CA K Sanjay BhargavChartered Accountant
Open menu
← All articles

The new safe harbour rules: what changed, and whether to take it

CA K Sanjay Bhargav, Chartered Accountant, Bengaluru

Membership No. 250054 · DISA (ICAI)

Published

Short answer: from 1 April 2026, IT and ITeS merged into one category at a unified 15.5% margin, and the eligibility threshold rose from ₹300 crore to ₹2,000 crore, with a five-year block for IT services. The margin fell and the door widened — but electing is still a decision, not a default.

What changed

BeforeFrom 1 April 2026
CategoriesIT and ITeS treated separately, with their own marginsMerged into a single Information Technology Services category
MarginCategory-specific, materially higher15.5%, unified
Transaction value threshold₹300 crore₹2,000 crore
Block periodShorterFive years for IT services
Threshold testingRecurringTested in the first year of the block only
ApprovalRules-based and automated

A separate safe harbour was also introduced for certain data centre services.

Why this is not a small change

Two of those rows change who the regime is even for.

The threshold moved by a factor of nearly seven. A large population of mid-size and large GCCs was structurally excluded from safe harbour under the ₹300 crore ceiling and never had reason to look at it again. That exclusion has gone. A conclusion reached against the old threshold carries no weight against the new one.

The block runs five years, tested once. Certainty for one year is administratively useful. Certainty for five, with the revenue test applied only at the start, is a genuinely different commercial proposition — it takes transfer pricing off the table for a planning horizon rather than a filing season.

The practical consequence: almost every GCC in Bengaluru that decided against safe harbour before April 2026 decided it on facts that no longer apply.

The part the coverage skips

A lower prescribed margin is widely reported as a saving. It is not, by itself.

Safe harbour sets a margin you must declare at or above in order to get certainty. It does not reduce your tax to that level — it sets a floor.

So:

  • If your defensible arm's length margin is above 15.5%, electing is straightforwardly attractive: you get certainty at a margin no worse than you would have argued for.
  • If your defensible margin is genuinely below 15.5%, electing means declaring more profit than a transfer pricing study would support, and paying tax on the difference. That is the price of the certainty, and for some GCCs it is a real price.

The margin coming down widens the set of GCCs for whom electing makes sense. It does not make electing correct for all of them. That is an arithmetic question on your own cost base, and it has an answer.

What you give up

The right to argue.

An election accepts the prescribed margin for the covered transactions. The comparables analysis, the disputes over what belongs in the cost base, and the appeal route all fall away with it.

That is the whole trade: certainty and no litigation, for a margin you did not negotiate. Whether it is a good trade turns on two things —

  1. how far your defensible margin sits below the prescribed one; and
  2. what your dispute history has actually cost you, in fees, in management time, and in cash locked up pending resolution.

Groups that have been through a full transfer pricing cycle to tribunal usually value the second far more highly than groups that have not.

Safe harbour, or an advance pricing agreement

Both buy certainty. They are not substitutes.

Safe harbourAdvance pricing agreement
HowElection by filing, against a prescribed marginNegotiated for your specific facts
SpeedFast, rules-basedConsiderably longer to conclude
ScopeSpecified transaction types, within thresholdsCan reach arrangements safe harbour does not
MarginPrescribedNegotiated

The 2026 changes make safe harbour viable for a much larger group, which shifts where the line between the two sits — but an APA remains the route where the arrangement is unusual, the value added in India is high, or the margin question is genuinely contested.

The decision, on your numbers

  1. Compute the actual margin the entity earns on its current cost base.
  2. Establish what margin a transfer pricing study would defend — which is not the same number.
  3. Test eligibility against the ₹2,000 crore threshold, remembering it is tested in the first year of the block.
  4. Price the difference between your defensible margin and 15.5%, over five years.
  5. Set that against what certainty is worth to you, given your dispute history.

Step 2 is the one most groups have not done recently and cannot skip, because without it the comparison has no left-hand side.

For how this sits alongside the other questions a captive faces, see GCC and captive unit taxation.

This is a working reference, not the statute. For anything you are relying on, confirm the notified rule text directly.

Frequently asked questions

What actually changed on 1 April 2026?

Three things that matter to a GCC. IT and ITeS were merged into a single Information Technology Services category carrying a unified margin of 15.5%, replacing the previous category-specific margins that ran considerably higher. The eligibility threshold for transaction value rose from ₹300 crore to ₹2,000 crore. And the block period for IT services is five years, with the revenue threshold tested only in the first year of that block rather than annually.

Does a lower safe harbour margin automatically mean lower tax?

No, and this is the point most coverage skips. Safe harbour sets a margin you must declare at or above to get certainty. If your actual arm's length margin is genuinely below 15.5%, electing means declaring more profit than a transfer pricing study would support, and paying tax on it. The margin coming down widens the set of GCCs for whom electing makes sense, but it does not make electing correct for everyone.

What do we give up by electing?

The right to argue. A safe harbour election is an acceptance of the prescribed margin, so the comparables analysis, the disputes about the cost base and the appeal route all fall away for the covered transactions. That is the entire trade: certainty and no litigation, in exchange for a margin you did not negotiate. Whether it is a good trade depends on how far your defensible margin sits below the prescribed one, and on what your dispute history has actually cost you.

We were above the old threshold, so we never looked at safe harbour. Should we now?

Yes. The threshold moved from ₹300 crore to ₹2,000 crore, which is a different order of magnitude and brings in a large population of mid-size and large GCCs that were structurally excluded before. A conclusion reached against the old threshold carries no weight against the new one, and the five-year block means the decision now buys considerably more certainty than it used to.

Is safe harbour the same as an advance pricing agreement?

No. Safe harbour is a standardised election against a prescribed margin, taken by filing rather than by negotiation, and available only for specified transaction types within thresholds. An advance pricing agreement is negotiated with the authorities for your specific facts, takes considerably longer to conclude, and can cover arrangements safe harbour does not reach. They solve the same problem — certainty — at different price points in time and effort.

Worth re-running the safe harbour decision?

Send your cost base, current markup and transaction value. Whether you are now eligible, what electing costs against your actual margin, and what the five-year block commits you to are worked out on your numbers before an election is made.

Related service: GCC & Captive Units