The research credit for biotech and pharma: from bench to Phase III

Drug and biologic development is close to the paradigm case for Section 41: long, expensive, systematically documented experimentation on questions — does the molecule work, is it safe, can it be manufactured at scale — that stay genuinely uncertain for years. Most preclinical and clinical spend through approval can generate qualified research expenses. The claims that fail do so at the edges: contract research organization agreements read too casually, orphan drug credit dollars double-counted, foreign trial sites swept in, and collaboration deals that turn the sponsor's research into someone else's funded research.

The business component: molecule, indication, and formulation

Under Section 41(d)(2)(B), a business component is a product, process, formula, technique, invention, or software item. In life sciences the natural units are the drug candidate itself (a product and, frequently, a formula), its formulation and delivery mechanism, and the manufacturing process — cell line development, purification, lyophilization, scale-up — which is a separate process business component with its own qualification analysis, exactly as in manufacturing claims.

The four-part test is rarely the hard part here. Uncertainty about capability and appropriate design is the industry's defining condition; hypotheses, protocols, and statistical analysis plans are a textbook process of experimentation; and the work is technological in nature by any measure. The discipline is in scoping: a new indication for an approved compound restarts uncertainty and can requalify clinical work, while line extensions driven purely by marketing (a new package size, a flavor change) generally do not.

Phase by phase: what qualifies when

How each stage of the development pipeline typically fares under Section 41(d):

Phase III deserves a note because examiners occasionally argue late-stage trials are "confirmatory." The better reading — and industry practice — is that efficacy and safety at scale remain uncertain until the pivotal data exist; a program that failed in Phase III was, by definition, still uncertain. Document the open questions each protocol was designed to answer and the argument largely answers itself.

Section 41 versus the orphan drug credit

Section 45C gives a 25% credit for qualified clinical testing expenses incurred between FDA orphan designation and approval for a rare-disease indication. It is more generous than Section 41 in two ways: the rate, and the fact that contract research counts at 100% rather than 65%. The catch is Section 45C(c): expenses taken into account for the orphan drug credit cannot also be QREs, and vice versa. There is no double dip.

The planning consequence is an allocation exercise, not an either-or. A sponsor running an orphan-designated Phase II alongside non-orphan discovery work routes the designated clinical testing expenses to Section 45C and everything else — discovery, preclinical, CMC, non-orphan indications — to Section 41. Expenses outside 45C's definition (clinical testing, post-designation, for the designated indication, inside the U.S. absent an insufficient-population exception) stay available for Section 41 regardless. Both credits land on the general business credit and both are claimed with their own forms; the Section 41 side runs through Form 6765, with Section G business-component reporting now standard.

CRO agreements: rights, risk, and who claims the research

Almost every sponsor outsources: CROs run trials, central labs run assays, CDMOs develop and make clinical material. Payments qualify as contract research expenses under Section 41(b)(3) — includible at 65% — only if the research would be qualified had the sponsor performed it, the sponsor retains substantial rights in the results, and the sponsor bears the economic risk. The mirror image is the funded research exclusion: to the extent the CRO is paid regardless of outcome and keeps no substantial rights, the research is funded as to the CRO, which is why CROs themselves rarely have valid claims on sponsor work.

The typical fee-for-service CRO master agreement places both factors with the sponsor — the sponsor pays per the budget whether or not the trial succeeds, and owns the data and inventions. But contracts vary, and Populous Holdings v. Commissioner (T.C. 2019), though an architecture case, states the framework examiners apply to any services contract: look at payment terms for risk and at the IP and data clauses for rights — see the case brief. Terms that should trigger review: success fees or milestone-contingent CRO compensation (shifts risk toward the CRO), CRO retention of platform IP or data rights (dilutes the sponsor's rights), and pass-through investigator grants, which are generally still contract research at 65% when paid through the CRO.

Worked numbers for a Phase II year, illustrating the 65% haircut and the exclusions:

Wage QREs$2,400,000

Supply QREs$900,000

CRO fees at 65%$1,950,000

Excluded: foreign sites$1,200,000

Excluded: regulatory admin$500,000

Illustrative program from the worked example; foreign sites and regulatory administration generate no QREs, and CRO fees enter at 65%.

Supplies in trials: drug product, comparators, and lab consumables

Supply QREs under Section 41(b)(2)(C) are tangible property, other than land and depreciable property, used in the conduct of qualified research. In a clinical program that means clinical trial material — API, formulated drug product, placebo — consumed in the trial, plus lab consumables, reagents, and animals in preclinical work. Purchased comparator drug used in an active-control arm is consumed in the experiment and is generally a supply QRE, which matters because comparator sourcing can be one of the largest line items in a head-to-head trial. What is not a supply: depreciable lab equipment, capitalized manufacturing equipment, and — a recurring adjustment — cloud and software costs, which are analyzed separately. The full expense taxonomy is in the QRE guide.

Manufacturing clinical material in-house adds a wrinkle: the materials consumed in process development batches (engineering runs, failed lots, stability samples) are supplies of the process business component, while material produced by a locked, validated process for a late trial edges toward production. As with manufacturing pilot runs, document when the process met its basic functional and economic requirements.

FDA-driven work: the qualifying core and the excluded shell

That an activity is required by FDA neither qualifies nor disqualifies it. The regulator's demands mostly force more experimentation — tox packages, validation studies, human factors testing — and that experimentation qualifies on its own terms. The exclusions bite on the administrative shell around the science:

• Qualifies: studies designed to resolve safety, efficacy, stability, or manufacturability uncertainty, even when run solely because FDA requires them; assay development and validation; process validation batches where the outcome is genuinely uncertain.
• Does not qualify: compiling and formatting submissions; responding to administrative information requests; GMP quality assurance on routine production; post-approval adverse-event surveillance; advisory-committee and label negotiations; market-access and pricing studies (Section 41(d)(4)(D) excludes surveys and studies of that character).

Funded research on collaboration and milestone deals

Licensing and collaboration structures are where biotech claims get complicated. Under Section 41(d)(4)(H) and Treas. Reg. §1.41-4A(d), research is funded to the extent the researcher neither bears financial risk nor retains substantial rights. Applied to common structures:

• Upfront payments with a committed research plan. If the license obligates the biotech to perform defined research and the upfront compensates it regardless of scientific outcome, that research is funded to that extent — the partner is buying the work.
• Milestone payments contingent on scientific success. Payment only upon IND acceptance, positive Phase II data, or approval is payment contingent on success. The biotech bore the risk of the spend between milestones; that research is generally unfunded and remains claimable, subject to the rights analysis.
• Cost-sharing collaborations. Funded to the extent of the partner's reimbursement; the biotech's unreimbursed share can still qualify, expense by expense.
• Retained rights. A biotech that licenses a compound exclusively, worldwide, in all fields, and keeps no right to use the results in its own ongoing research may fail the substantial-rights prong even where it bore risk. Field- or territory-limited licenses that leave the biotech using the underlying platform usually preserve substantial rights.

The analysis is contract-by-contract and clause-by-clause, and it cuts both ways — a pharma partner reimbursing a biotech's research on a pay-regardless basis may itself be entitled to contract research QREs at 65%.

Where biotech claims fail

The recurring exam adjustments in life sciences, roughly in order of frequency: foreign trial sites claimed as QREs; the 45C/41 overlap unreconciled; executive and business-development wages allocated to research without support; CRO invoices claimed at 100% rather than 65%; and post-approval work claimed on the theory that FDA involvement equals research. None of these is a science dispute — they are contract-reading and bookkeeping disputes, which is the good news: they are preventable at claim-build time. Start from the IRS research credit overview, scope business components at the program-and-indication level, and read every third-party agreement before its dollars enter the pool.