Balancing Trade-Offs in Hybrid Cloud and the Infrastructure Behind Scalable AI – with Jason Hardy of Hitachi Vantara

This interview analysis is sponsored by Hitachi Vantara and was written, edited, and published in alignment with our Emerj sponsored content guidelines. Learn more about our thought leadership and content creation services on our Emerj Media Services page.

Organizations across various industries are making significant investments in enterprise AI capabilities to enhance their efficiency and gain a competitive edge in their respective markets. However, many face substantial hurdles, including integrating AI into existing systems, addressing data quality issues, and aligning AI initiatives with business objectives.

These challenges often lead to increased costs and delayed returns on investment. Recent studies show that a substantial number of AI projects fail to meet expectations: One report by S&P Global Market Intelligence found that 42% of businesses abandoned most of their AI initiatives in 2025, up from 17% the previous year. 

Similarly, an MIT study revealed that 95% of generative AI (GenAI) deployment efforts fail to achieve desired outcomes. These statistics underscore the importance of strategic planning and execution in the adoption of AI.

On a recent episode of the ‘AI in Business’ podcast, Emerj Editorial Director Matthew DeMello sat down with Jason Hardy, Chief Technology Officer of AI at Hitachi Vantara, to discuss how organizations can adopt AI strategically, balancing infrastructure, data readiness, cloud use, compliance, and long-term ROI.

This article brings out two essential insights every organization needs for successfully adopting and scaling AI:

• Balancing cloud and data center use: Running core AI in data centers while using the cloud for low-priority tasks controls costs, boosts ROI, and supports sustainability.
• Reframing ROI in AI for long-term success: Valuing lessons from failures alongside wins and adopting a long-term, three-year view, where persistence builds the foundation for competitive advantage.

Guest: Jason Hardy, Chief Technology Officer of AI at Hitachi Vantara

Expertise: Artificial Intelligence, Generative AI, Cloud Computing

Brief Recognition: Jason leads the development of Hitachi Vantara’s AI strategy and portfolio. With over 20 years of experience in data-driven technology, Jason has previously served as the CTO for Hitachi Vantara’s Data Intelligence portfolio. Jason is also a seasoned consultant, having advised numerous global clients on data strategy, and is a frequent speaker at industry events.

Balancing Cloud and Data Center Use

Jason opens the conversation by explaining the dilemma companies face when adopting AI infrastructure: whether to commit upfront or start gradually and thoroughly. Initially, there was a rush to buy GPUs without a clear plan; however, the approach is now more methodical. 

Jason emphasizes to the enterprise audience that it’s not just about GPUs, he says; organizations must consider where their data comes from, how systems integrate, and what outcomes they aim for. These complexities often become clear only midway through the journey. 

Additionally, GPUs demand significant power, raising sustainability concerns until the technology becomes more efficient. Many businesses are adopting hybrid strategies, combining on-premises and cloud resources to strike a balance between performance, cost, and ROI, without overspending or compromising sustainability.

He also emphasizes that implementing AI requires a structured and strategic approach, rather than rushing in without a plan. Companies can’t simply expand data center space or treat AI as a plug-and-play product; they are not something you “buy” and instantly turn on. 

Instead, he believes that AI adoption is an enterprise journey that requires planning across physical infrastructure, data management, and talent. 

Fundamentally, Jason advises enterprise leaders on the need to align execution with business strategy, considering ESG impact, hybrid cloud strategies, and overall ROI. The era of experimenting to see what sticks is over, he notes emphatically. Boards and CEOs demanding AI must understand that success comes from deliberate planning and phased execution, not quick fixes.

Jason also explains that businesses initially turned to the cloud as a convenient space for AI experimentation, and it remains a good option for specific workloads. However, the cloud works best for lower-priority, horizontal use cases, such as HR or finance, where flexibility matters more than speed.

The problem is cost control; cloud expenses rise quickly as user activity increases, and companies can’t easily limit usage without shutting down services:

“The ROI starts to erode if you’re not getting a high amount of value out of your cloud platform, but you’re paying a considerable amount of cost for your per token, or however you’re paying for it.

That’s why we’re starting to see a bit of repatriation, bringing data back into the data center, and then using the cloud more strategically. In the short term, leaders will run the data center because it’s a finite cost.

They want the investment to be one-and-done, and then, obviously, power and cooling still matter. So they’ll use the cloud for that quick burst, or use the cloud for low priority assets as a matter of strategy, as I said.”

— Jason Hardy, Chief Technology Officer of AI at Hitachi Vantara

Reframing ROI in AI for Long-Term Success

Jason explains that companies are adopting a hybrid approach, using the cloud for bursts, low-priority tasks, or specific strategic needs—while moving core workloads back on-premises. 

The shift, Jason says, enables enterprises to access sensitive data locally for compliance and security reasons, while delivering real-time outputs, which is critical in industries such as manufacturing and energy, where latency from cloud round-trips is too slow. He notes that generative and energy-focused AI applications demand this on-site capability. 

At the same time, he feels geopolitical and regulatory pressures are accelerating “sovereign AI” initiatives, where nations build in-country AI infrastructure to maintain control over critical technology.

Jason’s examples of these kinds of initiatives include government-backed investments in the Middle East and Japan. These efforts represent repatriation “on steroids,” aiming to create cloud-like experiences within national boundaries while preserving autonomy, compliance, and cultural alignment.

Hen continues by saying that ROI in AI can’t be measured in the traditional sense of cutting costs or reducing headcount. Instead, it needs to be reframed to account for learning through failure. Most AI projects — around 90% — fail or never make it to production, but those failures only generate insights that improve the chances of success in future initiatives when business leaders learn from their mistakes: 

“What we’ve also found is that roughly 86 percent of pilots that succeed prove that the juice is worth the squeeze.

Whereas with outright failure, all those projects that won’t see the light of day, we’re seeing the improvements from these projects that do make the light of day make up for the difference in investment.

So, there is a real benefit to AI for the sake of understanding your organization better than ever. But again, it’s that balance again, and leaders have to make judgments appropriately against business goals.”

– Jason Hardy, Chief Technology Officer of AI at Hitachi Vantara

He stresses that AI investments should be evaluated over a three-year horizon, rather than based on short-term results. AI is a competitive advantage, so if competitors succeed first, they’ll reap the rewards. ROI should factor in both successes and the lessons from failures. 

Companies must accept some “burn” along the way instead of abandoning projects too quickly, because persistence, even through failures, ultimately leads to payoff, making the effort worthwhile.