Rule 46(8) of the new Income-tax Rules, 2026, in force since the first of April, asks something that sounds modest — back up your accounts every day on a machine physically located in India.

The simplicity is the problem. The wording appears to fit an era of one machine, in one place, backed up by someone each evening, a setup most businesses abandoned years ago.

The gap between the data the rule imagines and the data businesses keep is where the trouble starts.

Where The Words Break

The rule says the backup must be updated "on a daily basis", and then says nothing else. It does not say what daily means for a business that never closes, the software firm billing across three time zones, the exchange, the payments company, the delivery platform running around the clock.

Advisers have filled the silence, telling such firms to fix an end-of-day moment and document it, which is sensible and is also the adviser's decision of a moment-of-day rule that is not explicitly provided for under the Income Tax Rules.

"Daily" also doesn't say what a backup must contain. A full copy every night, or just that day's changes? Overwrite last night's copy with tonight's, and you have technically taken your daily backup while erasing what the books looked like a week ago. Which is what the rule is meant to protect in the first place.

Nor does "daily" say what a backup must contain. A fresh copy each night, or only the day's changes? Save over last night's copy with tonight's, and you have, read literally, made your daily backup while wiping out any view of what the books showed a week ago, the very thing the rule means to protect.

Then the next phrase: a server, physically located in India. A modern accounting system does not sit on a server. It sits spread across many machines at once, shuffled between locations as routine engineering, precisely so no single machine matters and nothing is lost if one fails.

"Physically located" assumes a box you can point at, and in a cloud setup, there is no such box, by design. The rule means one of two things and will not say which: an Indian branch of a big cloud service, sensible but not what the words say, or a single physical box, which is what the words say but which almost no serious business uses to run its accounts.

By literal reading, the Indian backup can be a second copy on the very same laptop the accounting software runs on. Same machine, in India, daily, the words satisfied and every purpose behind them are defeated.

The Shifty Nature

A copy sharing a hard drive dies in the same theft, the same spilt coffee, the same dead disk. A rule meant to guarantee that records survive, therefore can be met by the one arrangement that guarantees they do not. That’s because the words fix a place and a frequency; they say nothing about keeping the copy separate, intact, or recoverable, which are the only things that make a backup worth having.

The audit form makes the muddle concrete. The new report, which from this year replaces the three old forms, asks the auditor to write down, in the words of the tax department's own guidance, the internet address of the machine holding your accounts, the country it sits in, and the address of your Indian backup. In a cloud setup, that address is not fixed; the one your auditor records at filing time is not the one serving your data next Tuesday.

A fixed address comes only with an old in-house kit or one lone rented server, the very things a careful engineer is moving away from. The form rewards the worst choice: the fixed box fills it in cleanly and looks compliant, while a sturdier, spread-out setup cannot give one answer and ends up looking shifty.

The rule never asks the obvious question: what happens when a backup fails? Backups fail. There is nothing about checking that the backup opens, testing that it works, or how long a gap is allowed before you are in breach. If your one backup corrupts on a Tuesday and nobody notices until Thursday, were you in breach on Wednesday?

Technology doesn’t run on a never-miss-a-night standard, so the rule will be enforced at the officer's discretion. A standard nobody can perfectly meet, enforced at discretion, is a lever. It hands an officer who wants to throw out a set of accounts a fault he can nearly always find.

The Smallest Firm In The Largest Queue

Take the example of a consultancy that raises four or five invoices a month, keeps a clean ledger, files its income tax, GST and TDS without a hitch, and is still audited, because it is a company, or crossed a turnover line, or declares lower profits than previously estimated.

Its whole financial year could fit on one spreadsheet, but the rule does not care. It applies to every person who keeps accounts on a computer, from a one-person shop to a listed giant. A firm with five transactions a month and one with five million faces the same demands, the sign of a rule that measures the wrong thing.

The spreadsheet itself sits in a grey area. The common reading among tax professionals is that a loose list of figures is not really "accounts" unless it is built to show the full position day by day, with the change history intact. That is not possible with a spreadsheet.

This is not an abstract unfairness. Micro, small and medium enterprises account for around thirty per cent of GDP and close to half of exports, and with over thirty-two crore people, they are the country's largest employer after farming.

A listed company has an IT department to reconfigure its cloud and a budget to absorb whatever the auditor demands.

A two-person trading firm running Tally on one computer, with a part-time accountant who comes in on Saturdays, has none of that. For it, the rule is a new fixed cost, a new vendor to find, a new thing to get wrong, bolted on top of the GST returns, TDS and annual audit it already strains to meet.

Why The Noise Now

If the rule has been the law since April, why is every accountant and software firm upset about it now?

That’s because the accounts most businesses are filing this year are the previous fiscal year's. Since those accounts are judged under the old law, the new rule does not touch them. Its first test comes a year from now.

The year the new rule first covers is the one currently running, the one that began on 1 April 2026. A daily backup cannot be created once a day passes, nor a year of edit history invented once the year closes on 31 March 2027. Set the wiring up wrong from day one, and you spend twelve months building records that will not survive next year's audit, with no way back.

The second reason is plainer. The people sounding the alarm often sell the cure: the Indian backup servers, the audit-ready software, the advisory hours. A new cap limiting each accountant to sixty tax audits a year per firm partner, also from April, only sharpens the rush to sign clients early.

The effect is regressive by design: the same obligation weighs almost nothing on the giant and a great deal on the small, and it lands hardest on the firms the wider economy most depends on. A rule that nudges the smallest businesses towards "too much trouble to stay formal" works against the very formalisation the tax system has spent a decade in claiming.

None of this needed to happen because the fix is not complicated. A proportionate rule would set a turnover floor below which the daily-backup machinery does not apply, as the law already exempts the smallest firms from keeping formal books at all. It would describe the outcome it wants rather than the hardware, a recoverable copy held away from the original and reachable from India, which a cloud backup satisfies cleanly, instead of a "server physically located in India" that sends everyone hunting for a box. And it would say plainly that an Indian region of a recognised cloud service counts.

None of these is exotic; India's own rules use thresholds, outcome-based drafting and cloud-aware language elsewhere. The drafters reached for none of them here.

Who gains? India is in the middle of an enormous data-centre boom, led by Telangana, Maharashtra and a handful of other states, with the firms collecting the investment forming a small club. A rule forcing every business on a foreign-only cloud to acquire an Indian base creates captive demand for that capacity, just as the industry races to fill it.

It might be unfair, though, to dismiss Rule 46(8) as a money-funnel dressed up as a safeguard. The boom is driven overwhelmingly by artificial intelligence and cloud demand; a reading should not be made off an outcome. But India's localisation rules have a habit of arriving in step with the home-grown industry that profits from them.

The weight of this rule falls hardest on the small firms, the country that most needs to grow, while the benefit flows to the owners of buildings full of servers on Indian soil.

The alignment exists, whatever the merit of any conspiracy theory. It needs no one to have schemed, but it’s a rule written in the wrong decade, by people who didn’t ask what happens when the disk fails, or noticed that the smallest firm was made to stand in the same queue as the largest.