One nonprofit, nine front doors
DEFENSEWERX is a 501(c)(3) nonprofit, established in 2012, headquartered in Niceville, Florida, and named for James "Jimmy" Doolittle. It does not build technology and it does not hold a program budget. What it holds is a set of agreements with federal sponsors, and under each agreement it operates an innovation hub: a branded venue where a command or a laboratory publishes a problem and companies respond. SOFWERX is the one most people have heard of. It is one of nine.
That structure is the first thing to understand, because it explains behavior that otherwise looks inconsistent. Each hub takes its rules, its money, and its urgency from its own sponsor. SOFWERX moves the way U.S. Special Operations Command moves. ENERGYWERX moves the way the Department of Energy's Office of Technology Transitions moves, which is a different rhythm entirely and mostly does not involve cash going to the company. Two hubs sharing a parent organization and a naming convention can behave like two unrelated institutions, and treating a good experience at one as predictive of another is a common and expensive mistake.
The second thing to understand is what the parent organization is for. A federal laboratory or center can talk to industry directly. It usually does so slowly, through channels built for a different era, with a contracting shop that has a queue. The nonprofit exists to be the party that can move at a different speed: hold the event, take the submissions, run the logistics, sign a business-to-business agreement, and pay a company without the sponsor first opening a procurement action. The legal basis for that is narrow and specific, and it is worth reading before you build a plan around it.
The current roster
Hubs open and close. A widely circulated 2023 listing of DEFENSEWERX hubs included names that do not appear on the organization's current page, and the Naval Postgraduate School relationship now runs through FLEETWERX in Monterey. Check the roster before you cite it. As of this writing the parent organization lists nine.
| Hub | Federal sponsor | Location |
|---|---|---|
| SOFWERX | U.S. Special Operations Command | Tampa, FL |
| Doolittle Institute | Air Force Research Laboratory, Munitions Directorate (AFRL/RW) | Niceville, FL |
| ERDCWERX | U.S. Army Engineer Research and Development Center | Vicksburg, MS |
| Cyber Fusion Innovation Center | U.S. Army Cyber Command | Augusta, GA |
| FLEETWERX | Naval Postgraduate School | Monterey, CA |
| Nautilus | Office of Naval Research and NavalX | Arlington, VA |
| HSWERX | DHS Science and Technology Directorate | Niceville, FL |
| ICWERX | Intelligence Community, launched with CIA Labs | Tampa, FL |
| ENERGYWERX | DOE Office of Technology Transitions | Washington, DC |
Two of these are worth flagging for anyone building a pipeline. HSWERX was stood up as a pilot hub for DHS S&T rather than as a permanent fixture, which is a different kind of commitment than a decade-old laboratory relationship. And ICWERX is the only one Congress has publicly second-guessed: section 6722 of the FY2023 intelligence authorization, enacted as part of Public Law 117-263, directed the Director of National Intelligence, with CIA and NSA, to report on "whether the intelligence community would benefit from the establishment of a new organization to be known as 'ICWERX'" — a report Congress required after the hub already existed.
What a partnership intermediary agreement actually is
The authority is 15 U.S.C. § 3715, a section of the Stevenson-Wydler Technology Innovation Act. It lets the director of a federal laboratory enter a contract or memorandum of understanding with a partnership intermediary to perform services "that increase the likelihood of success in the conduct of cooperative or joint activities of such Federal laboratory with small business firms," paid out of funds available for the technology transfer function.
Read the definition carefully, because it is the part most summaries skip. A partnership intermediary is "an agency of a State or local government, or a nonprofit entity owned in whole or in part by, chartered by, funded in whole or in part by, or operated in whole or in part by or on behalf of a State or local government." The statute did not contemplate a national network of hubs. It contemplated a state economic development office or a university-adjacent nonprofit helping small firms get value out of a nearby laboratory. Everything the model does now sits on that sentence.
DoD has its own parallel provision at 10 U.S.C. § 4124(f), which lets the director of a Center for Science, Technology, and Engineering Partnership enter "a contract, memorandum of understanding or other transaction with a partnership intermediary," subject to the approval of the Secretary. That provision was widened by section 214 of the FY2024 National Defense Authorization Act (Public Law 118-31), which added two functions to the definition: facilitating technology transfer from industry or academic institutions to a Center, and workforce development in critical technology areas. The original statute pointed outward, from the laboratory to industry. The 2023 amendment made the inbound direction explicit, which is a fair description of what the hubs were already doing.
What none of this authorizes is the intermediary buying things for the government. The nonprofit performs services. It runs events, scouts technology, handles submissions, and executes agreements in its own name. The government's requirement, the government's money, and the government's decision about who wins remain the government's.
How money actually reaches a company
The single most useful habit when reading a hub announcement is to scroll to the section that lists possible award instruments. Hub events publish it, and it tells you more about the opportunity than the technical description does. A SOFWERX assessment event page, for example, lists five possibilities.
A business-to-business agreement as a sub-award under the PIA. The company contracts with the nonprofit, not with the government, under the intermediary's own agreement at 15 U.S.C. § 3715. This is the fastest instrument in the set and the one with the least federal-contract character attached to it.
An other transaction for research or a prototype project. Cited as 10 U.S.C. §§ 4021 and 4022. This is a federal agreement with the government, negotiated term by term, with no FAR clauses unless the agreement adopts them.
Procurement for experimental purposes. 10 U.S.C. § 4023, an old and underused authority limited to quantities needed for prototyping, experimentation, technical evaluation, assessment of operational utility or safety, or a residual operational capability.
A cooperative research and development agreement. A CRADA moves people, facilities, and intellectual property between a laboratory and a company. It does not move federal money to the company.
A prize, or an ordinary FAR contract. Prize authority appears as 10 U.S.C. § 4025 and 15 U.S.C. § 3719. And the last item on the list is always the plain federal contract, which is a reminder that a hub is a front door, not a substitute for the acquisition system behind it.
Each of those five has a different consequence for your intellectual property, your accounting, your audit exposure, and your ability to point at the work later. A PIA sub-award and a § 4022 prototype other transaction are not variations on a theme. They are different instruments with different downstream lives, and the difference between them is the reason to ask early rather than at negotiation.
One item on that list carries real strategic weight. SOFWERX event materials state plainly that "an award under 10 U.S. Code, Section 4022 may result in the further award of a follow-on production agreement without additional competition based on successful prototype completion." That is the § 4022(f) pathway, and it is the mechanism by which a modest prototype becomes a production award without another competition. If a hub event can produce a § 4022 agreement, the event is worth more than its dollar value suggests.
The assessment event, phase by phase
The hubs converged on a common event structure, and SOFWERX publishes it in the most detail. The pattern is recognizable across hubs even where the labels differ.
The published assessment event sequence
A USSOCOM event published for 2026 puts the collaboration event in mid-July, opens submissions a month later for roughly a month, downselects inside ten days, and holds the assessment event in late October. Call it three months from the first public signal to a room with the evaluation panel, and an unstated further period to an executed agreement. That is genuinely fast for a competitive federal process. It is not fast enough to be a cash-flow plan.
The event pages also carry the eligibility fine print, and it is not decorative. Events commonly restrict participation to U.S. persons and state that an award may carry NIST SP 800-171 obligations. If your security posture is not ready for that, the time to find out is during phase 1, not during phase 5.
What the hub model does not give you
The speed is real. So is the cost, and it is paid in things that matter later rather than things that hurt now.
Protest rights are thin to absent. GAO's bid protest jurisdiction under the Competition in Contracting Act reaches solicitations and awards of procurement contracts by federal agencies. A business-to-business agreement between a nonprofit and a company is not one. Neither is an other transaction. If a downselect goes against you, expect a courtesy explanation rather than a remedy, and price that into how much you spend chasing one.
Past performance does not accumulate the way you expect. A federal contract generates a CPARS record you can point at for years. A sub-award from a nonprofit generally does not. The work is real, the reference is real, and the government stakeholders who watched the demonstration are real, but the machinery that turns delivery into a citable rating is not automatically engaged. Ask, early, what record of performance will exist when the work is done, and get an answer from the sponsor rather than from the hub.
The public record shows the intermediary, not you. This is the structural feature, and GAO has documented what it looks like at scale in a related context. In COVID-19 Contracting: Actions Needed to Enhance Transparency and Oversight of Selected Awards (GAO-21-501, July 26, 2021), GAO found that DoD obligated $7.2 billion through a single consortium management firm, which distributed nearly all of it to five pharmaceutical companies receiving between $450 million and $2 billion each — while the federal procurement database showed the funds going to the intermediary. GAO issued 14 recommendations aimed at transparency and oversight. That report is about consortium management firms rather than partnership intermediaries, but the reporting mechanic is identical, and it explains why competitor research that relies on USAspending will not see this money.
The last point cuts both ways. Your competitors cannot see your hub work either. For firms that would rather not advertise where their capability is landing, that is a feature. For firms that need a visible award history to open the next door, it is a real gap, and it argues for pairing hub activity with at least one instrument that produces a public record.
A non-FAR pathway is not a rules-free pathway
The government's money still has a purpose, a time, and an amount, and fiscal law still governs all three. Export control follows the technology regardless of instrument. Classification and controlled-unclassified-information handling follow the data. The False Claims Act reaches claims for federal funds however they are packaged. And because the FAR clause that would have imposed a security baseline is absent, sponsors write the requirement into the agreement by hand — which is why NIST SP 800-171 language shows up on event pages next to the technical description.
The SOCOM pilot, and why its status matters
There is a piece of this story that most write-ups miss, and it is the one a company planning around SOFWERX should check first.
Section 851 of the FY2020 National Defense Authorization Act (Public Law 116-92) created a pilot program letting the Commander of U.S. Special Operations Command increase small business participation by working through a partnership intermediary. Section 852 of the FY2022 NDAA (Public Law 117-81) rewrote it into something more specific and more interesting: under the amended section, the Commander makes covered awards, defined as awards under the Small Business Innovation Research program, to small business concerns through a partnership intermediary, funded by SBIR dollars transferred to SOCOM from the services and other DoD components under 15 U.S.C. § 638(f).
The guardrails are written into the statute. No component may transfer more than 10 percent of its SBIR funds in a fiscal year. The aggregate transferred to the Commander may not exceed $20,000,000. None of those funds may pay the intermediary for any costs associated with the pilot. And the Commander reports annually to the defense committees and both small business committees, naming each recipient, each amount, and the role the intermediary played.
The termination provision reads: "The authority to carry out a pilot program under this section shall terminate on September 30, 2025." We reviewed the text of the FY2026 National Defense Authorization Act (Public Law 119-60, enacted December 18, 2025) and the phrase "partnership intermediary" does not appear in it.
Draw the right conclusion from that. It does not mean SOFWERX is winding down — the hub is publishing events for 2026 and listing PIA sub-awards among its instruments, and the underlying partnership intermediary agreement rests on 15 U.S.C. § 3715, which has no sunset. What lapsed on paper is the narrow pilot authority to route SBIR money through an intermediary at SOCOM. If your plan depends on that specific pathway rather than on the hub generally, confirm its current status with the hub before you build a schedule on it. This is exactly the kind of detail that separates a company that has read the statute from one that has read a conference slide.
Hub, consortium, or contract
The hubs are one of several non-FAR doors, and they are not interchangeable. The comparison that matters is on the axes that change your decisions.
| Axis | DEFENSEWERX hub | OTA consortium | SBIR | FAR contract |
|---|---|---|---|---|
| Who you contract with | The nonprofit, or the government, depending on the instrument selected | The government, via a project agreement under the consortium's base OTA | The government | The government |
| Cost to get in the room | Generally none; ERDCWERX states there is no cost to join its ecosystem | Membership dues, from nothing to a few thousand dollars a year | None | None |
| What starts the clock | A sponsor with a problem and an event calendar | A request for prototype proposals released to the membership | A published solicitation on a fixed annual cycle | A funded, documented requirement |
| Intellectual property | Negotiated per instrument; a PIA sub-award and a § 4022 agreement have different defaults | Negotiated in the project agreement; the base agreement sets the frame | SBIR data rights under DFARS 252.227-7018, carried forward | DFARS 252.227-7013 and -7014 defaults apply |
| Public record of the award | Often none visible; the intermediary appears rather than the performer | Frequently shows the consortium manager rather than the member | Public award data by firm and phase | Full FPDS record |
| Route to production | Only if the instrument is a § 4022 agreement written to contemplate a follow-on | Non-competitive follow-on under § 4022(f) when the criteria and language line up | Sole-source Phase III with no ceiling | Recompete, or priced options |
Most firms that use these pathways well use more than one. A hub gets you in front of operators and program staff who will tell you what is actually broken. An SBIR gives you funded development with data rights that survive. A prototype other transaction gives you a production pathway. The mistake is expecting any one of them to do all three.
A practical way in
The hubs are unusually open, which is both the point and the trap. Anyone can join. That means being on the list is worth close to nothing on its own.
Join the ecosystems, all of them, in an afternoon. Registration is a form and it is free. ERDCWERX states plainly that there is no cost to join and that any individual, company, or agency may participate. The value of registration is the announcement email, which arrives days or weeks before an opportunity becomes widely known.
Read the sponsor, not the hub. The hub is a venue. The sponsor decides. Before you spend a week on a submission, find out which program office owns the problem, whether it has money this year, and whether the event is a real acquisition step or a market survey. Those are different events and they look identical from the outside.
Attend the collaboration event. It is the cheapest advantage in the whole process. The operators describe the problem in their own words, the constraints they care about surface, and you learn whether your capability is close or merely adjacent — before you write anything.
Ask which instrument is intended, and ask early. The published list has five entries and they are not equivalent. A sponsor that expects a § 4022 prototype agreement is planning differently than one expecting a CRADA. You are entitled to ask, and asking marks you as someone who has done this before.
Bring the background intellectual property schedule with you. Whatever the instrument, someone will want a license to something. Arrive with a written list of the models, libraries, datasets, and components that predate the project. Anything unlisted risks being treated as developed under the agreement.
Pair the hub with an instrument that leaves a record. Because hub work is largely invisible in public award data, decide up front what artifact will prove the work happened: a named government reference, a documented demonstration result, a CRADA, or a parallel federal award.
Common objections, answered plainly
Is this just a way around competition?
The events themselves are competitive and the sponsors say so. SOFWERX event materials state that an assessment event "is considered competitive and solutions will be evaluated independently of one another primarily for technical merit." What the model avoids is the FAR's procedural apparatus, not the comparison between offerors. The genuine oversight question is not whether anyone competed. It is whether the public can see who won, and the answer there is often no.
Does the nonprofit decide who gets the award?
No. The sponsor's evaluation panel downselects and the sponsor's stakeholders sit in the one-on-one sessions. The intermediary runs the process and can execute an agreement in its own name. That distinction is exactly why the model works and also why it draws oversight attention — a private party is doing process work that sits close to an award decision.
Do the hubs replace SBIR?
They do not. SBIR is a statutory program at 15 U.S.C. § 638 with its own data rights and its own sole-source route to Phase III. The hubs are a discovery-and-engagement layer that can end in several instrument types, one of which, under the SOCOM pilot, was an SBIR award itself. Firms that treat them as alternatives usually end up with neither.
Is one membership enough to reach every hub?
The parent organization promotes joining the ecosystem of multiple hubs at once, and the sign-up burden is low either way. But the announcement traffic, the sponsor's priorities, and the instruments in play differ hub by hub. Register broadly and then pay attention narrowly, to the one or two sponsors whose problems you can actually solve.
Bottom line
The DEFENSEWERX hubs solve a real problem. A laboratory or a command that wants to see what industry has built should not need nine months and a source selection plan to look at it, and the intermediary model compresses that to about three. For a company with a working capability and no federal history, a hub is one of the few places where a first-rate demonstration counts for more than a first-rate proposal.
What the model does not do is generate the things a federal business is built from. It rarely leaves a public award record. It generally does not produce a performance rating. It gives you almost no recourse when a decision goes the wrong way. Those are acceptable trades for access and speed, and they are bad trades if access and speed are all you get.
Use the hubs to find out what a program office actually needs, and to be in the room when it says so. Then be deliberate about which instrument closes the work, because that choice — not the event, not the pitch — decides what you own, what you can point at, and whether the prototype has anywhere to go.
Frequently asked questions
An agreement under 15 U.S.C. § 3715 between a federal laboratory and a state or local government agency, or a qualifying nonprofit, under which the intermediary performs services that increase the likelihood of success in the laboratory's cooperative activities with small business firms. DoD has a parallel provision at 10 U.S.C. § 4124(f) for its Centers for Science, Technology, and Engineering Partnership, broadened by section 214 of the FY2024 NDAA to cover inbound technology transfer and workforce development.
Registration is a form and the hubs present it as free. ERDCWERX states there is no cost to join its ecosystem and that any individual, company, or agency may participate. Registration gets you the announcement email. It does not get you an award, and every competitor can do the same thing in the same afternoon.
Realistically, no. GAO's protest jurisdiction under the Competition in Contracting Act reaches solicitations and awards of procurement contracts. A business-to-business sub-award from a nonprofit is not one, and neither is an other transaction. Plan for a courtesy explanation rather than a remedy.
Frequently not in a way that names the performer. When money moves through an intermediary, the federal procurement database tends to record the intermediary. GAO documented that pattern for consortium management firms in GAO-21-501, where $7.2 billion obligated through one firm reached five companies that the database did not identify. Assume your hub work will be invisible to outside researchers and plan your evidence accordingly.
It depends on what you are trying to build. A prototype agreement under 10 U.S.C. § 4022 is the one with a non-competitive follow-on production pathway attached, which makes it the most strategically valuable when the sponsor writes the agreement to contemplate one. A PIA sub-award is faster and lighter but carries less downstream structure. Ask which one the sponsor intends before the submission window closes.