- NTT Data and AWS' expanded partnership is focused on the growing cost and complexity of scaling AI
- Priorities include eliminating data siloes and ensuring AI can scale without blowing the budget
- Overcoming these hurdles could unlock billions of dollars in additional agentic AI growth
AI is supposed to make life easier for enterprises, but getting the tech off the ground is proving harder – and more expensive – than hoped. That’s why NTT Data and AWS are stepping in, expanding their longstanding collaboration to solve two key hurdles enterprises face in deploying AI at scale: data and dollars.
Shashi Gupta, managing director and global head AWS capability lead for Cloud and Security at NTT Data, told Fierce data siloes remain a critical roadblock to successful AI implementation.
Take a financial services company as an example. In a typical trade transaction lifecycle, Gupta said there are “a hundred applications that come into the picture.” The problem is that the data for each of these is segmented, and these segments are not connected well in today’s world.
“That’s where the layer between the user and the multiple data [sources] becomes an important part,” he said. “What we are doing with AWS here is creating the right layer for our customers to use the right taxonomy to pull their data and get the right insights from that.”
Getting this data layer right, he added, also has implications for enterprises’ ability to train specialized small language models and tap into the innovation those can provide.
But data is only one part of the AI equation for enterprises. There’s also the pesky issue of money, balancing AI costs with revenue generation. That’s the “return on investment” everyone is hunting for.
“What gets harder as systems scale that people don’t really expect is that it’s not just [about] the technology,” Gupta said. The technology can be there, but if it doesn’t deliver on the complex outcomes a business wants then it’s no good.
The other problem is that there’s a price tag involved with taking a pilot to production scale.
“The other challenge that the industry is facing today is with agentic, the pilots and the experiments are super successful, but when it goes to production, it gets too expensive, it gets too unpredictable,” he said. “What’s important for us is to focus on not just the architecture but the adoption.”
FinOps, he added, has to be part of the discussion from Day 1.
That’s where AWS’ Process-to-Agent (P2A) framework for converting traditional business workflows into agentic AI systems comes into play. As AWS exec Chandra Pinapala noted on LinkedIn, P2A spans “process discovery, business reimagination, systems integration, governance frameworks [and] change management.”
Why do collaborations like the one between NTT Data and AWS matter?
Because while the global agentic AI market could grow from $8.5 billion this year to $35 billion by 2030, Deloitte said the latter number has the potential to reach $45 billion. But that’ll happen “only if enterprises and providers perform proper orchestration” of AI agents.