If your firm bills clients for the work that also happens to generate your R&D tax credit, a new Tax Court decision should make you nervous. In June 2026, the Tax Court ruled on Smith v. Commissioner (T.C. Memo. 2026-50) — a case involving one of the world's most acclaimed architecture firms, the studio behind some of the tallest buildings on earth — and the outcome was a wake-up call for architects, engineers, consultants, and any professional services firm that innovates under a client contract rather than on its own dime.
The firm lost the R&D credit on two of six sample projects entirely. On four more, it could only claim a credit for the sliver of expenses that exceeded what the client actually paid. If a firm with this much research sophistication and legal support couldn't fully clear the bar, a lot of smaller design, engineering, and consulting shops are almost certainly claiming credits they can't defend.
The Case in Brief
The taxpayers were the partners of a boutique architecture firm known for "supertall" buildings — structures that push the limits of physics, requiring genuine research into thermodynamics, geotechnics, fluid dynamics, and microclimate behavior just to make the design work. This isn't a firm sketching floor plans; it's a firm running the kind of iterative, uncertain technical investigation that the R&D credit was designed to reward.
The IRS challenged the firm's research credit claims across multiple tax years, arguing that the research behind these building projects was "funded" by the clients who commissioned the buildings — and funded research doesn't qualify for the credit, no matter how innovative it is. The Tax Court examined six representative projects in detail and applied a two-step framework that every client-funded firm needs to understand.
The Two-Step Test That Decides Everything
Under Internal Revenue Code Section 41(d)(4)(H), research paid for by someone else — a client, a grant, a government contract — is excluded from the R&D credit unless the taxpayer can clear two separate hurdles.
Step 1: Do You Retain "Substantial Rights"?
The first question is whether your firm keeps meaningful rights to use and benefit from the research it performs, independent of the client relationship. In Smith, two of the six sample projects failed here immediately: the client contracts assigned copyright in the work product to the client and required the firm to get written client approval before reusing anything it had learned or developed on future projects.
That's a common clause. Plenty of standard architecture, engineering, and design contracts include work-for-hire or IP-assignment language because clients want to own what they paid for. The Tax Court's message is blunt: if you sign away the right to reuse your own research without asking permission first, you've given up "substantial rights" — and the credit is gone for that project, full stop, regardless of how innovative the work was.
Step 2: Did You Bear the Economic Risk of Failure?
For the four projects where the firm did retain substantial rights, the court moved to the second test: was payment contingent on the research actually succeeding? Here the firm lost again, at least partially. Its contracts paid out on a phase-completion or monthly-progress basis — a standard billing structure for design and consulting work — rather than tying payment to whether the underlying research or innovation actually worked.
The court held that phase- or time-based billing does not count as bearing economic risk, because the firm got paid whether or not the research panned out. That's the whole point of milestone billing from a cash-flow perspective, but it's exactly the kind of protection from failure that the funded-research exclusion is designed to catch.
The one piece of good news: the court didn't disallow those four projects outright. It ruled that the firm could still claim a credit for research expenses that exceeded what the client paid — the theory being that any spending beyond the contracted price was genuinely put at the firm's own risk, unfunded by the client. That's a meaningful partial win, but it also means most of the credit on client-billed engagements evaporates unless your actual research costs are running ahead of client payments.
Why This Should Worry More Than Just Architects
The funded-research exclusion isn't new — it's been part of Section 41 for decades, and courts have applied the substantial-rights and economic-risk tests before (the Eighth Circuit's 2024 Meyer, Borgman & Johnson decision covered similar ground for a structural engineering firm). What makes Smith significant is how specifically and skeptically it applied both prongs to ordinary professional-services contract language: IP-assignment clauses, client-approval-for-reuse provisions, and phase-based billing. Those aren't exotic terms — they're standard boilerplate across architecture, engineering, industrial design, software consulting, and R&D-for-hire businesses of every size.
If your firm claims the R&D credit for work performed under client contracts and your engagement letters or master service agreements include any of the following, Smith suggests your credit is exposed:
- Client owns the deliverables and IP outright, with no carve-out letting you retain rights to methods, know-how, or reusable research.
- You need client permission to reuse learnings from one project on future work.
- Billing is tied to time, phases, or milestones rather than to whether the research goal was actually achieved.
- You've never separated "the fee for delivering the project" from "the cost of the research risk you personally bore." If total research spend never exceeds total client payment, there may be little or nothing left to claim under the Smith framework.
What Small and Mid-Size Firms Should Do Now
You don't need to be a supertall-building architecture firm to be affected by this ruling — any small engineering shop, product design studio, or technical consultancy claiming the credit on client work should treat Smith as a documentation and contract-drafting wake-up call, not just case-law trivia.
- Review your standard contract templates for IP and reuse language. If every engagement assigns full IP to the client with no reservation of rights, talk to counsel about adding a limited license-back or "retained know-how" clause before your next tax year closes.
- Track research costs separately from billed fees, project by project. Under the Smith partial-credit theory, you can only claim a credit for spend that exceeds what the client paid. Without project-level cost tracking, you have no way to prove that excess exists.
- Don't assume milestone billing disqualifies you outright — but don't assume it's safe either. The test is about whether payment depended on success, not just about when cash arrived. Contracts that explicitly tie some portion of payment to achieving a defined technical outcome are on stronger footing than pure time-and-materials or phase billing.
- Revisit R&D credit claims for open tax years with a professional who has read the full opinion, especially if your firm's contracts look like the ones in Smith. An overstated credit claim discovered at audit is a far more expensive problem than a conservative claim caught in review.
Where Clean Books Fit In
Every piece of this analysis — which projects had research spend exceeding client payments, which contracts carried IP-assignment language, which engagements were billed on milestones versus outcomes — depends on being able to trace costs back to individual projects and contracts with confidence. Firms that lump all client billings and research expenses into a handful of general ledger accounts have no way to reconstruct the project-by-project evidence a Tax Court opinion like this one demands.
Plain-text accounting makes that kind of project-level tracking a natural habit rather than a scramble at audit time. Beancount.io lets you tag transactions by project, client, and contract from day one, and keeps the full history in version-controlled, auditable files — so when a research credit question comes up two or three years later, you can pull the exact spend-versus-payment breakdown for a single engagement instead of trying to reconstruct it from memory. Get started for free and see why developers and finance professionals are switching to plain-text accounting.