Author:
R&D Tax Advisors
Role:
CPAs
Publish Date:
Mar 27, 2026
As the talent war for elite AI and frontier technology engineers drives compensation packages into the seven and even nine-figure range, the Internal Revenue Service has shifted its focus. For the C-suite of growing tech companies, this means that claiming Research and Development (R&D) tax credits for highly paid individuals (HPIs) now requires a significantly higher burden of proof.
Recent developments, including the landmark George v. Commissioner ruling and updated IRS audit techniques, signal that the traditional methods of substantiating these credits are no longer sufficient to survive a modern examination.
The Shift: From Project to Person
Historically, R&D audits focused on whether a company’s projects met the statutory "four-part test". However, when an individual’s W-2 reflects millions in stock-based compensation, the IRS now prioritizes the specific daily activities of that individual.
IRS examiners are increasingly trained to look past prestigious titles like "Chief AI Scientist" to determine if the individual was performing hands-on experimentation or merely providing high-level strategy, recruiting, and external evangelism. To the IRS, high compensation functions like a venture capital investment in a person’s future potential; however, for tax credit purposes, only the time spent on qualified research is eligible.
The Death of the "Shortcut Estimate"
The most significant risk to these high-value claims is the recent clarification regarding the Cohan rule. For decades, taxpayers used this common-law principle to bridge documentation gaps with reasonable estimates and testimony.
In George v. Commissioner, the U.S. Tax Court effectively ended the era of the "shortcut estimate" for R&D credits. The court stated it would not apply the Cohan rule if a taxpayer provides "no evidence at all that would permit an informed estimate". Furthermore, the court signaled that oral testimony without corresponding documentary evidence will be discounted or excluded entirely. For founders, this means that your lead engineer's memory of their work is no longer a substitute for a contemporaneous paper trail.
Building a Defensible "Dossier"
To mitigate the risk of disallowance, leadership must ensure their technical teams are producing engineering-grade artifacts in real-time. This requires a proactive plan that moves beyond retrospective annual studies and instead captures the "how" of the research as it happens.
A defensible R&D credit position for an HPI must include three layers of substantiation:
Business Value Context: Explaining why the compensation is so high (e.g., scarcity value, retention risk) to address initial auditor skepticism.
Product/Capability Output: Proving the individual delivered a tangible technical outcome, such as a new model or cost-saving architecture.
Qualified Research Evidence: Providing the core audit defense through experiment logs, technical architectures, version control commits, and activity logs from AI-assisted coding platforms.
Strategic Action for Growing Tech Companies
For CEOs of rapidly scaling companies, the takeaway is clear: the high value of your engineers’ compensation drives the risk of an audit, but only hard evidence of their research activities proves eligibility. Waiting until tax season to document these activities is a high-risk strategy that likely will not survive a modern IRS challenge. Implementing sustainable processes to capture technical uncertainties and experimentation concurrently with the work is now a fundamental requirement for protecting non-dilutive capital.
