Thirteen days apart this July, two different courts struck down two different Treasury regulations, and neither needed the Administrative Procedure Act to do it. Both cases were decided on summary judgment as pure questions of law. Both courts expressly declined to reach the APA arguments before them. That’s notable. Because a regulation that fails on notice and comment can be re-proposed and fixed; a regulation that fails for want of authority cannot.
The two decisions are Keysight Technologies, Inc. & Subsidiaries v. United States, No. 25-137 (Fed. Cl. July 2, 2026), and Siemens Medical Solutions USA, Inc. v. Commissioner, 167 T.C. No. 5 (July 15, 2026). Read alongside Varian Medical Systems, Inc. & Subsidiaries v. Commissioner, 163 T.C. 76 (2024), they sketch a method for attacking a Treasury regulation that has very little to do with how the rule was written and everything to do with whether Treasury could write it at all.
What Did Keysight and Siemens Decide?
Keysight is a refund case. The taxpayer paid, claimed, and sued in the Court of Federal Claims under the Tucker Act for 2020 through 2022, arguing that Treas. Reg. § 1.951A-2(c)(5) — the disqualified basis rule, which strips depreciation and amortization deductions in computing tested income under the GILTI regime — exceeded Treasury’s authority. Judge Tapp agreed: “In sum, the Court finds that the Treasury lacked the authority to promulgate Treasury Regulation § 1.951A-2(c)(5).”
Siemens is a deficiency case. There, the Commissioner had disallowed $314,992,962 of a section 245A dividends-received deduction under the Extraordinary Disposition Rules in Temp. Treas. Reg. § 1.245A-5T, producing deficiencies of $5,581,518 for fiscal 2019 and $1,452,006 for fiscal 2021. Judge Kerrigan held the taxpayer entitled to the full deduction. The reasoning fits in one sentence: “And a regulation that purports to contradict the statute can be neither necessary nor appropriate.”
A prefatory note on their posture: Keysight granted partial summary judgment; no final judgment has entered. The case is still actively litigating, and a Court of Federal Claims decision doesn’t have much binding impact for cases outside of that court. Siemens is a court-reviewed Tax Court opinion and carries precedential weight in that forum, subject to the Golsen rule.
Why Section 7805(a) No Longer Carries the Weight
For decades, Treasury’s general rulemaking authority under section 7805(a) — the power to prescribe “all needful rules and regulations” — functioned as a blanket charter. Pair it with an ambiguous Code provision, add Chevron, and the government won–a lot. Keysight takes that apart.
The court held that “the general grant of authority in Section 7805(a), standing alone, is insufficient to supply the requisite authority to the Secretary,” and then explained why the government’s contrary reading eats itself: “If Section 7805(a) is in fact some kind of Loper Bright exception, then virtually every agency regulation would be presumptively valid, rendering Loper Bright meaningless.” The sentence practitioners will end up quoting most is shorter than either: “When Chevron fell, so too did the presumption that statutory ambiguity favors the agency.”
And ambiguity has switched sides. Under the old framework, a taxpayer who established that a statute was unclear had handed Treasury the case. Under this one, ambiguity is a reason to ask where the delegation came from. The court put it bluntly: allowing the Secretary “to define terms it deems ambiguous, untethered to congressional grants of authority or the underlying statutory context, is precisely the kind of agency overreach Loper Bright was designed to foreclose.” What’s left is Skidmore, and Skidmore is a real standard rather than a courtesy — an interpretation earns deference to the extent it persuades, and this one didn’t.
Two moves in Keysight transfer directly to other regulations. The first is reading the rulemaking record against itself: Treasury’s proposed rule rested principally on the anti-abuse authority in section 951A(d)(4), and the final rule rested almost entirely on section 7805(a). The court then held that section 951A(d)(4) reaches only “this subsection” — section 951A(d) — and not the tested income rules in section 951A(c). Treasury had changed horses, and neither horse could carry the rule. And the second move is structural: because Congress expressly delegated in sections 951A(d)(4) and 951A(f)(1)(B) and said nothing about the Secretary in section 951A(c), the silence cut against authority rather than creating room for it.
There’s also a line worth remembering the next time the government argues consequences. Told at oral argument that the taxpayer’s position might cost “upwards possibly of $500 million” through 2033, the court answered that “whether the outcome of litigation financially benefits one party over the other, even when one of the parties is the Department of Treasury, is not the Court’s concern.”
Why an Express Delegation Didn’t Save the Section 245A Rules
Siemens answers the obvious follow-up question. Section 245A(g) is an express grant — Treasury may issue “such regulations or other guidance as may be necessary or appropriate.” So a specific delegation existed here, and the rule fell anyway.
The Court’s arithmetic did the work: “Section 245A allows a 100% deduction for qualifying distributions after December 31, 2017. Treasury’s adopted regulation disallows 50% of the deduction for distributions that Treasury admits satisfy the plain terms of the statute, using criteria that appear nowhere in the statute.” Treasury’s concern was real enough. Foreign income earned in the gap between the transition tax measurement date and the start of GILTI could escape both regimes. But the Court held that the concern belonged to Congress. The effective dates were clear, and Treasury had no authority to impose either regime on income subject to neither.
The pattern across the two cases is the useful part. A general delegation won’t sustain a substantive rule on its own. A specific delegation won’t sustain a rule that contradicts the statute it implements. Between them, that leaves a fairly narrow lane, and a good many regulations that have never been tested. Anyone with exposure under the section 245A participation exemption or the GILTI tested-income rules should be reading both opinions with a pencil.
How Is an Authority Challenge Different From an APA Challenge?
Freeman Law has been writing about procedural attacks on IRS guidance for years — the notice-and-comment challenge that succeeded in Mann Construction and the micro-captive litigation that followed it are examples of our long-running interest in the topic. Those cases were won on how the rule was made. The agency skipped a step.
Which means perhaps the agency can go back and take the step. That’s the built-in ceiling on a procedural win: Treasury re-proposes, opens a comment period, responds to comments, and reissues the rule in nearly the same form. Taxpayers who won on procedure have watched exactly that happen.
But an authority holding has no such off-ramp. If Congress didn’t give Treasury the power, no amount of process supplies it, and Capitol Hill has to fix it. That’s why both courts’ refusal to reach the APA questions reads as a choice rather than an accident. Keysight declined the logical-outgrowth argument because “a finding either way would not change the Court’s conclusion that the Regulation is invalid,” and Siemens said it needn’t address the APA at all, having resolved the case on the statute. Same result on paper. Very different lasting impact.
How Do You Preserve the Position Before You Have a Case?
Here’s the most practical detail in either opinion. Siemens filed Form 8275-R, Regulation Disclosure Statement, with its fiscal 2019 return, setting out the facts and the legal analysis for a position contrary to the regulation. The position was preserved on the return, before the exam, before the notice of deficiency, and before anyone litigated anything.
The disclosure protects against certain accuracy-related penalties, and it frames the issue on the taxpayer’s terms from the outset.
And the companion move is the refund side. A favorable decision in someone else’s case does nothing for a year that has already closed, and section 6511 doesn’t pause for good news — generally running three years from filing or two years from payment, whichever is later. Where a regulation’s validity is genuinely in play and the limitations period is running, a protective refund claim under section 6511 is the mechanism that keeps the door open. Identify the affected years now, not after an appellate decision makes the issue obvious to everyone.
Forum choice deserves a thought too, and these two cases illustrate the fork neatly. Siemens came through the deficiency door: a notice of deficiency, a timely petition, no payment required. Keysight came through the refund door: pay first, claim, then sue. Our discussion of the full-payment rule and choice of forum covers the trade-offs, and in a regulation-validity case the choice can matter more than usual, because it determines which court of appeals eventually gets the question.
The Punchline
The generation of challenges that produced Mann Construction taught taxpayers to ask whether Treasury followed the rules. Keysight and Siemens teach them to ask a harder question first — whether Congress ever handed Treasury the pen. A procedural defect buys a taxpayer time and a re-proposal. A defect in authority ends the regulation, and only Congress can revive it.
Our tax controversy practice handles regulation-validity disputes from the return position through Tax Court litigation and refund suits.
This article is for general informational purposes only and is not legal or tax advice. Reading it does not create an attorney-client relationship with Freeman Law. The law is fact-specific and subject to change, and you should consult qualified counsel about your particular situation.
