PB27 · Department of War · Inside Analysis

FY 2027 Department of War Budget — what the request actually says

A portfolio-manager's reference to the FY27 President's Budget request. The documents use the rebranded "Department of War" nomenclature throughout (OUSW(C) rather than OUSD(C)), organized around the 2026 NDS "Peace Through Strength" framework and its four lines of effort. This page pulls out what matters for budget execution, force structure, and program-level planning.

As of FY27 President's Budget submissionOUSW(C) source materials

Executive overview

Total request
$1,450B
Disc $1100B + Mand $350B
Scrubbed base growth
+28%
$251B above FY26 enacted (CRFB)
DAWG year-over-year
243×
$0.225B → $53.6B
Sovereign AI infra
$46B
79% of the $58.5B AI envelope
The $1.45T headline obscures two different stories
The publicly reported 44% growth is inflated by adding the $350B in reconciliation-funded mandatory spending to the discretionary baseline. Scrubbed, the base defense discretionary increase is $251B (28%) over FY26 enacted — still historically large, but not unprecedented. The remaining $350B flows through a reconciliation vehicle that has not yet cleared Congress. A portfolio plan that depends on the mandatory tranche carries policy risk until that bill is signed.

Five bottom-line takeaways

  1. 01 Defense-Wide appropriations absorb 70% of the mandatory tranche ($245B). OUSD(A&S) and CDAO are not sized to execute at this scale — expect heavy MIPR pass-through to the Services and significant Q4 carryover.
  2. 02 Investment accounts (Procurement + RDT&E) exceed 50% of the request for the first time since the Reagan buildup.
  3. 03 The AI portfolio is infrastructure-heavy, not application-heavy. 79% is sovereign compute; 21% is applications, integration, and R&D. This inverts the commercial AI ratio.
  4. 04 The MAC munitions package is the clearest signal of real depletion from regional employment — PAC-3 MSE and THAAD alone get 2,009 missiles in one year.
  5. 05 Attritable mass is an explicit doctrine now. Drone Dominance at $53.6B, with a ~1:1 ratio of asset purchases to counter-UXS spending, suggests the Department expects defense to be economically dominant in the drone era.

Peace Through Strength — four lines of effort

The 2026 National Defense Strategy reorganizes the Department around four mutually reinforcing lines of effort. Budget dollars map to these LoEs unevenly — Deter China and DIB are disproportionately funded, reflecting the strategic shift away from a globally distributed posture.

Line of effort
Homeland Defense
~$185B
Golden Dome core, border, space domain awareness
Line of effort
Deter China
~$445B
INDOPACOM posture, munitions, AUKUS, Space Force
Line of effort
Burden-Sharing
~$120B
NATO 5% commitments, EDI reshape
Line of effort
Industrial Base
~$348B
Munitions, shipbuilding, microelectronics, critical minerals
FY27 resourcing by line of effort (illustrative $B)

Note: LoE rollups are analyst-assembled from programmatic content, not a first-class appropriation view. Programs frequently support multiple LoEs.

Topline architecture — $1.45T built from two sources

The request is architecturally bimodal. The discretionary $1.1T follows the traditional PPBE process — PB justification books, congressional markup, appropriations conference. The mandatory $350B depends on a reconciliation vehicle that bypasses the 60-vote Senate threshold but requires its own legislative path.

The split is not just procedural. Mandatory funding is baseline-counted — once authorized, it flows year over year without annual appropriation battles. For high-priority programs like Golden Dome, Drone Dominance, and the AI Arsenal, this is a strategic choice to protect the funding stream from future administrations.

CRFB's analysis is worth internalizing: subtract the $350B reconciliation add, and the scrubbed base defense discretionary growth is $251B or 28% — historically large but not unprecedented.

Decomposition
Total request$1,450B
Discretionary (traditional PPBE)$1,100B
Mandatory (via reconciliation)$350B
FY26 enacted base$900B
Scrubbed base growth+$251B (+28%)
Mandatory add on top+$350B
Nominal headline growth+44%

Discretionary by appropriation title — where $1.1T sits

The discretionary $1.1T divides into the familiar appropriation titles. The notable shift for FY27 is the growth of investment accounts (Procurement + RDT&E) relative to operating accounts (MILPERS + O&M). Investment accounts now exceed 50% of the total for the first time in decades.

Discretionary $1.1T by appropriation title
Total discretionary
$1100B
O&M — $322B
$190B core readiness + enablers, plus $22B DHP/PSCP, $20B COMP (new appropriation), and base operations. The COMP carve-out is a structural change from prior-year DHP flow.
MILPERS — $199B
Growth driven by the 7-6-5 pay raise (7% E-1 through E-4, 6% E-5 through O-6, 5% above), plus retention investments and expanded recruiting bonuses.
RDT&E — $179B discretionary
Add another $100B in mandatory RDT&E,DW and the total RDT&E envelope is approaching $280B. The 6.1-6.3 science-and-technology line, however, is cut 8% — the imbalance is in applied development.

Mandatory $350B — where the novel signal is

The mandatory tranche is where the strategic intent shows most clearly. Two sections — Defense Industrial Base and Next-Gen Tech & Autonomy — account for 61% ($216B) of the mandatory ask. Munitions takes another 13%. The remaining five sections split the final 26%.

Mandatory $350B by section
70% flows through Defense-Wide appropriations
Of the $350B mandatory, approximately $245B runs through Defense-Wide accounts: Procurement,DW ($39.4B), RDT&E,DW ($100.5B), DPA Purchases ($30.0B), Strategic Capital Credit ($20.0B), and a new Golden Dome for America Fund ($17.1B). That concentration in OSD-controlled accounts is without recent historical precedent. Traditionally the Services would own this much new money.

Procurement deep-dive — $413B in weapons buys

The combined procurement ask (discretionary + mandatory) is the most important chart if you track steel-and-silicon flows. Four anomalies stand out from the traditional pattern.

Procurement by appropriation title, disc vs mand ($B)
Missile Procurement, Army — $36.6B
$12.1B disc + $24.5B mand = 3.4× multiplier
Driven almost entirely by the MAC munitions acceleration. PAC-3 MSE quantities alone reach 1,636 missiles across Army and Navy — a ~3× step over the historical cadence of 500-650/year.
Aircraft Procurement, Army collapse
$4.1B FY26 → $1.9B FY27
Zero Apache Block IIIA Reman funded, UH-60M drops from 29 → 1 aircraft, CH-47F from 11 → 5. This is the Army rotary-wing drawdown made explicit in budget dollars.
Shipbuilding, Navy — $65.8B
Single largest procurement line in the budget
But DDG-51 drops to 1 hull/year from 2. Constellation-class frigate is cancelled outright — replaced by a new FF(X) program scheduled for PB28.
Space Force procurement doubled
$10.1B → $19.1B
$7.7B of that is the new Space-based Air Moving Target Indicator (SB-AMTI) global coverage program, which did not exist as a FY26 line item. Proliferated LEO continues to grow.
DPA Purchases at $30.4B — 14× historical norm
$2-3B typical → $30.4B FY27
This is effectively a new category of 'procurement' for the Department — buying industrial capacity rather than end items. Largely reflects critical minerals, microelectronics, and munitions substrate.
Office of Strategic Capital — $20.2B credit subsidy
Backing ~$200B lending authority
OSC uses the credit-subsidy model to issue federal loan guarantees to industrial base suppliers. The $20B is the subsidy cost; the downstream lending is roughly 10× that.

MAC munitions — 18 programs, $47B concentrated signal

The Munitions Acceleration Council package is the most concentrated procurement signal in the budget. It combines new-program funding with "FY26 Procurement Disconnects" — incremental money to accelerate prior-year contracts. The latter line alone is $19.6B, or 42% of the total MAC ask.

MAC munitions ranked by mandatory dollars
PAC-3 MSE — 1,636 missiles in one year
Army gets 1,429, Navy gets 207. Historical PAC-3 cadence is 500-650/year. This is a ~2.5-3× production step and will require Lockheed's Camden facility expansion to be online on schedule.
THAAD — 830 interceptors total
373 new plus 457 via the FY26 Disconnects acceleration. The replenishment profile reflects real depletion from regional employment — not a strategic reserve buildup.
Low-Cost Cruise Missile — the industrial-policy signal
$1.6B for 1,000 units at ~$300K unit cost, plus integration funding for 2,000 more. This is CCA-adjacent thinking applied to cruise missiles: attritable mass over exquisite precision.
Blackbeard-GL hypersonic — $326M R&D
Small line, but the most strategically important development signal. Explicit attempt to drive hypersonic unit cost down by an order of magnitude from CPS/LRHW economics.
Conspicuous absence: SLCM-N zeroed
Sea-Launched Cruise Missile, Nuclear goes from $1.69B FY26 to $0 FY27. The stated justification is deterrence portfolio rationalization. The practical effect is to end a program that had substantial congressional support — expect markup activity.

AI, autonomy, and the Advana → War Data Platform transformation

This is where the budget is doing the most structurally novel work, and it's the section most likely to outlive any particular administration. The Department is simultaneously (a) building sovereign compute infrastructure, (b) restructuring its enterprise data platform, (c) standing up a dedicated autonomy organization with three-orders-of-magnitude budget growth, and (d) elevating a single CTO over the AI/innovation stack.

Frontier-tech envelope
$160B
AI $58.5B + Autonomy $53.6B + S&T $48B
Infra share of AI envelope
79%
$46B of $58.5B. Inverts commercial AI ratio.
DAWG year-over-year growth
243×
$0.225B FY26 → $53.6B FY27
AI Arsenal vs Drone Dominance breakdown
AI Arsenal
Drone Dominance

Advana restructuring into the War Data Platform

Per the January 2026 Hegseth/Feinberg memos (which this budget is built to fund), Advana is being trifurcated. The technical problem is real: audit-quality data lineage and warfighting-latency data serving have incompatible requirements.

Advana (legacy)Since 2019 · 400+ systemsWar Data PlatformWarfighting / intel dataStandardized AI data accessAdvana for Fin MgmtAudit remediationFY27 DWCF → FY28 agencyWDP App ServicesApp rationalizationSelf-service AI toolingAI Arsenal · $46.0BSovereign GPU infrastructureClean audit · FY28 target$1.7B audit investmentsGenAI.mil · MSS · JFN$2.3B CJADC2 surfaceDefense Autonomous Warfare Group (DAWG)$225M FY26 → $53.6B FY27 — 243× growthDrone Dominance target: 200,000+ autonomous systems by 2027

What an ML/AI practitioner should notice

  1. Infrastructure-to-application ratio is inverted vs commercial. 79% of the AI envelope is infrastructure. A commercial AI build-out at comparable scale would typically be 40-50% infrastructure, 50-60% applications. The Department is betting the binding constraint is government-owned, hardened, air-gapped compute — not model quality.
  2. GenAI.mil is frontier-vendor-agnostic by design. Budget language says the Department will have "access to the latest models from the top American frontier AI labs." That's a model-plurality strategy — Anthropic, OpenAI, Google, xAI, Meta — with a thin integration layer rather than a single-vendor bet. The cancelled $15B AAMAC was the last major single-vendor-scale play.
  3. The WDP/Advana split is architecturally correct. Separating warfighting data serving (sub-second latency, agentic AI) from audit trails (reproducibility, lineage) is the right move. Expect WDP to look like a modern lakehouse + vector store + feature store + ML serving stack.
  4. The $46B sovereign infra has a hidden supply-chain dependency. $46B at 2026 prices implies 100Ks of H100/B200-class accelerators in government-owned facilities. The DIB section calls out $48.8B+ in critical-minerals/IBAS investment for the substrate and packaging. These two lines are coupled — watch CHIPS Act follow-on legislation.
  5. Drone Dominance is a bet on attritable mass. The ratio of Asset Purchases ($16.9B) to Counter-UXS ($14.4B) is ~1:1 — the Department is spending roughly as much defending against drones as buying them. That's diagnostic of a world where defense dominates offense economically. Target of 200K+ autonomous systems by 2027 at this budget implies average unit cost ~$85K — Group 2-3 small UAS, not exquisite MQ-9-class platforms.
  6. Agentic AI is the real frontier bet. The $500M "AI Pace Setting Projects" line is a bellwether. The seven PSPs are agentic workflows: AI-enabled cyber, AI-enabled intelligence reporting, AI-enabled manufacturing, AI-enabled maintenance. Tool use and automation, not generative content. This is where commercial AI is still immature.

Program winners and losers

Winners

Golden Dome for America
$17.9B
New homeland missile defense architecture
Drone Dominance / DAWG
$53.6B
Up from $225M FY26 — 243x
Sovereign AI Arsenal
$46B
Gov-owned accelerator infrastructure
Critical Minerals / IBAS
$48.8B
Industrial base investment
MAC Munitions
$47B
18 programs, PAC-3 + THAAD-led
Space Force Procurement
$19.1B
Nearly doubled. SB-AMTI at $7.7B
DPA Purchases
$30.4B
~14x historical norm
Office of Strategic Capital
$20.2B
$200B credit authority behind it

Cuts, terminations, drawdowns

SLCM-N (Sea-Launched Cruise Missile, Nuclear)
−$1.69B → $0
Zeroed
B-21 Raider procurement
−$3.96B
$10.07B → $6.11B
Sentinel ICBM
−$379M
Continued schedule pressure
Constellation-class frigate
Cancelled
Replaced with new FF(X) program
Army UH-60M Black Hawk
29 → 1 aircraft
Rotary-wing drawdown
Army CH-47F Chinook
11 → 5 aircraft
Rotary-wing drawdown
AH-64E Apache Reman
→ $0
Zeroed in FY27
Science & Technology (6.1-6.3)
−8%
Below-threshold per 3346 covenant
DDG-51 Arleigh Burke
2 → 1 hull/yr
Production step-down
Contract Services
−$13.3B
FTE implications significant

Execution risks and congressional flags

The forward risk hierarchy is clear. Three items are at the high end — two legislative, one programmatic — and any of them can materially change the FY27 execution picture.

Reconciliation vehiclehigh
$350B (24% of request) depends on a reconciliation bill that has not yet moved. If blocked, the department loses mandatory flexibility and falls back to discretionary caps.
BLI / PE consolidationhigh
Proposed consolidation of Program Elements and Budget Line Items reduces congressional line-item visibility. Markup may reject and restore granularity.
Defense-Wide span of controlmedium
70% of mandatory ($245B) flows through OSD-controlled appropriations. OUSD(A&S) and CDAO are not sized to execute at this scale — expect significant MIPR pass-through.
DAWG scale-up executionmedium
243x year-over-year growth. Contracts, oversight, and workforce cannot ramp at that rate — expect Q4 carryover and OUO reprogramming.
Semiconductor dependencymedium
Sovereign AI Arsenal at $46B implies 100Ks of accelerators. Dependent on CHIPS Act follow-on legislation for substrate/packaging.
Reform savings realismlow
Claimed ~$50B in Pentagon savings embedded in topline. Historical realization rate of announced savings is typically 30-50%.

Portfolio management implications

Near-term (FY27 execution)

  • • Expect heavy Q2-Q3 obligation activity in Defense-Wide mandatory accounts. Stand up MIPR infrastructure early.
  • • Program plans that depend on reconciliation money must have a discretionary-only fallback path. Build both.
  • • DAWG will not execute $53.6B in one fiscal year. Plan for 40-60% carryover and reprogramming actions into FY28.
  • • Any program cut in PB27 should be treated as potentially restorable in conference. Monitor HASC and SASC markup.

Mid-term (PB28 formulation)

  • • BLI consolidation outcomes from this cycle determine the PB28 line-item structure. Track markup closely.
  • • Advana for Financial Management moves from DWCF to agency appropriations in FY28 — plan for cost share reallocation.
  • • Golden Dome architecture decisions made in FY27 will lock FY28-30 procurement profiles for SM-3 IIA, NGI, and sensor networks.
  • • Sovereign AI infrastructure spending will create operational funding requirements (power, cooling, sustainment) that don't appear until FY29+.

Long-term (2030+ force structure)

  • • The DDG-51 to 1 hull/year decision, if sustained, changes the 2035 surface combatant force structure materially.
  • • Army rotary-wing drawdown (UH-60, CH-47, AH-64 reman) is a structural bet on FLRAA and uncrewed teaming — the 2030 aviation mix looks different.
  • • The DAWG scale-up creates an organizational dependency — if it fails to execute, the small-UAS mass strategy fails with it.
  • • Warrior Ethos spending patterns (recruiting, retention, training) will determine whether the end-strength plan holds through 2028.

Sources and further reading

OUSW(C) FY2027 Budget Materials
Primary source — Overview Book, Mandatory Funding Overview, P-1/R-1
Committee for a Responsible Federal Budget
Scrubbed base defense discretionary growth analysis
Congressional Budget Office
Baseline projections and reconciliation scoring
Government Accountability Office
GAO-26 series on acquisition, audit readiness, munitions industrial base
Congressional Research Service
R-series and IF-series reports on specific programs
OMB Budget Materials
Cross-agency view, OMB Circular A-11 passback guidance

This analysis represents the author's reading of public PB27 materials and is not an official Department of War or agency product. Dollar figures round to one decimal place except where source material warrants greater precision. Where source material uses "Department of War" terminology, this page preserves it; where "DoD" is standard usage, both terms may appear.

Back to dashboard