← VuduVationsThe Intelligence Bureau
VuduIntelConsulting-AI Intelligence
← The Archive
VuduIntel · The Archive
No. 10
Week of May 27, 2026
The Week in Consulting AI

New AI work booms while old contracts shrink: the renewal squeeze begins.

ISG puts numbers on it: new-scope value up 20%, renewals down 24%, and incumbency is now a disadvantage.

VuduVations Intelligence Bureau · May 27, 2026 · 4 min read
The RepricingGovernance & Authority
Share

Start with the number that settled the argument. In the first quarter of 2026, ISG’s Jones reported that new scope contract value reached $8 billion, up 20% year over year, while renewal and extension contract value fell to $3.2 billion, down 24%. He paired those figures with the line the field had been circling for six weeks: in some ways, he wrote, you could argue that incumbency has actually become a disadvantage in this market. The point is not that a competitor showed up with a better pitch. Enterprises are taking work to market and redesigning how they operate rather than renewing what they already have.

Jones is precise about the gap. Buyers want providers to lead on AI, and many are willing to redesign processes to get better outcomes, but most do not yet appear ready to govern the operating model that comes with it. The buyer is ahead of the provider on willingness. The provider is behind on governance. That single gap explains why new scope is climbing while renewals are sliding. Wipro’s Bartoletti gave the architectural reason. Agentic systems, she said, do not just process information, they act, plan, decide, and execute often faster than any human oversight mechanism can follow. This is a statement about speed, not a compliance note. The human review cycle, the audit checkpoint, the governance committee were all built for systems that waited for approval. Agents do not wait, so the oversight loop is already behind the moment it starts.

Bartoletti’s answer is what she calls Trust by Design, governance built into the system before it runs rather than reviewed after the fact. Agentic AI, she notes, produces outcomes through chains of tiny decisions that are not individually explainable in any meaningful sense, so what is needed are frameworks that judge systems by their behavior over time rather than decision by decision. Three separate loops are failing under that pressure. The procurement loop: Hackett’s Sawchuk found that fewer than half of chief procurement officers feel confident they can monitor or control agentic AI, because the qualification cycle is structurally slower than agent deployment. The production loop: Cognizant’s Kumar named it from the delivery side, saying the more the technology advances, the wider the gap to enterprise value, and that bridge will come from firms like his. He added that the AI-native deployment shops have capital and engineers but were not built for scale.

The third is the relationship loop, and here Jones’s data is the verdict. Incumbents hold account visibility, executive relationships, and board access, and when they win, adjacent scope often gets rolled up into a much larger transaction lasting five, seven, even ten years. Yet renewals are down 24%. Buyers are redesigning rather than extending, and the tenure premium is not converting. So what is the market buying instead? KPMG’s Chandrasekaran said it plainly. What enterprises want now is not just AI capability but operational accountability, and as autonomous agents move into financial workflows they need clear boundaries, identity-bound execution, and auditability that runs with every agent action. The buyer has named the product. It is not another capability layer. It is the accountability architecture for agents that are already running.

The engine layer confirmed the shift by going quiet. Across two tracked runs on May 25 and May 27, zero of fifty-one tracked platform executives were active. After NVIDIA’s earnings week, with compute framed as revenue, a $20 billion processor opportunity, and Huawei conceding the China chip market, the infrastructure layer had nothing left to announce. The noise has migrated entirely to the accountability and delivery layer where these firms operate. The labs built the engines. The open question is who builds the governance architecture fast enough to matter.

Jones put a clock on it. The trend is likely to continue through the rest of 2026, he said, as providers compete aggressively for wallet share in a low single-digit growth environment. Eight billion dollars of new scope, growing 20%, in a flat total market, and the share that moves is moving to firms that can answer the accountability question with real governance frameworks, identity-bound execution, and behavior-level audit. On the May 27 run, 47 of 68 tracked consulting executives were silent. They are not absent from the market. They are still running the old loop, the one Jones says is losing, the one Bartoletti says is too slow, the one Sawchuk says procurement cannot govern. The new scope is not waiting for that loop to catch up.

Agentic systems do not just process information. They act, plan, decide, and execute faster than any oversight built to follow them.
The VuduVations Read
This edition is the clock on the wall for Consulting-as-Code. Jones has new scope growing 20% and renewals falling 24% because buyers want accountability that runs with every agent action, and the old review loop arrives after the agents are already executing. Consulting-as-Code, delivered by MCOS, delivers precisely that: a procedure the client owns outright, source-cited and auditable at the level of behavior over time, that runs without the vendor on the call or the bench in the seat. Every firm watching renewals fall still rents the engine and bills the hours against a loop too slow to govern it. The one that ships the client a procedure they own is already standing where the new scope is going.
Share
Also in this Edition
Market Signal
The contract data just flipped: being the incumbent is now the disadvantage.
New work is up about 20 percent while renewals fell about 24 percent. Buyers are redesigning the operating model, not re-signing the vendor.
Architecture
Wipro's governance chief says agents now move faster than anyone can watch them.
If agents act before review can catch up, governance has to be built into the system, not bolted on after.
Enterprise Demand
KPMG says enterprises stopped buying AI capability and started buying accountability.
The new requirement is agents with clear limits, identity-bound execution, and an audit trail on every action.
May 27, 2026← Back to the Archive