There’s something deceptively smooth about the phrase enterprise AI breakthrough—as if one could neatly package the sprawling reality of corporate data silos, reluctant boardrooms, and half-panicked CIOs into a tidy headline.
Still the phrase set the pace of this on-demand NVIDIA GTC conversation featuring Christine Ahn, NVIDIA Chief Commercial Officer & Lead Client Service Partner at Deloitte; Steve Brown, Managing Director, AI and Data Operations, Deloitte; and moderator John Mao, VP of Business Development and Alliances at VAST Data.
The wide-ranging chat painted a subtler, more compelling picture: enterprise AI isn’t just a story of triumphant innovation; it’s also still a narrative of anxiety, practical limitations, and hard-won technical readiness—or (gulp) lack thereof.
What emerged from this dialogue was a nuanced picture: beneath the sheen of new technology lies a serious reckoning with technical readiness, strategic clarity, and organizational capacity.
We’ll unpack, question by question, what Deloitte is hearing from its enterprise base and what is, albeit slowly, becoming the new normal.
Navigating the AI “DeepSeek” Shockwave
Mao kicked off the conversation with a newsy sticking point: DeepSeek and Stargate are highly visible but are they laying new groundwork for companies to follow?
Ahn admitted candidly that DeepSeek initially triggered widespread anxiety, pushing enterprises into a defensive posture:
“I think DeepSeek sort of sent everyone into a spiral… every company is really circling around the power source, which is NVIDIA” but did lay a broader awareness that enterprises must build technical agility into their operations to navigate future shocks—expected or otherwise.
Brown pinpointed DeepSeek’s significance as not merely another incremental GPU upgrade, but as a shift in AI efficiency from the ground up, saying the innovation here was “optimizing at the lower end of the stack,” reshaping how efficiently enterprises deploy their AI hardware resources.
Meanwhile, all agreed the simultaneous emergence of Stargate, a colossal $500 billion global commitment toward sovereign AI infrastructure, evoked a sense of urgency akin to a high-tech Cold War.
AI’s True Challenge: Beyond the Proof-of-Concept
Steering the conversation back toward enterprise realities, Mao pressed Ahn and Brown on what they’re actually hearing from clients about more grounded AI adoption challenges.
Ahn acknowledged that a core difficulty isn’t simply technical—it’s educational. Enterprises are struggling to internalize the real potential of AI beyond abstract promises.
“It seems so, so easy,” she reflected, “but that’s the biggest challenge we have… educating our team and our account leadership… bringing the art of the possible thinking of what AI is going to do”. This underscores a subtler truth: readiness isn’t just infrastructure—it’s a wholesale shift in strategic mindset.
Brown agreed, highlighting a pervasive uncertainty he sees repeatedly across industries: Many enterprises are stuck in endless experimentation cycles, uncertain how or when to move beyond cautious pilot projects. “They are either unaware or still experimenting with different ways to support AI workloads,” Brown observed, noting Deloitte’s role is to help cut through this fog: “We seek to bring some clarity… technically and financially.”
The message is clear: AI success depends as much on strategic clarity as it does on technological capability.
AI at the Boardroom Level: All Aboard the Hype Train?
According to Ahn, AI now dominates the highest-level conversations across nearly every industry Deloitte encounters.
There isn’t a client that I’ve spoken with that has not had this conversation at their board,” Ahn asserted, signaling a marked shift in how seriously enterprise leaders are taking AI strategy.
Yet within this excitement remains a nagging uncertainty: enterprises grasp AI’s promise, but translating that promise into practical, measurable results remains complex and elusive.
Enter the Technical Realities
The conversation veered toward AI’s chief technical bottleneck—data.
Mao pressed on the greatest barriers enterprises face, and Ahn identified the age-old nemesis of fragmented, messy data as central to the challenge.
Enterprises, Ahn noted, often perceive the state of their data as too chaotic to support meaningful AI adoption. Yet, paradoxically, she warned against paralysis: “There’s fear that they have to fix all their data first before they can really adopt AI… you’re going to be left behind if you don’t start now.”
The insight here is subtle yet crucial: technical readiness isn’t about attaining data perfection; it’s about developing the flexibility to work effectively despite imperfections.
Brown echoed this pragmatic perspective, explaining that addressing enterprise data challenges requires continuous, incremental improvement rather than expecting a one-time fix.
The AI Factory concept, which Deloitte has developed, seeks precisely this—to establish enterprise-scale clarity amid chaos, enabling businesses to access and integrate data from multiple sources and multiple modes of infrastructure—cloud, on-prem, hybrid—without sacrificing compliance, security, or performance.
Agents Everywhere: The Quiet Automation Revolution
Mao pivoted toward practical trends, asking Ahn and Brown what specific AI use cases were resonating with enterprises.
Ahn outlined a clear trend: software providers like SAP, Oracle, Salesforce, and Microsoft embedding agents directly into their platforms to automate critical back-office processes. She painted an imminent future where enterprises manage thousands of AI agents streamlining workflows such as insurance claims, banking fraud detection, and internal supply chain management.
All of the enterprise software ISVs are rapidly creating agents on top of their software platforms,” Ahn said, adding, “Insurance is looking at completely changing claims processing.
It’s a subtle but significant transformation, embedding AI directly into the core technical fabric of enterprises.
Looking Ahead: AI’s Enterprise Future
Finally, Mao asked the panelists to speculate on enterprise AI’s trajectory over the next three years.
Ahn predicted widespread adoption of industry-specific language models, an explosive proliferation of back-office AI agents, and, intriguingly, AI’s increasing physical presence through robotics: “Back office functions will have thousands of agents processing hire-to-retire, purchase-order-to-payment… moving from invisible agents to physical agents.”
Brown added complementary perspectives, forecasting an insatiable appetite for multimodal data, increasing deployment of edge AI solutions to handle latency-sensitive applications, and growing governmental involvement shaping AI’s future regulatory landscape.
“The appetite for data will become insatiable… AI at the edge is going to become increasingly important,” Brown concluded.
Bringing it All Together…
The GTC session offered a balanced, though subtly unsettling, reflection on enterprise AI’s future.
Beneath the optimism about technological possibilities lay a deeper concern: true technical readiness requires confronting messy realities—imperfect data, perpetual uncertainty, organizational silos—that AI alone cannot instantly resolve.
It’s clear from Deloitte’s insights that the coming AI breakthrough isn’t just a technological transformation but an ongoing, enterprise-wide reckoning.
Watch the full discussion here: https://www.nvidia.com/en-us/on-demand/session/gtc25-s74356/
