SDVOSB is leverage. Use it.
Veteran-owned small businesses are sitting on a federal AI contracting wedge in 2026 that the SDVOSB community is mostly under-monetizing. The reason is not a lack of opportunity — the opportunity is real, the eligibility paths are clear, and the set-aside flow is structurally favorable. The reason is that the community has internalized a narrative about set-asides that treats them as a participation mechanism rather than as competitive leverage, and the comms coming out of GSA and OSDBU briefings reinforce that framing.
This essay is for the SDVOSB founder who has a real AI capability, has been on a federal call where the contracting officer mentioned set-asides as a side comment, and walked away unsure whether to lean into it. Lean into it. The math is more favorable than the comms have implied, and the engagement-class fit between SDVOSB engagement shapes and the current federal AI buying spec is closer than anyone is saying out loud.
A note on specifics: this essay walks the structural argument and the eligibility logic. The specific numbers — sole-source ceilings, NAICS-specific size standards, set-aside dollar thresholds — change on a rolling cycle as SBA updates size standards and as administration priorities shift. I’ll point you at the right places to confirm current numbers rather than print numbers in a blog post that might be six months out of date by the time you read it. The structural argument is durable. The specific dollars on any given day, check the source.
What “SDVOSB is leverage” actually means
The framing most SDVOSB founders absorb from federal procurement outreach is that the set-aside is access — it lets you compete for work that you’d otherwise be shut out of. That framing is correct as far as it goes, but it’s incomplete. The set-aside is also price discipline — it constrains the field of competitors, which compresses the negotiation surface in a way that benefits the qualified small vendor in a procurement that would otherwise reward only scale.
Think about what a contracting officer is doing when they direct a procurement into an SDVOSB set-aside. They are explicitly choosing not to optimize for “lowest qualified bid from anyone in the marketplace.” They are optimizing for “best qualified bid from a constrained field that satisfies a statutory preference goal.” The contracting officer’s incentive in that scenario is to find a credible SDVOSB vendor and award. The bidding economics for the vendor inside the set-aside lane are dramatically more favorable than they would be in open competition, because the field is small, the competitors are sized similarly, and the contracting officer is structurally motivated to make the award land.
The wedge is sharper still for sole-source SDVOSB awards. Under current FAR provisions, contracting officers have authority to award sole-source SDVOSB contracts up to specific dollar thresholds without full competition, when the contracting officer has a reasonable basis to believe a qualified SDVOSB exists and the award is in the government’s interest. The sole-source threshold is meaningful — it covers the size of engagement most federal AI pilots actually want to procure, and it removes the procurement-cycle overhead that kills the timeline on a lot of mid-sized AI work. The specific dollar threshold for sole-source SDVOSB awards is set in the FAR and updates periodically — check FAR 19.1406 and the SBA’s current size standards table for the version applicable to your fiscal year and NAICS code.
This is the leverage. Not access alone. Access plus a contracting-officer incentive structure that favors making the award happen quickly to a qualified vendor in the lane. The SDVOSB founder who walks into a federal AI conversation positioning their capability and their set-aside eligibility is offering the contracting officer something that the large integrators structurally cannot offer: a fast, defensible path to procurement that satisfies both the technical requirement and the statutory preference goal in one award.
The eligibility path, briefly
SDVOSB eligibility runs through SBA certification. The path has four core requirements: at least 51% ownership by one or more service-disabled veterans, day-to-day control of the business by one or more service-disabled veterans, status as a small business under the SBA size standard applicable to the NAICS code of the work being pursued, and a service-connected disability rating from the VA (or proof of a service-connected disability for veterans whose discharge characterized the disability without a numerical rating).
The certification process moved to SBA in 2023 under the Veteran Small Business Certification (VetCert) program. Self-certification for sole-source and set-aside contracts is no longer sufficient as of the transition deadlines; certified SDVOSB status through SBA is required for set-aside eligibility on most federal solicitations. The certification window typically takes a few months end-to-end, and the documentation requirements are exactly what you’d expect — DD-214, VA rating letter or equivalent, ownership documentation, financial statements, control documentation showing the SDV holds the highest-ranking officer role and the highest compensation. SBA’s VetCert portal walks the process. Do not rely on self-certification or expired VA-issued certifications for set-aside work; the contracting officer will check current SBA status before award and a mismatch will disqualify the bid.
The piece that gets missed: certified SDVOSB status also enables participation in joint ventures (JVs) with non-SDVOSB partners under specific structural rules (the SBA Mentor-Protégé Program is the most common vehicle). A small certified SDVOSB can JV with a much larger non-SDVOSB mentor under specific conditions and the JV is treated as small for set-aside purposes. The structural details (51%/49% workshare requirements, performance-of-work requirements, MPA approval timelines) are specific and the SBA documentation is the authority — but the strategic point is that a small SDVOSB capability can be paired with a larger partner’s delivery capacity through a structurally-sound JV without losing set-aside eligibility. This is the structural answer to “we’re too small to credibly deliver against a $5M ceiling on our own.” You’re not — not if you JV correctly.
The engagement-class fit
In last quarter’s essay on OMB M-25-21 and M-25-22, I argued that the federal AI acquisition memos read as a buying spec for vendor-portable, fixed-scope, audit-traceable AI engagements. The structural overlap between that buying spec and the SDVOSB engagement shape is the part most of the SDVOSB community hasn’t fully internalized.
Large integrators are not well-positioned to deliver fixed-scope, deliverable-based AI engagements in the $250k–$5M range. Their cost structure requires staffed long-duration contracts. Their proposal economics don’t pencil on a six-week build. Their internal review and approval processes consume more elapsed time than the engagement itself. They will continue to win the large MAC-vehicle work and the $50M+ enterprise modernization programs, because that’s what they’re built for. They will not aggressively pursue the $500k pilot that ships working AI in 90 days and hands the agency a clean IP transfer at the end, because that engagement does not feed the partner-leverage pyramid they need to keep operating.
The mid-sized integrators (the next tier down, 200–2,000 employees) can technically deliver this engagement class, but most of their federal practices are still organized around the staff-augmentation model that defined federal IT contracting for the last twenty years. They are adapting, but the adaptation cycle is slow and the proposal posture is wrong for the current memo language. A small certified SDVOSB with the right architecture and a sharp proposal can frequently outscore a mid-sized integrator on the same RFP, because the SDVOSB’s proposal is purpose-built for the engagement class while the integrator’s proposal is a slightly-customized version of their standard pitch.
The set-aside math compounds this. When the procurement is steered into an SDVOSB set-aside lane (full or partial), the integrator is out of the field entirely. The competition is among certified SDVOSBs. The competitive dimension that wins is fit — does the vendor’s specific AI capability match what the agency wants to build, do they have the architecture, do they have credible past performance, can they show up with a clean fixed-scope proposal that the contracting officer can defend on the back end. The competitive dimension that does not win, in that lane, is scale.
This is the wedge. The agency’s buying spec under M-25-22 favors small fixed-scope engagements. The set-aside math favors certified SDVOSBs in that engagement class. The combination is a structural advantage for the SDVOSB founder who has the capability and can execute the proposal correctly.
NAICS, size standards, and what to register for
The NAICS codes that matter for AI-specific federal work are 541511 (Custom Computer Programming Services), 541512 (Computer Systems Design Services), 541611 (Administrative Management and General Management Consulting), 541690 (Other Scientific and Technical Consulting), and 541990 (All Other Professional, Scientific, and Technical Services). The SBA size standards for these codes are set in the SBA Size Standards Table and they update on a rolling cycle — typically every five years, sometimes more frequently when SBA does category-specific reviews. The specific revenue or employee thresholds applicable in your fiscal year are at sba.gov/size-standards.
The strategic point on NAICS selection is that your primary NAICS on SAM.gov drives a lot of downstream surfacing — agencies search for vendors by NAICS, market research databases sort by NAICS, set-aside notifications get scoped by NAICS. Choose your primary NAICS based on the work you most want to be found for, not the work you most often actually do. Many SDVOSB founders set their primary NAICS to 541512 (Computer Systems Design) by default and miss that 541511 (Custom Computer Programming) is a stronger surface for application-build work, or that 541690 is the right primary for AI advisory engagements that don’t involve software delivery. The other NAICS codes go in your additional list — register for everything plausibly applicable, not just your primary.
SAM.gov registration itself is the floor. If you’re not registered with current size certifications, current NAICS, current capability narrative, and current past performance, you are invisible to the federal market regardless of your capability. The registration is free and the renewal cycle is annual; if you’ve let it lapse, your first move is to fix that, before anything else in this essay.
What “use it” actually looks like in practice
Concrete advice, because the structural argument is useless without a playbook.
One: Lead with set-aside eligibility in every federal conversation. Most SDVOSB founders are too modest about this. The contracting officer’s day is easier when they know you’re certified. Say it on the first call. Put it in the email signature. Put it on the website above the fold, not buried in an About page.
Two: Build a sole-source dossier, not just a capability statement. Capability statements are the right artifact for getting on a vendor list. Sole-source dossiers are the right artifact for being awarded sole-source work. The dossier includes the capability narrative, but also explicit past performance language with named agencies and named outcomes, a NAICS-coded list of the work you’ve delivered, a price reference (typical engagement ranges, not specific quotes), and an explicit statement of your sole-source eligibility under the applicable FAR provisions. The contracting officer who is considering a sole-source award needs all of this to defend the award on the back end. Make it easy.
Three: Pursue Mentor-Protégé strategically. If you have a capability that’s strong enough to win sole-source work in the $250k–$2M range on your own, you may not need an MPA. If you want to pursue ceilings above your independent comfort level, identify a mentor whose delivery capacity complements your capability, structure the JV against the SBA rules, and present the combination to contracting officers as a packaged option. The mentor benefits because they get exposure to set-aside work they couldn’t otherwise touch. You benefit because you can credibly bid larger ceilings without rewriting your operational model.
Four: Get your past performance language tight. Federal procurement runs on past performance. The vendor with three named agency past-performance references on a specific capability beats the vendor with twenty commercial logos every time. If you don’t have direct federal past performance yet, indirect past performance from federal-adjacent work (state, local, university, federally-funded research) is acceptable and you should write it up specifically. If you have no federal-adjacent past performance, your first move is to bid into an SBIR Phase I or an OTA — both have lower past-performance barriers and produce federal references for the next bid.
Five: Pick your lanes. The federal AI work surface is huge. The SDVOSB capacity is finite. Pick three to five agencies whose mission you actually understand, build relationships with the OSDBU offices and the relevant contracting officers, and work those lanes deliberately. The founder who shows up at one agency four times this year closes more than the founder who shows up at twenty agencies once.
The wedge has a clock
The current alignment between the federal AI buying spec (M-25-21/22), the engagement-class size that favors small vendors, and the set-aside math that favors SDVOSBs specifically is a moment. It will not last forever. The large integrators are adapting their delivery models. The mid-sized integrators are restructuring proposal teams. The certification environment is becoming more competitive as more veteran-owned firms enter the AI space. The wedge is real, the wedge is open, and the wedge is finite.
The SDVOSB founders who treat set-aside eligibility as competitive leverage rather than just access, who position against the current buying spec rather than the last decade’s, and who build sole-source dossiers and MPAs and named-agency relationships in the next twelve months will capture meaningful federal AI revenue at margins the open market would not pay. The ones who wait for the comms to tell them the math is favorable will read about it after the wedge has closed.
Use it.