Indemnities and the Tax Covenant

12.1The problem warranties can't solve

In Chapter 11 you met the warranty: a contractual statement of fact about the target which, if untrue, gives the buyer a claim in damages. You also met its built-in escape hatch — disclosure. Under clause 8.3 of the Sunrise SPA, the warranties are qualified by everything Disclosed. Once the seller has fairly disclosed a problem, the warranty is read subject to it, so the statement is no longer untrue and a warranty claim about that problem is dead on arrival.

That is exactly how the system is meant to work: disclosure moves the risk of a known problem back onto the buyer, who is expected to price it or walk away. But now put yourself in the buyer's chair. Due diligence has surfaced a genuine skeleton — one that arose entirely on the seller's watch, will cost real money to fix, and cannot be priced with any confidence. The seller has disclosed it, so the warranties won't bite. Does the buyer just swallow it?

Commercially, usually not. The buyer's position is simple: this is your historic problem; I am not paying for it. The tool that gives effect to that position is the indemnity: a freestanding contractual promise by the seller to make a specific loss good if it materialises. The promise pays euro for euro, whether or not the problem was disclosed, and without the buyer having to prove the shares were worth less than it paid.

12.2Indemnity vs warranty — what actually changes

The two protections look similar on the page — both are seller promises tucked into the SPA — but they behave very differently when a claim is brought:

Juniors are sometimes taught the folklore version of that last bullet. An indemnity is still a contract term, and English courts construe it like any other: the words used decide what is covered. Wide causation language such as “arising out of or in connection with” stretches the net. But arguments about whether a particular cost really falls within the indemnity, or was inflated by the buyer's own conduct, do not vanish. They just become questions of construction rather than of the law of damages. The drafting does the work, which is why buyers draft indemnities widely and sellers fight over every phrase.

One diagram to hold on to — which tool answers which kind of risk. The fourth box, the tax covenant, gets the second half of this chapter:

12.3Where indemnities come from: due diligence

Indemnities are not boilerplate. Almost every specific indemnity in a real SPA is the fossil of a due diligence finding — a problem someone unearthed in the data room (Chapter 7) that the parties then had to allocate. Project Sunrise has a textbook example.

Neither of the usual tools worked. A price reduction failed because the risk was unquantifiable: Atlas's opening guess and Meridian's opening guess were millions apart, and neither could prove the other wrong. A warranty failed because the matter was disclosed. The inspection letter is Disclosed, so clause 8.3 qualifies the compliance warranties in paragraph 5 of Schedule 3. Paragraph 5.3 even carves the Legacy AML Matter out expressly. So no warranty claim could ever run. And the risk sat squarely in the seller's period of ownership — the onboarding happened years before Atlas appeared.

The negotiated answer was a specific indemnity: Meridian keeps the risk, Atlas gets paid if it crystallises, and neither side has to guess the number today. The seller accepts this reluctantly but rationally — the alternative was a price chip based on the buyer's worst-case number, or no deal.

12.4Dissecting clause 10.2, phrase by phrase

Here is the indemnity the parties landed on. It is four lines long and every phrase earned its place in a negotiation:

“shall indemnify … on demand.” The obligation to pay is triggered by the buyer asking, not by the buyer winning a lawsuit. If the CBI fines Solaris €3m, Atlas sends a demand and the money is due — there is no preliminary round of proving breach, loss and causation as there would be in a damages claim. If the seller refuses, the buyer still has to sue, but it sues for a promised sum, which is a far shorter and safer road. For a buyer, “on demand” is cash-flow protection.

“the Buyer, for itself and on behalf of each Group Company.” This is subtler and more important than it looks. When the CBI fines Solaris, the loss lands in Solaris — the target — not in Atlas. An indemnity drafted only in the buyer's favour invites the seller to argue that the buyer itself has suffered no loss (or only some smaller, indirect one). Extending the indemnity to losses of each Group Company shuts that argument down: the parties have agreed whose loss counts, in advance.

“all Losses arising out of or in connection with.” Two pieces of width here. The defined term Losses sweeps in fines, penalties, costs and professional fees — items a damages claim might struggle to reach. And “arising out of or in connection with” is deliberately loose causation language: the loss need only be connected to the Legacy AML Matter, not directly caused by it in the strict damages sense.

The enumerated heads. Sub-paragraphs (a) to (c) — CBI fines, mandated remediation costs, professional fees — are introduced by “including”, which under clause 1.2(d) does not limit the general words. They are there for certainty: the three costs everyone expects are named, so no one can later argue they fall outside the clause.

Now the guardrails, because Meridian did not sign a blank cheque:

• A cap of €8,000,000 for all claims under clause 10.2 (clause 9.4(b)) — high enough to cover realistic outcomes, low enough that Meridian can quantify its worst day.
• A four-year notice window (clause 9.1(c)) — longer than the 18 months for business warranties, because regulatory processes grind slowly, but not open-ended.
• No de minimis, no basket, no buyer-knowledge exclusion (clause 10.3) — obviously, since the whole clause exists because the buyer knows.
• Conduct obligations on the buyer (clause 10.4): the remediation programme must be run diligently and cost-effectively, and the seller must be kept reasonably informed of material CBI correspondence.

12.5Payments as a reduction of the price

One short clause deserves a mention before we reach tax, because it exists for tax. Clause 3.4 says that any payment by the Seller under the locked box, warranty or clause 10 provisions “shall, so far as possible, be treated as a reduction of the Consideration”. The logic: if an indemnity payment were simply income arriving in the buyer's hands, there is a risk the tax authorities would want a slice of it — leaving the buyer under-compensated. Recharacterising the payment as an adjustment to the price the buyer paid for the shares avoids that: the buyer is treated as having simply paid less, which is not a taxable receipt. It is a two-line clause doing quiet, valuable work, and you will see it (or a longer version) in almost every SPA.

12.6The tax covenant — why tax gets a regime of its own

If indemnities are targeted strikes, the tax covenant is a standing army. It is an indemnity-style promise under which the seller covenants to pay the buyer an amount equal to the target's tax liabilities for the seller's period of ownership. Historically the covenant was a separate “tax deed”; now it is usually a clause or schedule. Practically every private share deal has one.

Why does tax, alone among liabilities, get blanket euro-for-euro cover rather than mere warranties? Three features make it special. Tax liabilities are long-tailed: a tax authority can open an enquiry and raise an assessment years after the period in question. They arise without fault: a filing position everyone thought safe can simply turn out to be wrong. And they are assessed later, so the accounts drawn up at the deal date can only ever estimate them. Buyers therefore refuse to take historic tax risk on trust, and sellers — who controlled the company when the profits were earned — find it hard to argue they should not stand behind it.

Walk it through. The core promise covers Tax on profits earned, or events occurring, on or before the Locked Box Date — 31 December 2025, the line in the sand from Chapter 9. Everything before the line is the seller's; everything after is the buyer's. Then four exceptions trim it to what is fair. Exception (a): if the Accounts already provided for the liability, the buyer priced the deal knowing about it — no double recovery. Exception (b): ordinary-course tax arising after the box date is just the cost of running the business the buyer now owns. Exception (c): tax on Permitted Leakage was agreed as part of the leakage package. Exception (d): if a government changes the law after signing, that is nobody's historic sin, so the buyer bears it.

How does the covenant sit with the Tax Warranties in paragraph 12 of Schedule 3? They do different jobs, and you need both. The warranties (returns filed, tax paid, no disputes, residence where expected) exist mainly to force information: the seller must disclose against them, so any live tax problems surface before signing. The covenant exists to move money: whatever pre-box tax emerges, disclosed or not, the seller pays. Both share the seven-year time limit in clause 9.1(b) — deliberately matched to the long horizons over which tax authorities can raise assessments, and far longer than the 18 months given to the business warranties.

Finally, notice how neatly the covenant dovetails with the locked box. If Solaris paid Meridian a dividend in February 2026, the dividend is Leakage — and so is any tax the company suffers on it, under limb (h) of the definition. If instead the Irish Revenue assesses Solaris for underpaid tax on its 2024 profits, that is pre-box tax and the covenant pays. Between leakage and the tax covenant, the seller stands behind the whole pre-completion tax picture — which is exactly what a locked box, with its fixed price and no completion accounts, requires.

12.7The wider indemnity menu — and the pushback

Once buyers taste euro-for-euro cover they tend to ask for more of it. Common requests: an indemnity for ongoing litigation found in diligence; for a data breach or security incident; for a problematic customer contract; for underpaid employment taxes on contractors. The seller's stock responses are equally familiar: “the price already reflects that”, or “you have a warranty; if it's not disclosed, you're covered — warranty or nothing”. Every specific indemnity is a small defeat for the seller's clean-exit ambitions, so expect each one to be fought.

The negotiation rarely ends at yes or no, because indemnities sit on a spectrum of alternatives:

Notice the pattern: the choice of tool tracks two questions. Can we price it? If yes, chip the price. Will the seller be there to pay? If in doubt, take security. The indemnity is the answer when the number is unknowable but the counterparty is solid. That is why it fitted the Legacy AML Matter, with a VC-backed English holdco on the hook. It is also why the negotiation of who-bears-what is really a negotiation about confidence (Chapter 14).

12.8Conduct of claims — who holds the steering wheel

When the loss comes from a third party — a claimant suing the target, or a regulator fining it — a second question follows the money: who controls the defence? The economics point one way. Whoever ultimately pays wants the steering wheel: the right to run the defence, choose the lawyers, and decide whether to settle. A buyer who is fully indemnified has little reason to fight hard; a seller footing the bill has every reason.

For warranty claims arising from third-party claims, clause 9.7 strikes the usual compromise. The buyer must notify the seller promptly, keep it informed, consider its reasonable representations, and not settle without the seller's consent (not to be unreasonably withheld). The clause also reminds you that the buyer's common law duty to mitigate survives.

But look what happens when the third party is a regulator. Solaris cannot “defend” a CBI enforcement process the way it would defend a commercial lawsuit. The CBI is its supervisor: Solaris must keep its licence, its relationship and its credibility long after this matter closes. A seller with full conduct rights might be tempted to fight the regulator to the last euro — scorched earth that the seller walks away from and the buyer lives with. That is precisely why clause 10.4 is drafted as consultation — diligent, cost-effective conduct by the buyer, with information flowing to the seller — rather than seller control.

12.9The Legacy AML Matter, with numbers

Run the film forward. In 2028 the CBI concludes its process: a fine of €3,000,000, and the remediation programme has cost Solaris €1,500,000. Atlas notifies a claim under clause 10.2 — well inside the four-year window — and demands €4,500,000. The de minimis (€50,000) and the basket (€420,000) are disapplied by clause 10.3, so they take nothing off the top; the total sits comfortably under the €8m cap; and the money is due on demand. Now the counterfactual: no indemnity, warranties only. The matter was Disclosed, so the compliance warranty is qualified and the claim fails — not reduced, but gone. Recovery: nil. One clause, €4.5m of difference.

That is the whole chapter in one table: for a known risk, the warranty regime is designed to fail, and the indemnity is designed to pay. Next we leave the risk-allocation machinery and follow the deal into the gap between signing and completion — Chapter 13.