R&D Expensing Is Back: Deduct Now or Amortize Under Section 174A? A Side-by-Side Comparison for Your 2025 Return

A software company spends $400,000 a year building its product. From 2022 through 2024, the tax code forced it to capitalize those costs and deduct them over five years — paying tax on profit that existed only on paper. Many startups wrote real checks to the IRS in years they lost money. That era is over: under new Section 174A, domestic R&D expensing is back for tax years beginning after December 31, 2024 — and the IRS has now confirmed the mechanics in its updated guidance.

But "back" doesn't mean "automatic," and it doesn't mean one-size-fits-all. The law gives you a choice: deduct domestic research costs immediately, or elect to amortize them over at least 60 months. It also gives you a separate, expiring choice about the 2022–2024 costs still sitting on your balance sheet. Which combination is right depends on your numbers — and for calendar-year businesses on extension, the decision lands on the return due September 15 or October 15, 2026.

Having spent years as a CFO before moving into tax practice, I read this as a modeling exercise, not a paperwork question. Let's compare the options the way you would in a board deck: side by side, with dollars.

What Changed: Section 174A in One Minute

The One Big Beautiful Bill Act created Section 174A, which permanently restores immediate R&D expensing for domestic research and experimental costs — software development explicitly included. Three rules frame every decision:

1. Domestic R&E paid or incurred in tax years beginning after December 31, 2024: deduct currently, or elect to capitalize and amortize over 60+ months.

2. Foreign R&E — work performed outside the U.S. — still must be amortized over 15 years. Nothing improved here.

3. Old capitalized costs from 2022–2024 don't have to keep dripping out over their original schedule: you may accelerate the remaining balance into 2025, or split it evenly across 2025 and 2026.

One door has already closed: eligible small businesses (under ~$31 million average gross receipts) could retroactively amend 2022–2024 returns and claim refunds, but that window generally ended July 6, 2026. If you missed it, the catch-up on the 2025 return is now your only route to the stranded deductions — which makes this filing season the decision point.

Comparison 1: Deduct Now vs. Amortize — 2025 Costs Onward

	Deduct immediately	Elect 60-month amortization
Cash flow	Full tax benefit this year	Benefit spread over 5+ years
Best for	Profitable companies; owners needing cash	Loss-year startups with expiring attributes; companies expecting much higher rates/profits later
Taxable income	Lowest now	Higher now, lower later
Reversibility	Default — no election needed	Election binds you; changing back requires IRS consent

The base case is simple. A profitable S-corp agency spends $300,000 on domestic development in 2025. Deducting it immediately saves roughly $111,000 in federal tax at a 37% owner bracket — this year, in cash. Amortizing the same costs yields only about $30,000 of first-year deductions (60-month schedule, mid-year start), pushing ~$100,000 of tax a year into the future. For most profitable businesses, immediate R&D expensing wins, full stop: a dollar of tax deferred is a dollar you can reinvest.

So who would ever elect amortization? Companies where a bigger deduction today is wasted or even harmful. A pre-revenue startup already running losses gets nothing extra from a larger 2025 loss today — NOL carryforwards only offset 80% of future income, and if a funding round triggers an ownership change, Section 382 can throttle how fast those NOLs can ever be used. Spreading deductions into the years when revenue actually arrives can deliver more after-tax value. The same logic applies if you expect to graduate into materially higher brackets, or where a state that hasn't conformed to 174A creates a two-sets-of-books problem worth simplifying.

The honest summary: immediate expensing is right for perhaps four out of five businesses — but the fifth one, usually a venture-backed startup, can leave six figures on the table by defaulting.

Comparison 2: The Stranded 2022–2024 Costs — One Year or Two?

If you capitalized R&E in 2022–2024, the undeducted balance is waiting. Take that SaaS company with $400,000 of annual domestic R&E: after the old five-year schedule ran through 2024, roughly $780,000 of its $1.2 million remains unamortized. The catch-up election (made as an accounting method change with the 2025 return) offers two paths:

	All in 2025	Split 2025 / 2026
2025 deduction	$780,000	$390,000
Best for	Strongly profitable 2025; owners wanting maximum refund now	Smoothing income across two years; avoiding a wasted deduction in a weak 2025
Risk	Deduction may exceed income → trapped in NOL limits	Betting 2026 stays profitable

If 2025 was a strong year, the one-shot deduction can wipe out most of the tax bill — at a 21% corporate rate, $780,000 of catch-up is worth about $164,000. If 2025 profit is thin, deducting more than you earn just converts cash-value deductions into NOL carryforwards; the two-year split keeps both years' deductions working at full rates. This is a fifteen-minute projection exercise that regularly swings tens of thousands of dollars — run it before you file, not after. If your books can't tell you what you actually spent on development by year and by country, fix that first — I've written about why bookkeeping is the non-optional foundation.

Comparison 3: The Hidden One — U.S. vs. Offshore Development Teams

Here is the angle most coverage skips, and the one I flag first for founders with distributed teams. Section 174A restored expensing only for domestic research. Foreign R&E still amortizes over 15 years — with only ~3% deductible in year one.

Compare $200,000 of developer cost in 2025:

• U.S.-based team: $200,000 deducted in 2025. Tax benefit at 21%: $42,000 now.

• Offshore team (Ukraine, India, Poland — anywhere abroad): roughly $6,700 deductible in 2025. Tax benefit now: ~$1,400. The rest arrives in slices through 2040.

Offshore developers may still cost less in salary — but the after-tax gap is now much narrower than the invoice suggests, and for U.S. companies owned by foreign founders the structure questions stack on top of this (if that's you, see my guide to Form 5472 and the $25,000 penalty trap for foreign-owned U.S. LLCs). Every offshoring decision made on 2021-era math deserves a fresh model.

Don't Forget the R&D Credit — It Stacks

Expensing and the Section 41 research credit are not either/or. You can deduct costs and claim the credit; Section 280C simply requires reducing the deduction by the credit amount or electing a reduced credit instead. For a company with $500,000 of qualified research and a ~$50,000 credit, the combination — full expensing plus the credit — routinely beats either benefit alone. Startups can even apply up to $500,000 of the credit against payroll taxes, turning research spend into cash relief before profitability. And once the deductions land, remember the follow-on effects: a leaner taxable income changes your QBI deduction math for pass-throughs, while a C corporation piling up refunded cash should keep an eye on the accumulated earnings tax.

What to Do Before Your Extended Deadline

Four steps, in order: pull the fixed-asset or tax-return schedule showing your unamortized 2022–2024 R&E balance; project 2025 and 2026 taxable income under the one-year and two-year catch-up scenarios; split your 2025 research spend into domestic and foreign buckets (the answer changes the deduction by a factor of thirty); and decide affirmatively — the default treatment is rarely the optimized one. You can verify what the IRS has on file for your business, including prior-year transcripts, through your IRS Business Tax Account.

Where it gets complicated enough to need help: mixed domestic-foreign teams, pending funding rounds with Section 382 exposure, partnerships where excess deductions can strand at the partner level, and any situation where the 280C credit election interacts with prior-year choices. That's exactly the modeling we do at TaxesZenPro — a two-scenario projection usually takes less time than one wrong filing takes to unwind.

Official guidance: the IRS's Rev. Proc. 2025-28 and the IRS newsroom update on OBBBA provisions.