Court axes Drax reading of cap – interpreting limitations of liability

August 23, 2023
Pen and paper

A High Court case[1] has once again highlighted the importance of clear and consistent drafting in commercial contracts, particularly in the context of provisions seeking to limit the parties’ liability.  

Where language is open to conflicting readings, the court will apply the usual rules of contractual interpretation. If a party’s preferred reading of a liability clause has not been plainly and adequately expressed, then it risks being undermined by the court’s determination, even where such preferred reading appears to make contractual and commercial sense.


Drax, an energy supplier, appointed Wipro to provide software services under a master services agreement dated 20 January 2017. Drax terminated the agreement early for what it claimed were repudiatory breaches by Wipro, and issued proceedings.  

Drax issued a claim against Wipro, seeking total damages of £31 million for quality issues, delay, issues arising from the termination, and misrepresentation, each comprising several specific causes of action and separate losses (except misrepresentation, which overlapped with the others).


One of the preliminary issues to be determined at trial (i.e. ahead of the main trial, which is due to take place next year) was whether the agreement imposed a single, aggregate cap on Wipro’s liability, or multiple caps with a separate financial limit applying to each of Drax’s claims.

The clause in question (clause 33.2 of the agreement) read as follows:

“Subject to clauses 33.1, 33.3... the Supplier's total liability to the Customer... arising out of or in connection with this Agreement... shall be limited to an amount equivalent to 150% of the Charges paid or payable in the preceding twelve months from the date the claim first arose. If the claim arises in the first Contract Year then the amount shall be calculated as 150% of an estimate of the Charges paid and payable for a full twelve months.”

Wipro read this as a single cap for all claims. Each of Drax’s claims had arisen in the first contract year, in which 150% of the charges were £11.5m, so Wipro contended that this should be its maximum liability. Drax, on the other hand, argued that clause 33.2 imposed a separate liability cap on each of its claims, meaning that Wipro’s maximum possible liability would be £23m.


In giving judgment, Mr Justice Waksman referred to Wood v Capita[2] and earlier cases setting out well-established principles of contractual interpretation. He also referred to Triple Point Technology v PTT[3] (the leading decision on interpretation of clauses that exclude or limit liability) and Royal Devon and Exeter NHS Trust v ATOS[4] (a case on which we have commented previously).

Turning to the language of clause 33.2 itself and applying the principles from those cases, the judge noted that the wording up to “limited to” strongly suggested that the cap should apply to all claims, with the reference to “total liability” and absence of the words “for each claim” (or similar). Had there followed a specific sum, rather than a formula, the language would clearly indicate a single cap. 

The phrase “the date the claim first arose” also pointed towards a single cap: while it would have been better to state “first claim” expressly, the fact that it was “the claim” rather than “each claim” was helpful to Wipro, as was the fact that the same wording was used in the neighbouring clause 33.3. It was common ground that clause 33.3 imposed a single cap for all claims relating to breaches of the data-protection clause of the agreement, meaning that reference in that clause to the date when “the claim first arose” had to mean when the first of the claims arose, even though that was not explicitly stated. So the same expression in clause 33.2 had to be interpreted in the same way, and the judge placed considerable weight on the use of the same words in the two clauses.

On the other hand, clause 33.3 included the words “aggregate” and “for all and any claims”, both of which expressions were absent from clause 33.2. That could suggest that, absent those words, it was not entirely clear that clause 33.2 imposed a single cap rather than multiple caps. Yet, while acknowledging the point, the judge held that this was not a significant indicator that clause 33.2 did not impose a single cap.

He also referred to clause 33.1 of the agreement, which set out a separate liability cap for loss of or damage to tangible property. Here, the parties had demonstrated they were quite capable of using explicit language like “per event”. They could have done this at clause 33.2, but they did not. Again, comparison with a neighbouring limitation of liability indicated that clause 33.2 should be read as providing for a single cap.

The judge did give consideration to the fact that clause 33.2 was “subject to” clause 33.3. If clause 33.2 were read as a single cap, arguably it would not make sense for it to be “subject to” a provision allowing a much greater cap for only one group of claims. If clause 33.2 were instead read as a separate cap for each claim, and assuming that a data protection claim would prima facie fall within clause 33.2, clause 33.3 could reduce what was otherwise recoverable under 33.2.

Yet the similar language in the two clauses around when the claim “first arose” resulted in linguistic tension if clause 33.2 were read as providing for multiple caps, and this was heightened if clause 33.3 was simply a proviso to clause 33.2. As such, the judge held that the correct interpretation of the language in and around clause 33.2 was that there was a single cap, despite some “linguistic quirks”. Drax’s claimed interpretation was at least possible, but in the judge’s view it was wrong.

He took into account Drax’s commercial argument that the agreement was intended to be long-term and could potentially cover multiple different projects in the future of varying values, and as such there was little business sense in the parties’ being “stuck with” a gradually reducing single cap based on charges made early on in the agreement term (if the first claim arose then). While this was theoretically true, the judge considered it was unrealistic: if there were severe issues with the project early on in the agreement term, it was highly unlikely that Drax would choose to commission large amounts of further work under the agreement, and far likelier that they would exercise their termination rights (as, indeed, they had done). That practical reality outweighed any theoretical commercial position.


While commercial and common-sense interpretations were taken into consideration in this judgment, choice of language was central – and, in particular, consistency in choice of language. It almost goes without saying, then, that the utmost care should be taken when drafting limitations of liability, and especially where a provision has multiple sub-clauses.

If extra, bespoke limbs are added to a standard-form liability clause, it is important to ensure that the addition harmonises with the existing drafting. Special care is needed when linking a limit on liability to a numerator of charges, rather than a fixed sum: the way in which the formula is expressed may affect potential interpretations of the provision. And as demonstrated in this case, the possibility of conflicting interpretations of a limitation of liability can complicate any future contractual dispute – and can have far-reaching commercial consequences.

[1] Drax Energy Solutions Limited (formerly Haven Power Limited) v Wipro Limited [2023] EWHC 1342 (TCC).

[2] Wood v Capita [2017] UKSC 24.

[3] Triple Point Technology v PTT [2021] AC 1148.

[4] Royal Devon and Exeter NHS Trust v ATOS [2017] EWCA Civ 2196.

Helena FranklinHelena Franklin
Helena Franklin
Helena Franklin

News & Insights