Burn, Not Balance Sheet: What a Carve-Out Really Does for the CFO

Investment committee, Tuesday afternoon. The head of innovation presents a carve-out: a technology from the company's own lab, working, but with no home in the core business. She has saved her strongest argument for last: “And we avoid the write-off.”

The CFO glances up. “What write-off? That's internal development. We expensed it long ago.”

Silence in the room.

Come to the CFO with the wrong argument and you lose them in the first minute. Yet there is a right argument. And it's stronger.

Where the write-off argument comes from

It isn't entirely wrong. It's just true less often than people think.

Under IFRS there's a clear split. Research is expensed immediately and never creates an asset. Development, on the other hand, may be capitalized, but only if the company can demonstrate all six criteria of IAS 38: technical feasibility, intention to complete, ability to use or sell, probable future benefit, adequate resources and reliable measurement of costs. If such a project is later stopped, the impairment test under IAS 36 kicks in. Then comes the impairment, and it hits EBIT in the current year.

The HGB, the German Commercial Code, works similarly, just softer. Self-created intangible assets may be capitalized, but don't have to be (§ 248(2) HGB). And if research and development can't be reliably separated, capitalization is ruled out (§ 255(2a) HGB).

In other words: in a European group that has capitalized its development, a stopped project really can trigger a write-off. That's where the argument comes from. And there it's legitimate.

Why it almost always falls flat under US GAAP

For a company reporting under US GAAP, things look different. ASC 730 is unambiguous: research and development costs are expensed as incurred. The main exception concerns software. For a self-developed technology in the lab, that means: there's no asset on the balance sheet. So there's nothing to write off either. The costs ran through earlier P&Ls long ago.

There is one real exception. Technology that came in with an acquisition is capitalized at fair value as in-process research and development and remains an indefinite-lived intangible asset until completion or abandonment. If it's abandoned, the impairment follows. Here the write-off argument works in the US too. If, on the other hand, you bought the technology on its own and without alternative future use, you expensed it immediately.

There's nothing to gain on the tax side in the US either. For tax years beginning after December 31, 2024, domestic research expenditure is once again immediately deductible (§ 174A IRC). If a project is abandoned, there's no additional deduction (§ 174(d) IRC).

And even where the argument formally applies, a warning belongs with it: nobody can promise “no write-off”. A funded company with a valuation from a real funding round gives the impairment test a reference point. The auditors decide.

Lever 1: The burn stops

Now to the argument that really counts. A project on the shelf is rarely dead. It keeps running on a low flame: two or three people, a lab bench, materials, patent fees. Under US GAAP all of that is expense, every quarter, fully in operating income. In IFRS groups it's no different once the project no longer meets the capitalization criteria.

A carve-out ends this run. The team moves onto the new company's payroll, and external investors fund the development from that point on. For a project that would otherwise have kept running for years, that's real, recurring relief for earnings.

A write-off is a one-time accounting event. The burn is real money, every quarter. That's exactly the lever a CFO gets immediately.

Lever 2: Three revenue streams instead of one cost center

What used to only cost money becomes three potential revenue streams: a minority stake, a royalty from the license, ideally kicking in only above a revenue threshold, and often a new customer, if the group supplies the young company. And all of that without a single additional headcount.

Lever 3: A stake without start-up losses

This is where it gets technical, and where precision pays off. Under US GAAP, a stake below 20 percent is presumed to carry no significant influence (ASC 323). Without significant influence, the stake is a simple equity investment (ASC 321). The group then takes no share of the young company's start-up losses. If there's no market price, it can carry the stake at cost less impairment and adjusts it as soon as there's an observable price, such as the next priced funding round. A good round can then show up as a gain in the P&L. A bad one as a loss.

Two caveats belong here. First, the 20 percent threshold is only a presumption. A board seat or strong say-so rights can establish significant influence below it, too. Second, on May 13, 2026, the FASB tentatively decided by a 5 to 2 vote to remove exactly this presumption for stakes below 20 percent. That isn't final yet; an exposure draft for comment is still pending.

The consequence is already clear, though: the accounting is decided less by the size of the stake than by the rights attached to it. If you want to keep losses out of your P&L, keep your say lean. Observer status instead of a board seat isn't modesty then. It's accounting policy.

Lever 4: No consolidation through the back door

The second stumbling block is consolidation. Under ASC 810, a company must consolidate an entity if it has two things at once: the power to direct the activities that most significantly drive its economic performance, and the obligation to absorb losses or the right to receive benefits that could be significant. A group that provides most of the money and secures strong control rights can end up exactly there. Then the young company's losses sit fully in its numbers.

The structure that avoids this isn't a trick. It's simple: minority stake, no obligation to fund further, external investors lead the rounds, at most an observer seat. That's the US translation of “off the balance sheet”.

Lever 5: A capped risk

And then the point every CFO likes. A carve-out with a fixed budget, released in stages and tied to decisions, has a known maximum. Every additional euro of risk sits with external investors. The alternative is to keep funding the project internally for another two years, with no end in sight.

And in Germany?

For German corporations there's a tax argument that often gets mentioned, and it holds: gains from a later sale of the shares are disregarded under § 8b(2) KStG (German Corporate Income Tax Act), with 5 percent treated as non-deductible business expenses. Effectively, 95 percent is tax-free.

But three traps come with it. Dividends from a stake below 10 percent are fully taxable (§ 8b(4) KStG). Losses on the shares aren't tax-deductible (§ 8b(3) KStG). And if you contribute your IP in exchange for shares, that's a swap. The fair market value of the asset given up then applies (§ 6(6) EStG, German Income Tax Act), and since self-created IP usually sits on the tax books at zero, a taxable gain can arise. One more reason why a license to the new company is often the better structure than a transfer. Which structure fits the individual case is for the tax advisor to decide.

The argument in two sentences

For a US group with in-house development, the right argument sounds like this: “This project shows up as expense in your P&L every quarter and has no path to revenue inside your company. From now on, external investors fund the development, your team builds it in a company of its own, and you keep a stake, a royalty and a customer, with no losses on your books and no consolidation.”

For a European group with capitalized development, or for acquired technology, one sentence gets added: “And a funded company with a real valuation gives your auditors a reference point, instead of you having to justify abandoning the project.” You can't promise the outcome.

A write-off happens once. The burn happens every quarter. Stop the burn and you have the CFO on your side.

This article is not tax or accounting advice. It does not replace consultation with auditors and tax advisors.

Sources

• KPMG, “Handbook: Research and development” (2025): wording of ASC 730-10-25-1 (expensed as incurred), software exception ASC 350-40/985-20, ASC 730-10-25-2(c) (acquired intangibles without alternative future use), acquired in-process R&D from business combinations under ASC 805 as an indefinite-lived intangible asset until completion or abandonment — kpmg.com
• IASC Foundation, “Technical Summary IAS 38 Intangible Assets”: research as expense, six criteria for capitalizing development, impairment under IAS 36 — Technical Summary
• Handelsgesetzbuch (German Commercial Code), § 248(2) (option to capitalize self-created intangible assets) — gesetze-im-internet.de
• Handelsgesetzbuch (German Commercial Code), § 255(2a) (distinction between research and development, prohibition on capitalization where they cannot be separated) — gesetze-im-internet.de
• FASB Staff, “PCC Topic 10: Equity Method of Accounting” (June 2, 2026): wording of the presumptions under ASC 323-10-15-8 (significant influence from 20%, none below, each rebuttable) — fasb.org
• FASB, “Minutes: Equity Method of Accounting: Targeted Improvements” (board meeting May 13, 2026): tentative decision (5:2) to remove the presumption of no significant influence below 20%; all decisions tentative, exposure draft with a 75-day comment period pending — fasb.org
• PwC Viewpoint, “Loans and investments”, ch. 2.3: measurement alternative under ASC 321-10-35-2 (cost less impairment, adjusted for observable price changes) — viewpoint.pwc.com
• Deloitte, “Roadmap: Consolidation”, ch. 7.1: primary beneficiary criteria under ASC 810-10-25-38A (power to direct and obligation to absorb losses or right to benefits) — dart.deloitte.com
• 26 U.S. Code § 174A (immediate deduction of domestic research expenditures for tax years from 2025, Pub. L. 119-21) — law.cornell.edu
• 26 U.S. Code § 174 (foreign research over 15 years, no additional deduction on abandonment under § 174(d)) — law.cornell.edu
• Körperschaftsteuergesetz (German Corporate Income Tax Act), § 8b(2)–(4) (capital gains disregarded, 5% non-deductible business expenses, losses not deductible, dividends taxable for stakes below 10%) — gesetze-im-internet.de
• Einkommensteuergesetz (German Income Tax Act), § 6(6) (swap: valuation at the fair market value of the asset given up) — gesetze-im-internet.de