AI infrastructure spending has entered an era of historic scale. Microsoft, Google, Amazon and others have collectively committed hundreds of billions of dollars to expand compute capacity, even as analysts warn that parts of the market may be racing ahead of sustainable demand. For enterprise leaders outside Big Tech, the stakes are just as high, but the margin for error is far smaller.
While AI investment continues to accelerate, many organizations struggle to connect infrastructure outlays to near-term financial returns, raising concerns about capital efficiency and long-term value creation.
Members of the Senior Executive AI Think Tank—a curated group of executives and leaders shaping enterprise AI strategy—argue that the debate should not center on whether to invest, but how. What follows is a playbook drawn directly from their insights—detailing how seasoned leaders evaluate billion-dollar bets, stage risk intelligently and ensure AI infrastructure becomes a durable advantage rather than an expensive monument to hype.
Building Capability That Converts to Business Value
For Raghu Para of Ford Motor Company, AI infrastructure decisions begin with a simple but unforgiving test: Can capability be converted into outcomes?
“When considering AI infrastructure investments, we focus on building lasting strategic advantage, not chasing hype,” Para says. “I use a ‘Capability-to-Conversion’ lens to evaluate how each component—compute, models, data pipelines—translates into real outcomes like automation gains, faster decisions or increased customer value.”
Para emphasizes that financial discipline does not mean small ambition. Instead, it requires explicit thresholds for ROI, including time-to-impact and scalability.
“Disciplined investment means aligning infrastructure to core business loops, not experimental side projects,” he says. “Ambitious bets are critical, but only when grounded in architectures that are explainable, cost-flexible and ready for regulatory scrutiny.”
“Approve data centers only when tied to specific revenue opportunities or cost savings—no infrastructure without use cases.”
No Infrastructure Without Use Cases
Jim Liddle, serial entrepreneur and enterprise AI strategist, has spent more than 25 years building and scaling technology companies, including a successful exit to a leading cloud data management unicorn. That experience informs his blunt approach to AI capital allocation.
“Approve data centers only when tied to specific revenue opportunities or cost savings,” Liddle says. “No infrastructure without use cases.”
Liddle advocates a “revenue per GPU” mindset, tracking utilization and business outcomes per dollar of compute. And to manage uncertainty, he recommends staging investments with six- to 12-month decision gates rather than committing years of capex up front.
He also applies a 70-20-10 allocation model: “Allocate 70% to proven AI use cases with clear ROI, 20% to promising experiments and 10% to moonshots. This will prevent both under-investment and reckless spending.”
Data Maturity as a Nonnegotiable Gate
At American Eagle Outfitters, Engineering Manager Uttam Kumar views data readiness as the ultimate gating factor for AI infrastructure approval.
“The primary nonnegotiable criterion is demonstrated organizational data maturity and readiness,” Kumar says. “A billion-dollar engine is useless without premium fuel, and in AI, data quality is that fuel.”
Kumar stresses that this ensures governance, data literacy and quality are robust enough to immediately leverage new AI capability.
Risk-Adjusted ROI Over Hype Cycles
For Aditya Vikram Kashyap, Vice President of Firmwide Innovation at Morgan Stanley, AI infrastructure represents nothing less than a new industrial backbone—but only if governed correctly.
“I greenlight only where capital delivers enduring advantage,” Kashyap says, “one that’s measured through risk-adjusted ROI, resilience under stress tests and cross-ecosystem impact.”
Kashyap warns that vision without discipline quickly turns into speculation. “The goal is not to chase the hype cycle, but to architect capabilities that compound over decades,” he says. “Legacy will not be judged by billions spent, but by whether we built foundations strong enough to outlast bubbles and define the next economy.”
“Authentic leadership doesn’t chase every new wave; it invests with clear intention.”
Investing With Intention, Not Fear
Dileep Rai, Manager of Oracle Cloud Technology at Hachette Book Group (HBG), frames AI infrastructure as both a technical and human leadership challenge.
“Authentic leadership doesn’t chase every new wave; it invests with clear intention,” Rai says. He urges executives to evaluate ROI, milestones and practical value while ensuring technology elevates people rather than replacing judgment.
Rai supports bold commitments—but only in disciplined phases tied to verified outcomes.
“Actual progress is conscious growth,” Rai says, “where technology amplifies human intelligence, not ego or automation for its own sake.”
Measuring ROI, TCO and Societal Impact
Chandrakanth Lekkala, Principal Data Engineer at Narwal.ai, urges executives to look beyond technical feasibility to full economic and social impact.
“The growing amount of money invested in this sector indicates a bubble,” Lekkala says. “Executives should ensure they realize profits through ROI and total cost of ownership analysis.”
He advocates intelligent monitoring systems that forecast benefits and risks while tying AI activity directly to corporate goals. Lekkala also raises a cautionary note on inequality and employment displacement.
“Effective ROI models eliminate monetary risk,” he says, “but leaders must also reflect on the broader social and economic impact when planning infrastructure.”
“True discipline isn’t spending less. It’s buying adaptability.”
Infrastructure as Strategic Optionality
At GARMIN, Bhubalan Mani, Lead for Supply Chain Technology and Analytics, frames AI infrastructure not as sunk cost, but as strategic optionality.
“The real challenge isn’t ROI precision,” Mani says. “It’s that organizations conflate spending with capability building.”
He applies real options thinking, staging investments as learning gates and establishing kill criteria upfront. “Courage to exit separates leaders from bubble casualties,” he says.
Beyond traditional ROI, Mani recommends measuring decision velocity and strategic degrees of freedom, as well as building modular, reversible architectures over “monolithic commitments.”
“True discipline isn’t spending less,” Mani says. “It’s buying adaptability.”
Turning Vision Into Structured Risk
Sandesh Gawali, Senior Director of Data and AI Products at Salesforce, views nine-figure AI investments as deliberate, structured risk.
“Hope is not a strategy,” Gawali says, “so the job is to turn vision into structured risk.”
He backs large capex only when three elements align: a sharp strategic thesis, a clear line from capability to economics and reusable architecture.
Boards, he notes, will underwrite long-term AI risk—but only when it is staged, focused and constantly tested.
Key Tips for Funding AI Infrastructure
- Tie infrastructure to outcomes, not hype. Every major AI investment should map directly to automation, revenue growth or decision advantage.
- Fund use cases, not capacity. Approve compute only when linked to specific economic value and utilization metrics.
- Gate investment on data readiness. Infrastructure without mature data governance amplifies risk rather than return.
- Measure risk-adjusted ROI. Stress-test AI investments for resilience, regulation and long-term advantage.
- Invest with intention, not FOMO. Phase spending around milestones that empower people and verified outcomes.
- Model full ROI and societal impact. Include TCO, monitoring systems and workforce considerations.
- Buy optionality, not rigidity. Stage investments with kill criteria and modular architectures.
- Turn vision into structured risk. Align strategy, economics and reusable platforms before committing capital.
Discipline Is the Differentiator
As AI infrastructure spending continues to accelerate, scale alone will not determine the winners. According to the members of the Senior Executive AI Think Tank, disciplined frameworks—rooted in outcomes, data readiness, staged risk and adaptability—separate enduring advantage from expensive regret.
For executives, investing boldly is merely one challenge. The real test of leadership is whether those investments can compound value long after the hype cycle fades. The leaders who succeed will be those who balance conviction with scrutiny and ambition with accountability.
