VA T4NG2 for SDVOSBs: How to Win AI and Automation Work on the $60.7B Vehicle
As of late March 2026, the VA’s largest IT services contract is open for business. A U.S. Court of Federal Claims ruling denied the last 18 bid protests, clearing Transformation Twenty-One Total Technology Next Generation 2 (T4NG2) — a 10-year, $60.7 billion multiple-award IDIQ run by the VA Technology Acquisition Center in Eatontown, NJ — to start issuing task orders against a final roster of 33 prime contractors, roughly 17 of them SDVOSB-eligible. If you build AI or automation and you hold a verified SDVOSB, this is the single most important VA vehicle to understand, because of where the money is about to flow and because of a subcontracting rule most vendors overlook.
This is a spoke under the VA AI modernization pillar. It’s written for three readers: a prime BD lead who holds a T4NG2 slot and needs SDVOSB AI depth to round out a task-order team; a VA contracting officer (CO) or program manager who wants to route AI work through T4NG2; and another SDVOSB sizing up the vehicle. The mechanics matter, so this goes deep.
What exactly is T4NG2?
T4NG2 is the VA’s flagship contract vehicle for buying IT and health-IT services — a multiple-award, indefinite-delivery/indefinite-quantity (IDIQ) contract issued under solicitation 36C10B23R0011, with a base ordering period of five years plus one five-year option and a program ceiling of $60.7 billion. It is managed by the VA Technology Acquisition Center (TAC) at 23 Christopher Way, Eatontown, NJ, under the Office of Procurement, Acquisition and Logistics (OPAL), and it accepts orders through roughly October 2033. The primary NAICS code is 541512 — Computer Systems Design Services, with a $34 million size standard.
The scope spans eleven functional areas quoted in the contract’s performance work statement: program management and enterprise architecture; systems and software engineering; software technology demonstration and transition; test and evaluation; independent verification and validation; enterprise network; enterprise management framework; operations and maintenance; cybersecurity; training; and IT facilities. AI and machine learning are not a separately named functional area — instead, AI, RPA, and automation work flows through the software-engineering, enterprise-management, O&M, and cybersecurity areas, which is exactly how AI/RPA task orders ran on the original T4NG vehicle.
A practical note for buyers: T4NG2 is a services vehicle. Task orders can include software and IT products incidental to a solution, but it isn’t designed for straight product buys — for that, a CO would look to NASA SEWP or a GSA vehicle.
How does a task order actually get awarded on T4NG2?
Orders are competed among the 33 primes under FAR 16.505 “fair opportunity.” A VA contracting officer defines the requirement with TAC, issues a Request for Task Execution Plan (RTEP) to the eligible primes, and evaluates technical and price proposals. Task orders typically run between $1 million and $1 billion, with common scopes including enterprise data management, low-code/no-code integration, IT systems development, DevSecOps, legacy modernization, and network and desktop support, over performance periods of three to five years.
The step that matters for an SDVOSB happens before the competition: the VA Rule of Two. Under 38 U.S.C. §8127(d), codified in VAAR 819.70, the contracting officer must set the order aside for SDVOSBs whenever they reasonably expect two or more verified SDVOSBs will submit fair and reasonable offers. VAAR also authorizes a tiered (cascading) evaluation, with SDVOSB as the first tier and other veteran-owned small businesses as the second. So on a meaningful share of T4NG2 orders, the competition is restricted to the SDVOSB primes from the outset — which is the whole reason the SDVOSB-reserved seats are valuable. The deeper mechanics of that hierarchy are in the Veterans First contracting guide.
Why is now the moment? The T4NG-to-T4NG2 reflow
The original T4NG — a $22.3 billion vehicle awarded in 2016 — has obligated roughly $17.4 billion in task orders over its life, and its ordering period expires June 30, 2026. That sunset is forcing a wave of marquee task orders to recompete onto T4NG2, and that wave is the near-term opportunity.
The largest single piece is PTEMS — Product and Technology Ecosystem Management Services, about $1.2 billion, currently held by Booz Allen Hamilton (originally Liberty IT Solutions). On Booz Allen’s Q4 FY2026 earnings call in May 2026, its CFO explicitly attributed a 23% year-over-year decline in the firm’s Civil business to “the PTEMS contract roll-off.” Behind it sits a queue that industry trackers have catalogued: ITOPS (~$1B), Operations Support and Security Compliance (~$698M), Franchise Fund Support Services (~$579M), Cloud Operations and Migration Services (~$186M), and roughly two dozen smaller expiring tasks.
Why this matters for a small AI shop: every reflow recompete is a teaming inflection point. Primes rebuilding bids for these task orders need fresh past performance, fresh resumes, and credible differentiators — and AI/automation specialization is one of the few differentiators a prime can bolt onto a 2026 recompete that a buyer actually rewards.
Is T4NG2 actually usable now, or still tied up in protests?
It’s usable. But the path here was long, and a buyer or teaming partner is right to ask.
VA made initial awards to 30 contractors in October 2023, out of 173 proposals. What followed was a multi-year protest fight at the U.S. Court of Federal Claims: Sierra7 was disqualified on a size review and replaced by CGI Federal in March 2024; the court narrowed the challenges in October 2024; and in May 2025 VA took corrective action that added three more primes — Taurian Consulting, Peregrine Digital Services, and Technatomy — while agreeing not to issue task orders until at least March 2026. Twenty-six protests narrowed to eighteen, and on March 30, 2026, Judge Molly R. Silfen granted the government judgment on the administrative record, rejecting challenges from protesters including GDIT, T4NG2 JV LLC, and Veteran First Technologies LLC. As of this writing no stay remains and task orders can flow, though appeals to the Federal Circuit are always possible and worth monitoring.
There’s a governance footnote worth knowing, because it shapes the vehicle’s future. In a January 2024 letter, the House Small Business Committee challenged the VA’s math: T4NG2’s ceiling rose 172% over T4NG’s $22.3B while the awardee count rose only about 43% — the Committee’s view being that 30 primes is too few for a $60.7B vehicle. That pressure is one reason to watch for any future roster expansion or on-ramp.
Who are the SDVOSB primes — and which ones need AI depth?
This is the teaming targeting list. Of the 33 final primes, roughly 17 are SDVOSB-eligible, most of them structured as joint ventures (often SBA Mentor-Protégé JVs pairing a small SDVOSB with a larger mentor). The ones most likely to need an AI/automation subcontractor are those competing on health-IT modernization, ServiceNow, low-code, document automation, and DevSecOps:
- Peregrine Digital Services — IronArch (SDVOSB, known for ServiceNow and cloud modernization) + Ad Hoc JV.
- T4 Designs — NDi (SDVOSB) managing partner; network, cyber, cloud.
- Veterans EZ Info (VetsEZ) — SDVOSB with a data-engineering footprint at VA’s Office of Enterprise Integration.
- Zetta Solutions — GigaTECH (SDVOSB protégé) + Sierra7 mentor JV; healthcare IT.
- Clear Vantage Point Solutions (CVPS) — PingWind (SDVOSB cyber) + Peraton mentor JV.
- SCIO JV — Sprezzatura (SDVOSB advisory) + GovCIO mentor JV.
- Taurian Consulting — Guidehouse + ReefPoint Group (SDVOSB) JV; digital and integration work.
Other SDVOSB-eligible primes on the roster include 1Tech JV, A2E Digital Solutions, ATL-NG, Canopy Health (LogC2), Digipathy, Dynamic Government Resources 2, H2 Technology Group (HigherEchelon + SAIC), Optimal Link (the WOSB-SDVOSB seat), RP and Partners, and VCH Partners (the HUBZone-SDVOSB seat). The large-business primes — useful as mentor-protégé targets — include Accenture Federal, Booz Allen, CGI Federal, Cognosante, Credence, Deloitte, ECS Federal, GovCIO, ManTech, and SAIC.
Verify any specific prime’s current SDVOSB status against SBA VetCert and SAM.gov before approaching — the roster shifted repeatedly through the protest process, and at least one award (JTech) was cancelled.
How does a new SDVOSB AI shop get on T4NG2?
You can’t prime it. There is no announced on-ramp and no rolling admission; the 33-prime roster is closed for the foreseeable future. The realistic paths are two, and the second is the lever most vendors miss.
Path one: subcontract to a T4NG2 prime. Primes assemble teams order by order. The entry requirements are a teaming agreement, a capability statement aligned to the specific task order’s performance work statement, ATO-ready personnel, and — critically — verified SDVOSB status. The full playbook is in the T4NG2 teaming guide.
Path two: a Mentor-Protégé Joint Venture. Several of the SDVOSB primes above are themselves mentor-protégé JVs. For a brand-new SDVOSB, partnering as a niche-AI sub to one of those protégés is often faster than chasing a mentor directly.
Here’s why a verified SDVOSB AI sub is worth more to a T4NG2 prime than a non-SDVOSB sub with identical skills. On an SDVOSB-set-aside services order, FAR 52.219-14 limits the prime to subcontracting no more than 50% of the work to firms that are not “similarly situated.” A similarly situated entity is a first-tier subcontractor with the same small-business program status as the prime — i.e., another verified SDVOSB that’s small under the order’s NAICS — and crucially, work performed by a similarly situated entity counts toward the prime’s required 50% self-performance, not against its subcontracting cap. VA’s own VAAR 819.7009 reinforces it: only VIP-listed SDVOSBs are considered “similarly situated” on an SDVOSB order.
The practical consequence: when a T4NG2 prime swaps a non-SDVOSB AI sub for a verified SDVOSB AI sub, it gains compliant subcontracting headroom for free. That’s the conversion lever — a credentialed SDVOSB AI specialist isn’t just another vendor on the prime’s bench; it’s a compliance asset on every Veterans-First-set-aside task order the prime chases.
What AI and automation work actually runs through T4NG2?
Enough precedent exists on the original T4NG to make the claim concrete rather than aspirational. The VA Office of Information Security ran an AI/RPA program as a T4NG SDVOSB set-aside, with a performance work statement requiring RPA bots built on TRM-approved tools (UiPath, PowerShell) deployed on the VA Enterprise Cloud’s Azure Government environment, using DevSecOps practices. Other T4NG precedents include a $12M Heuristic Behavior Analytics AI/data-analytics security order, and a set of Sierra7 SDVOSB orders — LEAF, Data Discovery Analytics and Labeling, ArcGIS Indoors support — that are exactly the kind of work reflowing onto T4NG2.
The VA’s published AI strategy names the workloads it’s buying: AI-driven RPA, OCR, and NLP for claims and document workflows, plus retrieval and assistant patterns across veterans-facing and internal systems. For an SDVOSB doing fixed-scope 90-day builds, the accessible slice is document processing, RAG over VA policy, RPA on administrative tasks, and contact-center augmentation — delivered as a sub-tier specialist on a prime’s task order. The broader build-vs-team picture is in the VA pillar.
What does performing on a T4NG2 order require for security — and is CMMC one of them?
No, CMMC is not required — and getting this right with a contracting officer matters.
A T4NG2 contractor inherits the VA’s stack: the VA Enterprise Cloud (VAEC) on AWS GovCloud and Azure Government, both at FedRAMP High, each with a VA-issued Authority to Operate. On top of that sit VA Handbook 6500 and 6517 (the information-security and risk-management overlays), NIST SP 800-53 controls, Section 508 accessibility, HIPAA and PHI handling for any VHA workload, and a project-level ATO that teams accelerate by inheriting VAEC documentation. CMMC — the Cybersecurity Maturity Model Certification — is a DoD construct under DFARS 252.204-7021 that doesn’t appear in VA contracts.
For a Cloudflare-native, FedRAMP-aware shop, the honest and complete positioning on T4NG2 is: deployed on FedRAMP High-aligned platforms, building to support the prime’s VA ATO and Handbook 6500/6517 package, FedRAMP-aware and not CMMC-certified — and CMMC is not a T4NG2 requirement. That honesty is a differentiator, not a gap; the full framing is in FedRAMP-aware, not CMMC-certified.
Frequently asked
What is T4NG2?
Is T4NG2 issuing task orders yet?
How many T4NG2 primes are SDVOSBs?
Can a new SDVOSB become a T4NG2 prime?
Does T4NG2 require CMMC?
What's the biggest task order moving onto T4NG2?
Working with Truvisory
Truvisory is an SBA-verified SDVOSB that builds working AI and automation on a Cloudflare-native, FedRAMP-aware architecture — fixed-scope, in 90 days.
If you’re a T4NG2 prime that needs verified-SDVOSB AI depth to strengthen a task-order team — and to expand your similarly-situated subcontracting headroom on Veterans First orders — start with a scoping call. For the full VA picture, see the VA AI modernization pillar; for the companion 100%-SDVOSB software vehicle, see SPRUCE.