CNBC Television Sep 17, 2026 4 min
With BK Kalra, President and CEO of Genpact
Genpact's customers are mostly Global 2000 companies, and BK Kalra said their ambitions for AI have run well ahead of what they can actually do with it.
The two chief executives CNBC had just heard from argued against tapping the brakes on the AI build-out, and the show noted that both of them sell into it. Kalra, whose company is paid to close the gap between what a large enterprise wants from AI and what it can run, said a pause would help.
"And if I look at our client base, which is largely Global 2000, I would say aspirations are high, because whatever happened in Silicon Valley, but readiness is low."
Kalra is president and CEO of Genpact, which he said has spent three decades running mission-critical operations for the largest enterprises: finance, supply chain, sales, human resources, banking and insurance. On CNBC Television he made the case that the constraint on AI now sits inside customers rather than inside the models.
The full segment is covered here so you can skip it.
Here are the 3 takeaways that matter.
Key Takeaways
Foundation-model progress is not the bottleneck, in his account; the Global 2000's readiness to use it is
He wants a pause, not a brake — enough breathing space for enterprise capability to catch up
The labs are talking about safety because capability arrived and governance did not
Genpact sells agentic operations, which moves work from human-processed and human-validated to machine-processed and human-validated
His answer to the regulation question is speed on governance rather than delay on capability
1. High Hopes, Low Readiness
CNBC put it to Kalra that the executives it had just interviewed, Nikesh Arora and Rene Haas, both make things the AI build-out has to buy, and so might have a reason to oppose any slowdown. Kalra answered in two parts. The first was that the foundation models themselves are not in question.
He said the progress there has been exceptional and is not open to argument. The problem is on the customer side, and he described it in the same three words twice.
"And if I look at our client base, which is largely Global 2000, I would say aspirations are high, because whatever happened in Silicon Valley, but readiness is low." — BK Kalra
What he wants is a specific and limited kind of slowdown.
"And I think taking a pause — I don't think slamming brakes, but taking a pause is not such a bad thing — and creating a breathing space so that the capabilities catch up in these Global 2000 companies." — BK Kalra
Genpact, he said, is one of the bridges that builds that momentum, which is also a description of what the company sells.
2. Governance Lags Capability
The host pushed the discussion past technical capability to how people view these systems and how governments will regulate them. Kalra agreed and gave his second perspective, this time about the research labs.
"And I think the other piece, from a research lab standpoint, is, look, they are talking about safety because the capability has been developed, but the governance frameworks have not been developed." — BK Kalra
His conclusion is not to slow the capability work down but to run the governance work faster.
"However, how do you speed up the governance, speed up all of the structural pieces on safety that you need to bring in?" — BK Kalra
He said governance is built into Genpact's own agentic offerings as a structural part rather than an add-on.
3. Sovereigns and Agentic Ops
The last question was how a company like his handles governments that each have their own view of how AI should be developed and deployed. Kalra did not answer it directly. He asked for what he called a 30-second plug and described the business instead: three decades of mission-critical operations for large enterprises, and a pivot that followed the technology.
"And with the turn of technology, we pivoted to agentic ops, which is moving from human-processed, human-validated to machine-processed, human-validated, still all wrapped in responsible AI." — BK Kalra
The distinction he is drawing is about where the human sits. The machine now does the processing and the person checks the output, rather than the person doing both.
Bonus Insights
The show supplied the skeptical framing itself, noting that executives who sell components into the AI build-out have a commercial reason to argue against a slowdown. Kalra did not contest it, and his own position has the same shape in reverse: a company paid to make enterprises ready for AI benefits from a pause in which enterprises get ready.
He also never returned to the question about sovereign regulation. The segment ended on the description of Genpact's business.
Kalra's claim is that the gap holding AI back is inside large enterprises rather than inside the models, and that the fix is faster governance rather than slower capability.
Products, Companies & Tools Mentioned
Genpact (Kalra's company; three decades of mission-critical enterprise operations, now sold as agentic operations with governance built in)
If this was worth your time, send it to someone closer to the industry than you are.
Get the latest market chatter as it happens: