Hugh Zabriskie spent years watching promising machines fail in the least glamorous corner of the defense business: the test event. A system would perform cleanly in a demo. Then the evaluation team would spend weeks reconstructing what actually happened, working from video, operator memory, and handwritten notes.

That reconstruction gap is the company he now runs. Valkryn, founded in 2025, came out of stealth on 17 November with $17M in total funding, including a $15M seed led by Andreessen Horowitz's American Dynamism fund. It does not build drones or robots. It instruments the tests that decide whether they get fielded.

The pitch is narrow. Test capacity is scarce, and scarcity has a price. "Companies wait months for an opportunity to demonstrate themselves," Zabriskie told Axios. The delay is not a scheduling nuisance. It is the speed limit on the entire procurement cycle.

Our read: the scarce asset in autonomy is evidence. Repeatable, comparable, auditable evidence sits closer to the budget line than another airframe, because every hardware award eventually needs a verification signature. Valkryn is selling that signature as software.

The test range is a data problem

The product is called Battle Lab. It has three layers. An Environment layer reproduces the conditions that matter: radio-frequency denial, dense airspace, threat-representative adversaries, weather. A Measurement layer captures telemetry and operator actions on a common clock. A Learning layer scores each event against government criteria and lets evaluators compare programs, sites, and software versions directly.

The alternative is the status quo described in the company's own materials. Evaluation events still lean on handwritten notes, and procurement decisions worth billions can rest on slide decks rather than structured data. That is a claim about process, not one we can audit from outside. It is also consistent with what test practitioners describe publicly.

The commercial insight is that the bottleneck moved. Hardware got cheaper and faster to iterate. Verification did not. A drone that ships a new software build every month needs a test regime that runs at the same cadence.


CHRONICLE: Valkryn
────────────────────────────
2025  ·  🏗  Founded by autonomy and defense engineers
MAY 2026  ·  📄  Army award W912CH26CA025, $250k ceiling
13 NOV  ·  🎯  Air Force Test Center sole-source notice
17 NOV  ·  🚀  Stealth exit, $17M total, seed led by a16z
23 NOV  ·  ⏳  Anticipated award date on FA2823-27-Q-A001

Procurement filings and company announcement, November 2026

The distinction that matters is between a demo and an evaluation. A demo is designed to succeed. An evaluation is designed to find the failure mode before a unit does. Instrumenting the second is a harder software problem, because data arrives from many vendors' sensors and must be synchronized, normalized, and scored against criteria the government can defend in an audit.

Twelve months, five dates, one system of record

The most useful document is not the funding announcement. It is a notice of intent published on 13 November by the Air Force Test Center's operational contracting division at Eglin Air Force Base. The center intends to award a sole-source, firm-fixed-price order to Valkryn Systems for TestOS enclave hosting, live sensor integration, and software licensing.

The justification is the interesting part. Valkryn owns the proprietary TestOS source code, and the platform is the baseline for JIATF-401 and ATEC Data Mesh counter-small-UAS testing. No other commercial source, the notice states, has the required pre-built ingestion architecture. Estimated value runs from $250,000 to $1,000,000. The anticipated award date is 23 November.

An earlier Army award, W912CH26CA025, placed the same TestOS evaluation system at a forward test site with a $250,000 ceiling. Neither contract is large. Together they establish something more valuable than revenue: a default position.

"Neutral means we don't build the systems that are under test."— Hugh Zabriskie, co-founder and CEO, Valkryn
⚠️
The conflict-of-interest math
An evaluator that also sold hardware would have an incentive to grade generously. Valkryn's structural choice removes that incentive and caps its addressable market in the same move.

Traction so far is contract-shaped, not revenue-shaped. Zabriskie described several awards, mostly Other Transaction Agreements, as funded and operational. "These aren't SBIRs," he told Tectonic. "This is quite real work." The distinction matters to investors because an OTA can move from prototype to production without a fresh competition, which shortens the path from a test contract to a program of record.

Why independence is the product

Operational testing of major defense systems is required by law to be independent of the program office. The program office and its vendor both benefit from a favorable result. That is why the independent check exists, and why a commercial vendor can sell it.

The market cap follows from the same logic. An independent evaluator cannot build the systems it tests, so its revenue tracks testing budgets, not procurement budgets. Investors usually prefer the bigger number. Here the smaller number is the moat.

Neutrality is also fragile. One disputed evaluation and the premise takes damage. The reputation risk is concentrated, and it compounds with every new program.

The institutional backdrop helps. The GAO warned in June 2026 that halving the staff of the independent operational test organization raised the risk of systems reaching troops with undocumented shortfalls. The Secretary of War has described testing as an operational risk rather than an administrative one. Capability is arriving faster than the capacity to check it. That is the wedge.

The competitive set is no longer empty. Nominal raised a $240M Series C in November for adjacent hardware-testing data infrastructure, and the Army has been opening its ranges to private operators. The gap Valkryn targets is narrower than testing in general. It is the evaluator's workflow, from test design to scored evidence.

What $15M buys before the standard exists

The demand signal sits in the budget request. The Pentagon asked for $53.6B for autonomy and drone platforms and $21B for counter-drone systems in fiscal 2027, up from $13.4B and $3.1B the year before, according to figures cited in the company's release. Defense tech venture funding reached a record $35.4B in the first half of 2026, with autonomous systems at 42% of second-quarter funding.

Line itemFiscal 2027 requestPrior year
Autonomy and drone platforms$53.6B$13.4B
Counter-drone systems$21.0B$3.1B

Figures cited in Valkryn's 17 November release, attributed to the Pentagon budget request

A $15M seed does not buy a standard. It buys engineering capacity, a second and third deployment, and the time to make the scoring methodology legible to contracting officers. Dealroom ranked the round in the 97th percentile of security seed rounds in a sample of 3,284 deals, according to reporting that cites its data. Percentile rankings flatter small samples of large checks.

The honest risk is budget gravity. If verification stays a line item that programs fund last, the ceiling is real. If independent evaluation becomes a standing requirement with its own appropriations, the buyer changes from a program office to the department. The September Project Agincourt memo, which stands up a command to test equipment under realistic opposition, points in that direction without settling the funding question.

The bet is that autonomy proliferation is already priced in, and that verification becomes a budget category. Valkryn has twelve months of contracts, a $15M runway extension, and one sole-source notice pointing at a November award. What it does not have is a published testing standard it can own. Whoever writes that standard inherits the category. Until then, the company is selling capacity into a market that has not decided how much proof it wants.

Air Force Test Center notice of intent: Valkryn TestOS C2Net
Sole-source justification, quantities, estimated value, and the 23 November anticipated award date for the evaluator's platform.
The primary document behind the traction story — a procurement notice, not a press release.
Defense-tech tester Valkryn emerges with $17 million
Founder background, the neutrality argument, and the test-capacity bottleneck in Zabriskie's own words.
Useful for the quotes and for how the company frames its scarce resource.
Hardware evaluation startup Valkryn emerges from stealth with $17M
The three-layer Battle Lab breakdown, the OTA contracts, and the founders' prior roles at Applied Intuition and Anduril.
The most detailed independent description of what the product actually does.
The gap between what works and what sells
Why software-defined systems break the old test model, plus the GAO warning and the market-cap objection.
The clearest statement of the bear case: a neutral evaluator cannot sell into the procurement budget.