstrates that a team of nine can deliver what a program of nine hundred has failed to deliver in three years and is told the result does not count because it was not done according to anointed methodology. The heretic is the person who reads the company’s innovation manifesto and takes the manifesto literally.

Taking the manifesto literally is, in most large organizations, a career-limiting move.

Kelly Johnson, the founder of Lockheed’s Skunk Works, delivered the P-80 Shooting Star in 143 days on a 180-day deadline. Decades later Johnson wrote the epitaph for the kind of organization he had built, expressing the fear that the way he designed and built airplanes may one day no longer be possible, perhaps not even for the Skunk Works. Ben Rich, Johnson’s successor, quantified the decay, noting that 240 people from industry and government developed the F-117 stealth fighter while a decade later more than two thousand auditors, engineers, and official kibitzers oversaw the B-2 bomber, compiling one million sheets of paper every day in reports and data nobody in the bureaucracy had either the time or the interest to read.

The Vocabulary of Stasis

Sankar is sharp on the language institutions use to disguise stasis as strategy. In the defense context, Sankar identifies the way “requirements creep,” “technology readiness levels,” and “milestone reviews” have ceased to manage complexity and serve instead as mechanisms for avoiding decisions. Each gate is a place where a program can be delayed without anyone bearing accountability for the delay. Each review is an opportunity to add requirements without anyone bearing accountability for the cost. The system produces not weapons but documents about weapons.

The corporate parallel is precise. Substitute “digital transformation” for “modernization initiative.” Substitute “strategic review” for “milestone gate.” Substitute “cross-functional alignment” for “requirements coordination.” The vocabulary changes; the function is identical, to create the appearance of motion while preserving the existing order. Palantir’s CEO Alex Karp, captures the dynamic in The Technological Republic with the observation that “those who say nothing wrong often say nothing much at all.” The administrative voice, clinical and careful and above all without feeling, is the voice of institutional self-preservation masquerading as measured judgment.

A particular kind of meeting will be familiar to anyone who has worked in a large technology company, though today the meeting most often takes place on Teams. Thirty people occupy a room. A slide deck runs sixty pages. Three hours of discussion concern the discussion needing to happen before the decision can be made about whether to schedule the review determining if the initiative should proceed to the next phase of evaluation. At no point does anyone ask the only question of consequence, can we build it and should we?

The meetings described above are not failures of execution but successes of institutional self-preservation. The meetings exist because the cost of a wrong decision falls on the person who made the decision, while the cost of no decision is distributed across the organization so thinly that no one feels the cost. Until the market does.

Foucault defined power not as direct coercion but as an action upon an action, the capacity to structure the field within which others act. The distinction illuminates the mechanism at work in Sankar’s analysis. The institution does not need to forbid building; the institution only needs to create an environment in which building requires seventeen approvals, four compliance reviews, and a three-year architecture study. The constraint operates through the shaping of possibility itself. Sankar documents the apparatus in concrete detail, the nearly four thousand auditors of the Defense Contract Audit Agency, the Cost Accounting Standards requiring dual bookkeeping across entire supply chains, the classification regimes used to gatekeep programs under the banner of national security. The defense primes have internalized the discipline so thoroughly that the primes impose government-style requirements on subcontractors even when no statute demands compliance, policing themselves out of sheer terror of being audited. Foucault would recognize the pattern immediately, power made invisible by its diffusion into process, enforced not by any sovereign authority but by the subjects themselves.

Alex Karp, Sankar’s co-founder at Palantir and a product of the Frankfurt School’s critical tradition, arrives at a complementary conclusion in The Technological Republic. Karp cites Steven Levy’s observation that bureaucracies are “designed to consolidate power” and “perceive the constructive impulse as a threat.” Foucault explains the mechanism; Karp names the target. The institution does not merely prevent building through accumulated process. The institution treats building itself as an act of insurrection.

The critical tradition has always excelled at diagnosis. From Adorno to Habermas, the Frankfurt School produced generations of scholars across every field and discipline in the modern university capable of dismantling the logic of institutional power with extraordinary precision. The tradition’s limitation was equally consistent, the diagnosis never culminated in construction. Critical theory could map every mechanism of institutional control and never once propose building something as the response. Karp arrived at the conclusion through the critical tradition; Sankar arrived at the same conclusion through two decades of building against institutional resistance. Both discovered that the real change required was not better critique, not better reform proposals, not better analysis of what was broken, but building. The act of construction itself became the argument that no amount of institutional analysis could refute.

The Insurgent’s Toolkit

Mobilize offers more than a diagnosis. Sankar proposes a way forward. The argument, stripped to its essentials, is that the defense industrial base needs to be rebuilt around speed, iteration, and the willingness to let small teams with real authority solve real problems even when doing so means bypassing the systems designed to prevent those solutions.

The proposal is heresy and Sankar knows it. The book’s power comes from the fact that Sankar is not proposing reform from the outside but rather revolution from within, timely given that Palantir’s Maven is becoming a Program of Record at the Pentagon. Palantir’s entire history is a case study in what happens when a company builds what the institution needs, delivers the product to the people who use it, demonstrates that the product works, and then spends a decade or more fighting the institution’s immune response.

The lesson generalizes. In any ossified institution, the path to building something real follows a predictable and punishing sequence.

First, build it anyway. Do not wait for permission. Do not wait for the architecture review. Do not wait for the strategic alignment workshop. Build a working thing that solves a real problem for a real person. Building is the single most subversive act available inside a large organization because building removes the most powerful weapon the bureaucracy possesses, the ability to argue that something is theoretically impossible while ensuring the thing is never practically attempted.

Sankar’s most vivid illustration is William McLean at the Naval Ordnance Test Station at China Lake. McLean’s division was explicitly told not to develop an air-to-air missile. McLean built the missile anyway, scrounging funding from an existing proximity fuze program and disguising the effort under the name “Local Fuze Project 602.” The resulting Sidewinder was developed at one-tenth the cost of its competitors and achieved an eighty-seven percent hit rate in combat. McLean hated formal requirements in advance of a product because what worked could seldom be known in advance. Requirements, McLean believed, were appropriate only when a system had already been proven. The insight inverts the standard institutional logic entirely, building precedes requirements rather than the reverse. The heretic never waits for the reference architecture. Sometimes the reference architecture is published after the working system has already shipped.

Second, find the warfighters. In Sankar’s framework, the warfighter is the person who actually needs the capability, the soldier, the analyst, the operator. In the corporate context, the warfighter is the customer, the field engineer, the person whose name is on the support ticket at 2 AM. Warfighters do not care about methodology. Warfighters do not care about architectural review boards. Warfighters care about whether the thing works. Find the warfighters. Put the thing in their hands. Let the warfighters report what the thing actually needs to do. Warfighter feedback is worth more than every governance committee in the organization combined.

Andrew Higgins grasped the principle before anyone gave it a name. Higgins designed landing craft not from an office but from the swamps of Louisiana, building boats for trappers and oil drillers who needed shallow-draft vessels capable of beaching and retracting in hostile conditions. When the Navy needed a landing craft for amphibious assault, Higgins already had the design because Higgins had already found the warfighters. By September 1943, 12,964 of the Navy’s 14,072 vessels, ninety-two percent of the fleet, had been designed by Higgins Industries. Eisenhower credited Higgins with winning the war. Every step of the way, Higgins had to fight the Navy bureaucracy for the right to compete.

Third, survive the antibodies. The institution will respond. The institution will respond not with a competing product, since competing would require building something, but with process. A review will follow. Concerns about “scalability” and “supportability” and “alignment with the strategic roadmap” will surface. People who have not built anything in years but who retain veto authority over builders will express their concerns. The antibody phase is where most heretics break. The breaking comes from the slow, grinding, demoralizing weight of an organization preferring conventional failure to unconventional success.

Colonel Drew Cukor’s experience running Project Maven is Sankar’s most detailed case study of the antibody response. Within ninety days of launch, Cukor had operational AI deployed in a combat zone in East Africa. The Pentagon’s response was retaliation. DARPA believed it owned AI and accused Cukor of insubordination. The Army, Air Force, and National Geospatial-Intelligence Agency objected that Cukor was encroaching on their respective territories. A stream of anonymous complaints accused Cukor of money laundering, illegal contracting, and harboring foreigners. The real offense, as Cukor himself understood, was that one group of people was doing exceptional work and the natural institutional reaction was to kill the effort and get everyone back in line.

Fourth, make it undeniable. The only defense against institutional antibodies is evidence so overwhelming that suppressing the evidence becomes more costly than adopting the solution. The defense requires metrics, users, and a production system already running, already delivering value, already creating dependencies making removal politically expensive. The heretic must make the thing so embedded, so obviously superior, so visibly connected to outcomes of consequence that the organization’s only choices are to embrace the thing or to publicly explain why the organization chose to abandon something already working.

The insurgent approach is an exhausting way to innovate and should not be necessary. In organizations where the approval process has become the product, insurgency is the only way forward.

The Institution’s Self-Narrative

The most striking parallel between Sankar’s defense narrative and the corporate experience is the mythology of self-awareness. The Pentagon knows it has an acquisition problem. The Pentagon has commissioned hundreds of studies confirming the problem. The Pentagon has created dozens of reform initiatives and has established innovation offices, rapid fielding directorates, software factories, and pathfinder programs. Each initiative is announced with genuine enthusiasm. Each initiative is slowly, patiently, thoroughly digested by the existing bureaucracy until the initiative becomes another node in the network the initiative was designed to disrupt.

Large technology companies follow the same pattern. The companies create innovation labs, establish “startup-within-a-startup” programs, hire chief transformation officers, and give TED talks about agility. The innovation lab’s output is then subjected to the same thirty-seven approval gates as everything else, the startup-within-a-startup discovers that external tools require a nine-month security review, and the chief transformation officer publishes a LinkedIn post eighteen months later about “incredible journey” and “amazing team.”

The institution is not unaware of its dysfunction. The institution is aware and has learned to metabolize awareness into yet another process. Self-diagnosis without treatment is not insight; self-diagnosis without treatment is theater.

The Question of Loyalty

An accusation follows every institutional heretic, the accusation of disloyalty. The person who challenges the process is told the challenger does not understand the complexity. The person who builds outside the approved architecture is told the builder is creating risk. The person who demonstrates that a small team can outperform a large program is told the demonstration undermines morale.

Sankar addresses the accusation directly in the defense context. Sankar argues that true loyalty to the mission requires disloyalty to the dysfunction, that a soldier who sees a broken supply chain and says nothing is not being loyal but being compliant, a different thing entirely. The deepest form of institutional commitment is the willingness to fight the institution when the institution is failing its own purpose. Admiral Hyman Rickover, the father of the nuclear Navy whom Karp profiles at length in The Technological Republic, captured the distinction in a single sentence, “My job was not to work within the system. My job was to get things done.” Loyalty to the system and loyalty to the mission diverge precisely at the point when the system can no longer accomplish the mission. The heretic chooses the mission.

The distinction between loyalty and compliance is one many large organizations refuse to make. The organizations conflate loyalty to the mission with loyalty to the org chart, treat disagreement as defection, and reward the person who manages the failure smoothly over the person who prevents the failure noisily.

Heretics are sidelined. The sidelining is gradual, one reassignment at a time, quietly, over years. The heretics find their projects defunded, their teams redistributed, their influence reduced to advisory roles with no authority. The institution strips away everything except the title while publishing a press release about its exciting new innovation initiative.

The Cost of Comfort

Sankar’s book carries an urgency the corporate parallel often lacks. The argument is that the broken defense industrial base is dangerous and not merely inefficient, that the inability to build and field weapons at speed is an existential national security risk, that China is not waiting for the DoD to complete its acquisition reform study.

The corporate version of the threat is less dramatic but no less real. Markets do not wait. Technology does not wait. The customer who needed the solution three years ago has already found the solution elsewhere. The competitor dismissed as “not enterprise-ready” has shipped six versions while the incumbent was finalizing its requirements document.

The comfort of a large institution, the steady paycheck, the known processes, the predictable career trajectory, comes at a cost invisible until the cost becomes catastrophic. The cost is measured not in what the organization does but in what the organization fails to do, in the products never built, in the markets never entered, in the talent never empowered, in the decade passing while the strategy is being finalized. The institution tracks every internal cost with obsessive precision, logging hours and auditing expenditures down to the decimal, while losing sight of the only number justifying the exercise in the first place, revenue. The institution can account for every dollar spent but cannot account for the dollars never earned.

Sankar quotes the old military maxim, the enemy gets a vote. So does the market. The market’s vote is not subject to review by the governance committee.

Building Anyway

The most radical proposition in Mobilize is not any specific policy recommendation. The most radical proposition is the underlying assertion that building things is what defines success, that the person who ships a working system in the face of institutional resistance is a patriot, and the willingness to break glass, to violate process in service of outcome, is an act of responsibility. Karp states the same principle with characteristic directness in his own book, “If a U.S. marine asks for a better rifle, we should build it. And the same goes for software.” The engineering elite, Karp argues, bears an affirmative obligation to participate in the defense of the nation, an obligation not contingent on institutional approval or procurement authorization but existing simply because the need exists.

The assertion is deeply uncomfortable for institutions that have spent decades building elaborate systems of control. The assertion implies that the systems themselves may be the problem, that the architecture of oversight designed to prevent failure has become the primary mechanism by which failure is guaranteed, that the safest-looking path, follow the process, attend the reviews, only sell what’s on the truck, and wait for approval, is in fact the most dangerous path because the safest-looking path ensures nothing is delivered while everything is documented.

Sankar extends the argument to its material conclusion with the assertion that the factory itself is a component of the fighting force, as integral to a weapon system as a radar or a fuselage. Mass production is the deterrent. A nation incapable of producing weapons at scale cannot credibly threaten to use force, eventually facing an adversary willing to call the bluff. The urgency is not abstract. China has already rebuilt its industrial base around the logic Sankar describes, manufacturing at a speed and scale the American defense establishment cannot currently match.

Rapid coding with generative AI as a development methodology has begun to alter the calculus in ways Sankar’s heretics could not have anticipated even five years ago. The distance from concept to working prototype, once measured in months or years of coordinated institutional effort, now collapses to days or hours for a single engineer with the right tools. A small team augmented by AI-driven code generation, architecture exploration, and iterative refinement can produce functional systems at a pace that would have required a department a decade ago. The shift undermines the institution’s most durable assumption, that building requires the institution’s resources, the institution’s timelines, and the institution’s approval. The heretic who once needed to scrounge funding and disguise efforts under false project names, as McLean did with Sidewinder, can now produce a working system before the architecture review committee has scheduled its first meeting. The institutional monopoly on the capacity to build is breaking and the implications for every ossified bureaucracy are profound.

The heretic understands the dynamic. The heretic has always understood the dynamic. The heretic is the person who looks at the gap between what the institution says it values and what the institution actually rewards. True values come from practice, not advocacy.

Every large organization says it wants innovation. The test is simple, what happens to the person who actually delivers innovation without asking permission first?

The answer says everything about whether the institution is still running or just idling.

The Way Out

Sankar’s prescription for the defense industrial base is, ultimately, a prescription for any institution that has lost the ability to build. The prescription is to empower the people closest to the problem to engage revolutionary approaches, reduce the distance between decision and action, tolerate imperfection in service of speed, and above all judge people by what they deliver rather than how well they navigate the process.

The obstacle is not ignorance; the obstacle is incentive. The people who benefit from the current system are the people with the authority to change the system and the beneficiaries have no reason to do so. The acquisition officer whose career depends on managing a twenty-year program has no incentive to demonstrate that the program could be completed in three years. The division president whose bonus is tied to headcount has no incentive to show that the work could be done by a tenth of the people. The architecture review board whose authority depends on serving as a gate has no incentive to leave it wide open.

Heretics matter for precisely the reason the incentive structure predicts. Heretics are not necessarily smarter or more talented than the colleagues around them, though heretics often are. Heretics are the only people in the system whose incentives are aligned with the mission rather than the machinery, the only people who would rather be sidelined for building something that works than promoted for managing something that does not.

Sankar’s book is a call to arms for the defense industrial base. The book’s truest audience may be the person sitting in a fluorescent-lit office in a building named after a founding patriarch, staring at a slide deck about the future of innovation, knowing that the prototype in the desk drawer already works, wondering whether the fight to prove the prototype’s value is justified.

The fight is justified.

The fight has always been justified.

The institutions that survive will be the ones learning to value their heretics before the heretics stop trying.

Shyam Sankar’s Mobilize: How to Reboot the American Industrial Base and Stop World War III and Alex Karp’s The Technological Republic are both available now and should be read together by defense professionals and anyone who has tried to build something real inside a machine forgetting how.