Artificial intelligence has become the fastest-moving investment category in the corporate world. Boards are asking about it, investors expect it and competitors are announcing new initiatives seemingly every week. For many Fortune 500 CEOs, however, the challenge isn’t deciding whether to invest in AI—it’s deciding where to place the first major bet.
The stakes are high because the wrong investment can consume millions of dollars while delivering little business value. Organizations across industries are launching AI labs, experimenting with custom models and deploying new tools at scale, yet many still struggle to achieve measurable returns.
That reality raises an important question: If you were making your first significant AI investment today, where would you focus—and what would you avoid?
To find out, we asked members of the Senior Executive AI Think Tank, a community of leaders and practitioners specializing in machine learning, generative AI and enterprise transformation. Their answers reveal a striking consensus about where AI creates value, why so many organizations get their priorities wrong and the foundational investments that should come before any large-scale AI deployment.
Start With Measurable Business Outcomes
Rodney Mason, Chief Marketing Officer for Minty, says CEOs should resist the temptation to begin with ambitious AI experiments and instead focus on initiatives tied directly to business performance.
“Start with AI that improves a core business outcome—revenue growth, customer experience or productivity—and has measurable ROI within 12 months,” Mason says.
He argues that organizations often become distracted by emerging technologies while overlooking the foundational elements required for success: “Invest in data quality, workflow integration and employee adoption before chasing advanced models.”
Mason points to practical applications such as customer service automation, sales intelligence, decision support systems and AI copilots as strong candidates for initial investment because they connect directly to measurable results. But equally important is knowing what not to fund.
“Avoid spending heavily on custom foundation models, AI ‘science projects’ or large-scale deployments without clear business ownership and success metrics,” he says.
For Mason, the primary cause of AI failure is rarely the technology itself.
“Most failures stem from poor data, unclear use cases and lack of change management—not lack of AI technology,” he says. “Winners solve business problems first and apply AI second.”
Build the Data Foundation Before the AI Layer
Sathish Anumula, Enterprise and Business Architect for IBM Corporation, believes the highest-return AI investment isn’t AI at all.
“My advice to the CEO would be to stick strictly to proprietary data infrastructure and internal workflow augmentation,” he says.
According to Anumula, many organizations underestimate how fragmented their data environments have become over time.
“Enterprise data is frequently unstructured and siloed. The highest ROI is achieved by structuring this data to drive Retrieval-Augmented Generation with off-the-shelf models,” he says.
Rather than launching customer-facing initiatives immediately, he recommends starting internally.
“First use AI internally—augmenting employees in high-friction areas to build cultural buy-in and immediate efficiency,” he says.
Anumula is particularly direct about where CEOs should avoid spending: “Don’t build foundational models. It’s a massive capital sink.”
His reasoning centers on competitive advantage: “The real competitive advantage isn’t owning the algorithm, but being the best at combining commoditized AI intelligence with pristine, proprietary business data.”
“AI should be treated as a transformation enabler, not a technology trophy.”
Fix Processes Before Scaling AI
Dileep Rai, Manager of Oracle Cloud Technology at Hachette Book Group (HBG), emphasizes that AI amplifies existing organizational strengths and weaknesses.
“I would advise a Fortune 500 CEO to invest first in data foundations and high-value operational use cases rather than AI itself,” Rai says.
Examples include forecasting, supply chain planning, workforce productivity and customer service functions where results can be measured quickly and clearly.
“Clean, governed data, strong integration architecture and targeted applications consistently deliver measurable ROI,” he says. “AI amplifies the quality of underlying processes; it does not fix broken ones.”
Rai also cautions against investing in AI for visibility or prestige.
“I would avoid spending heavily on custom foundation models, AI vanity projects or enterprise-wide deployments without clear business outcomes,” he says.
The most successful organizations, according to Rai, begin with a clear objective: “Many organizations overinvest in technology before identifying the decisions they want to improve. AI should be treated as a transformation enabler, not a technology trophy.”
Modernize Data and IT Before Buying More AI
Lynn Comp, Head of AI Center of Excellence at Intel, believes organizations frequently reverse the proper order of AI investment.
“AI is a productivity engine when the data fed into it is well structured and represents business reality,” Comp says.
She warns that organizations often spend heavily on AI software and infrastructure before modernizing the systems that support them.
“Spending millions on new AI equipment or services before investing to modernize their data architectures and approach to IT services results in garbage in, garbage out.”
Once foundational systems are in place, Comp recommends a disciplined approach to implementation.
“The next step after that is to focus on use cases and executing structured problem-solving strategies before starting with AI tools.”
For Comp, successful AI adoption begins with operational discipline rather than technology acquisition.
Prepare People and Governance Before Technology
Punit Bhatia, Founder of Grow Skills Store, argues that organizational readiness matters more than technology selection.
“If I were advising a Fortune 500 CEO making their first major AI investment, I would be very clear: Success will be decided less by the technology you buy and more by how well your organization is prepared to use it.”
For Bhatia, the priority is straightforward: “The priority is not tools. It is people, governance and leadership readiness.”
He recommends three immediate investments: workforce training, governance built on ISO 42001 principles and executive education.
“This is the most underestimated investment, and yet one of the most critical,” he says.
Just as importantly, Bhatia advises leaders to avoid highly visible but low-impact initiatives.
“Do not start with expensive technology platforms. Avoid visible but low-impact projects, and do not treat AI as an IT project,” he says.
His sequence for success is clear: “Start with your people. Build skills and confidence. Put governance in place to manage risk and build trust. Equip your leadership to make informed decisions. Then invest in clear business outcomes.”
Prioritize Trust, Ownership and Business Value
Venkata Kondepati, Manager of Data Architecture and Engineering at Ascentt, believes many AI initiatives fail long before a model is deployed.
“Most AI failures are not model failures; they are data quality, trust and adoption failures,” Kondepati says.
As a result, he recommends beginning with trusted data foundations, governance frameworks and clearly defined ownership structures.
“I would invest in creating a trusted data foundation, clear ownership and measurable use cases tied to revenue, cost reduction, customer experience or operational efficiency.”
Like several other experts, Kondepati advises CEOs to target outcomes achievable within six to 12 months.
He also urges restraint around experimentation: “I would avoid spending heavily on custom models, AI labs or large-scale experimentation without a clear path to value.”
His conclusion captures a central theme across the discussion.
“The winners will not be those with the most AI,” he says. “They will be those who connect AI investments to measurable business outcomes and sustainable competitive advantage.”
“If you inject cutting-edge automation into an enterprise plagued by fragmented data, you simply accelerate your mistakes at scale.”
Clean Data Beats Custom Models
Uttam Kumar, Engineering Manager at American Eagle Outfitters, believes CEOs should concentrate resources on enterprise data modernization rather than model development.
“A CEO must aggressively prioritize clean data infrastructure over flashy frontend applications,” Kumar says.
He recommends investments in enterprise data pipelines, governance frameworks and the elimination of information silos.
“Money should flow directly into modernizing the enterprise data pipeline, unifying siloed inventory databases and establishing robust governance frameworks.”
At the same time, Kumar sees little strategic value in developing proprietary large language models from scratch.
“Buying commercial foundational tools is far cheaper and more efficient,” he adds. “If you inject cutting-edge automation into an enterprise plagued by fragmented data, you simply accelerate your mistakes at scale.”
For CEOs evaluating first investments, Kumar’s recommendation is concise: “Fix the operational foundation first, utilize existing cloud models and protect your margins from empty capital expenditures.”
Demand a Credible Payback Period
Will Conaway, President of Tuxedo Cat Consulting, believes every AI investment should be evaluated through a financial lens.
“Focus first on AI use cases with a clear payback case: customer service, software engineering, sales support and document-heavy operations.”
These environments often deliver measurable gains in labor productivity, revenue conversion, error reduction and cycle-time improvement.
Conaway echoes the importance of foundational preparation: “Start with clean data, governance and workflow redesign, because these drive ROI more than model sophistication.”
He also warns against pursuing highly visible programs before demonstrating value.
“Avoid spending early on enterprise-wide AI programs, custom model builds or pilots designed for visibility rather than economics,” he says.
His investment test is simple: “If a use case cannot show a credible six- to 12-month return, it is probably not the right first investment.”
“Ask where AI can reduce friction, improve judgment, compress time, increase consistency or help people execute with more clarity.”
Improve the Work Before Buying the Technology
Divya Parekh, Founder of executive coaching brand DivyaParekh.com, believes leaders should begin with the work employees are already responsible for delivering.
“I’d tell them to start with the work people are already accountable for delivering, not with the technology they feel pressured to buy.”
Rather than evaluating tools first, she encourages leaders to map high-value business activities and identify areas where AI can improve execution.
“Map the highest-value role deliverables across the business: decisions, customer interactions, risk reviews, sales motions, reporting, strategy updates and operational handoffs.”
The next step is to determine where AI can reduce friction and improve outcomes.
“Ask where AI can reduce friction, improve judgment, compress time, increase consistency or help people execute with more clarity.”
Like many of her fellow Think Tank members, Parekh is skeptical of large investments in custom models and broad experimentation.
“Where I would not spend first: custom models, big AI labs, tool overload or scattered pilots that create excitement but never change how work gets done.”
For Parekh, the real measure of AI success is performance: “The real ROI is stronger role performance, faster execution and better decisions at scale.”
Nine Lessons for AI Investment Success
- Invest in business outcomes first. Prioritize AI initiatives tied directly to revenue growth, productivity or customer experience.
- Treat proprietary data as your competitive advantage. Structure and organize enterprise data before expanding AI capabilities.
- Fix processes before deploying AI. Technology amplifies existing strengths and weaknesses rather than correcting broken workflows.
- Modernize data architecture before purchasing new AI platforms. Better inputs consistently produce better outputs.
- Train people before scaling technology. Workforce readiness and governance are critical foundations for sustainable adoption.
- Create trusted data ownership structures. Governance and accountability drive adoption and business value.
- Spend on data infrastructure, not custom models. Commercial AI tools are often sufficient when paired with quality enterprise data.
- Demand measurable ROI within six to 12 months. Early wins create momentum and justify future investment.
- Focus on improving work execution. AI should help employees make better decisions and operate more effectively.
The First Investment Isn’t AI
For all the attention being paid to large language models, AI assistants and breakthrough technologies, the members of the Senior Executive AI Think Tank arrive at a surprisingly simple conclusion: AI success is rarely a technology problem. More often, it’s a leadership problem.
The organizations generating meaningful returns from AI aren’t winning because they found a secret model or made the largest investment. They’re winning because they know what business problem they’re trying to solve, they have the data to support it and they’ve prepared their people to adopt new ways of working. In other words, they treat AI as a business transformation strategy—not a technology purchase.
For CEOs making their first major AI investment, that’s an encouraging reality. The biggest competitive advantage isn’t hidden inside a model. It’s built through the decisions leaders make before the technology is ever deployed.
