The Same Work for 30% Less: Anatomy of the Client Ask

Clients are asking IT services providers for the same scope of work at 25% to 30% below previous rates, on the reasoning that AI has lowered the provider's cost of delivering it. The request is usually reasonable on its own terms. What determines the outcome is whether the provider arrives with an alternative commercial construct or only with a defence of the existing one.

The ask, in the client's words

Persistent Systems chief executive Sandeep Kalra summarised the pattern in August 2026: clients want "the same work for 25% to 30% less," delivered faster and with higher productivity. Everest Group's Jimit Arora described the resulting market as one where "the odds are very much in favour of clients."

It helps to understand why the client believes this is fair rather than opportunistic. Enterprise buyers have watched their own engineering teams adopt AI tooling and have seen internal throughput change. They read the same vendor marketing about productivity gains that providers circulate. When a provider's own website claims a 30% uplift, procurement treats that as an admission, and prices accordingly.

The provider is therefore negotiating against its own sales material.

Four clause types now appearing at renewal

The ask rarely arrives as a blunt demand for a lower rate. It arrives structured, and the structure matters more than the headline number.

1. Productivity pass-through. A clause committing the provider to reduce charges by a stated percentage each year on the assumption that AI-assisted delivery improves. It looks modest at 4–5% annually and compounds into a serious margin problem across a five-year term.

A reasonable answer: accept a pass-through only where it is tied to a measured baseline the two parties agree, and pair it with a floor below which rates cannot fall regardless of measured gain.

2. AI-adjusted rate benchmarking. The client benchmarks the provider's rates against a market reference that already reflects AI-era pricing, then requires alignment. The reference set is usually opaque.

A reasonable answer: ask to see the comparator basis and insist that benchmarking compare like-for-like scope and accountability, not headline rates alone. A provider carrying delivery risk is not comparable to one supplying capacity.

3. Savings-share language. The client proposes that AI-generated savings be split, which sounds equitable and often is. The difficulty is definitional: savings against what baseline, measured by whom, over what period.

A reasonable answer: engage willingly, but make the baseline a joint artefact agreed before signature. Cognizant's February 2026 agreement with Daimler Truck, which splits AI-driven savings between vendor and client, is a working example of the structure done deliberately rather than retrofitted.

4. Term compression. Rather than cutting rates, the client shortens the contract from five years to two, preserving the option to reprice sooner.

A reasonable answer: price the shorter term accordingly. A two-year contract carries different amortisation of onboarding and transition cost than a five-year one, and that difference is legitimate to reflect.

Why holding the line usually loses

The instinctive response is to defend the rate card and wait, on the view that clients have not yet asked in most accounts and quality will tell. This is a defensible reading of an uncertain market, and it is the position most mid-market firms have taken.

It carries a specific risk. A provider who concedes nothing until the client insists ends up conceding at the client's number, at renewal, under time pressure, with no alternative construct on the table to trade against. The discount arrives anyway, and arrives as a pure margin reduction rather than as part of a restructured deal.

Firms that have moved earlier are trading the discount for something. ExlService reports outcome-based arrangements at roughly 36% of revenue with a target in the mid-40s over three to five years. Genpact, at around 24%, attributes about 300 basis points of gross-margin expansion to agentic solutions. These are not firms that avoided the price conversation. They are firms that changed what was being priced.

What to have ready before the meeting

Preparation determines the outcome more than negotiation skill does. Four things are worth assembling before a renewal conversation opens.

The last of these is where most firms are weakest. A provider who cannot state the cost of delivering a given outcome cannot know whether a proposed discount is painful or fatal, and will therefore either refuse reasonable terms or accept unreasonable ones.

The line worth drawing

There is a version of this conversation that goes badly for reasons unrelated to price. Under commercial pressure, providers sometimes accept accountability for results that depend on decisions they do not control: user adoption, scope discipline, data quality, the client's own investment choices.

A provider can commit to capability, availability, throughput and quality. What the client achieves with those is shaped by the client's own choices. Being precise about that boundary is not timidity. It is the difference between a contract that can be delivered and one that cannot.

The firms that come out of this period well will not be the ones that held the old rate longest. They will be the ones that arrived at the renewal with a different thing to sell, priced with enough confidence to defend it.