What is funded research under Section 41?

Research is funded — and excluded from the performer's credit — to the extent another person pays for it in a way that does not depend on the research succeeding, or the performer retains no substantial rights in the results. If the performer bears the risk of failure and keeps rights to use the results, the research is unfunded and creditable to the performer.

Do fixed-price contracts qualify for the R&D credit?

Generally they favor the performer. Under Fairchild, payment contingent on delivering work that meets specifications — with the contractor eating the cost of failed efforts — is not funding, so the contractor bears risk and can claim the credit if it also retains substantial rights. Time-and-materials and cost-plus contracts usually shift risk to the customer instead.

Can both parties to a research contract claim the credit?

Not on the same dollars. The rights and risk tests are designed to assign the research to one side: the payer claims 65% of payments as contract research if it bears risk and holds rights; the performer claims its own costs if it bears risk and retains substantial rights. Examiners look for double claiming.

The funded research exclusion: who keeps the credit on contract R&D

Section 41(d)(4)(H) excludes from qualified research "any research to the extent funded by any grant, contract, or otherwise by another person (or governmental entity)." The word doing the work is funded, and forty years of regulations and case law have reduced it to two questions about the contract: who bears the financial risk of failure, and who keeps rights in the results. A contractor that answers "we do" to both can claim the credit on customer-paid research; a contractor that answers "the customer" to either cannot, however qualified the science.

The risk test: payment contingent on success

Treas. Reg. §1.41-4A(d) provides that amounts payable under an agreement that are contingent on the success of the research are not funding — the performer is, in substance, betting its own resources. Fairchild Industries v. United States (Fed. Cir. 1996) is the foundational application: under a fixed-price government development contract with detailed acceptance testing, Fairchild was paid only for work that met specifications and bore the cost of every failed attempt itself. The Federal Circuit held the inspection-and-acceptance and default clauses put the risk of failure on the contractor, so the research was unfunded to Fairchild — even though progress payments were made along the way, because those payments were recoverable if the work ultimately failed acceptance.

The contract-type heuristics follow directly, though none is conclusive without reading the clauses:

Fixed-price development contracts generally leave technical risk with the performer — cost overruns and failed iterations are the performer's problem.
Time-and-materials and cost-plus contracts generally pay for effort regardless of outcome; the performer is funded, and the customer may hold the creditable expense (at 65%, under the contract research rules).
Termination-for-convenience clauses, acceptance criteria, warranty and rework obligations, and milestone-payment refundability all move the needle and should be read, not assumed.

The rights test: substantial rights in the results

Independently, research is treated as fully funded if the performer retains no substantial rights in the results — §1.41-4A(d)(2). The rights need not be exclusive: a non-exclusive, royalty-free right to use the research in the performer's business suffices. But a contract assigning all intellectual property to the customer, with the performer needing permission or payment to reuse what it learned, fails the test outright — and unlike the risk test, which applies "to the extent" of funding, a total failure of rights disqualifies the project entirely regardless of risk allocation.

Geosyntec Consultants v. United States (11th Cir. 2015) illustrates the interplay. The engineering firm claimed credits for client-funded environmental projects; the Eleventh Circuit sorted the contracts by economic substance, holding the capped and fixed-price arrangements unfunded (Geosyntec bore the risk within the price) while its cost-reimbursable contracts were funded — the clients paid for effort, success or not. Later litigation (Dynetics, Populous Holdings, Meyer Borgman Johnson, Tangel, Smith, System Technologies) has continued the clause-level approach, with performers frequently winning on rights — courts have found retained know-how and reuse rights "substantial" even where clients owned project deliverables — and losing on risk where payment was assured.

Practical application

The funded research analysis belongs at the front of every claim by a firm doing custom work — engineering and architecture practices, contract manufacturers and developers, government contractors, grant-funded ventures (federal grants are funding; there is no risk of nonpayment). Three habits keep the position defensible:

Read every research-adjacent contract both ways. Inbound customer contracts determine what the firm may claim; outbound subcontracts determine what it may claim as contract research. Exam IDRs routinely request both — see audit defense.
Sort by clause, not label. A "fixed-price" contract with unlimited change orders billed at cost behaves like time-and-materials; a "T&M" contract with a hard cap and acceptance testing behaves like fixed-price. Geosyntec is the template.
Fix future contracts, not past ones. Where the pipeline is fixed-fee work, adding an express retained-rights clause (non-exclusive reuse of methods, tools, and know-how) is cheap and often decisive. Amending old contracts after the credit year persuades no one.

When the analysis goes against the performer, the dollars are not necessarily lost to everyone: the funding customer, if it bears the risk and holds the rights, may have a 65% contract research expense on the same arrangement. What the statute forbids is both parties claiming the same research — which is exactly what an examiner reconciling the two returns will look for.