Artificial intelligence (AI) has reached a turning point for enterprise leaders as the challenge now lies in proving that AI can generate measurable business value and, most importantly, earn trust. As organizations move from isolated pilots to scaling enterprise AI adoption across business units, the companies succeeding with AI monetization are building systems that connect AI strategy, operational integration, and responsible governance into repeatable business outcomes.
“You don’t want to start with the technology,” Ellison says. “You want to start with what actually drives value to the bottom line,” says David D. Ellison, Chief Data Scientist and Director of AI and High-Performance Computing (HPC) Engineering at Lenovo. Over the past eight years, Ellison established and grew Lenovo’s AI Discover Center of Excellence (COE) that now supports over $1 billion in revenue.
Building Enterprise AI Around Business Value
The temptation is to begin with the latest Generative AI model or the newest machine learning breakthrough. The stronger foundation comes from identifying high value business problems first. “People try to tackle the largest problem at the beginning,” says Ellison. “That is the wrong choice because you haven’t solved problems before. You don’t have the experience, and you don’t have the systems set up.”
Instead, it’s advisable to focus on quick wins that demonstrate return on investment (ROI) early. Those smaller, high impact projects create momentum, align stakeholders, and establish confidence in broader AI adoption. This process also allows organizations to refine workflows before scaling larger initiatives.
At Lenovo, that model evolved into a hybrid structure that combines a centralized COE with distributed data science teams embedded inside business units. The COE establishes workflows, governance, and repeatable systems, while domain experts inside the business adapt those frameworks to operational realities. “I believe in the hybrid approach. Some companies try to do the COE fully approach, and that can turn into a lack of awareness of what’s going on in the business,” Ellison says. “Others are fully decentralized, and then you never develop systematic and repeatable processes.”
Moving AI From Proof of Concept to Production
One of the biggest challenges in AI commercialization is the gap between pilot projects and deployment. Many AI initiatives generate excitement during demonstrations but fail to deliver long term value. “You’ve got to design for production, not demos,” Ellison says. AI systems must be trained using real business data, integrated into existing workflows, and tied to measurable key performance indicators (KPIs) established alongside business owners. However, there is a balance to strike. “You design too much for scale at the beginning, and you get lost in the weeds of developing a huge platform instead of solving the problem.”
It follows the “fail fast” mantra: organizations that move too slowly often miss opportunities, while those that scale recklessly create systems that cannot sustain long-term growth. “If you’re not embarrassed of your first product, you waited too long to deploy,” he says. At Lenovo, Ellison has helped move AI from proof of concept to production across multiple industries, including logistics, healthcare, enterprise technology, and motorsports. Among the company’s notable achievements are a NASCAR Smart Pit Box with 99.6% accuracy which helped deliver a first place finish, and an Elder Care solution that reduced emergency room visits by 50%. “You want something that scales and gets used daily, not occasionally,” Ellison says. “So you can find the errors, find the problems, and build on them.”
Responsible AI Became a Competitive Advantage
As enterprise AI adoption accelerates, governance has become inseparable from commercialization. “AI that nobody trusts, nobody uses,” he says. At Lenovo, Responsible AI is built around six core pillars: diversity and inclusion, privacy and security, accountability and reliability, transparency, and explainability, and environmental and social impact. Embedding those principles early creates a faster path to deployment because governance is integrated into the system itself.
Trust also directly affects adoption. Enterprise customers increasingly demand visibility into how AI systems make decisions, how data is managed, and how governance controls are enforced. Organizations that fail to establish that trust often struggle to move beyond experimentation. This is especially important as companies deploy generative AI and autonomous agents into regulated enterprise environments. Governance, orchestration, and integration have become just as important as the underlying models themselves. “The systems win,” Ellison says. “It’s not a standalone model that wins.”
The Future of AI Commercialization
Looking ahead, Ellison sees the future of enterprise AI strategy shifting toward agentic AI systems that automate real workflows rather than simply generating insights. At Lenovo, one initiative used agentic AI to reduce a four-hour network deployment process to just 12 seconds. The opportunity, however, extends far beyond automation alone. The next phase of AI monetization will come from combining agents, governance, orchestration, and enterprise data into systems capable of executing tasks autonomously, while remaining transparent and secure.
Equally important is leadership, and Ellison warns against separating AI strategy from execution through what he describes as an “ivory tower” approach, where technical experts advise from the sidelines without owning outcomes. “You need an AI expert that owns the execution also,” he says. “Otherwise, no matter how good the AI knowledge is, they’re not going to know what actually works in the field because they’re not executing.” As enterprise AI matures, that combination of technical expertise, operational accountability, and responsible governance may ultimately determine which organizations successfully build AI businesses from the ground up.
Follow David D. Ellison on LinkedIn for more insights on enterprise AI strategy, scaling AI from pilot to production, and real-world deployment of AI and HPC systems.
