Not Every AI Workload Belongs in the Same Place

Artificial intelligence (AI) is rapidly becoming part of the higher education landscape. Institutions are exploring AI-enhanced advising, research computing, cybersecurity operations, student success initiatives, administrative automation, and many other applications. Yet as these efforts move from experimentation to implementation, many leaders are discovering that one of their greatest challenges is not simply identifying valuable use cases but determining how AI capabilities should be supported, governed, and sustained over time.

The conversation is often framed around AI itself: Which models should be used? Which platforms should be adopted? Which capabilities should be prioritized? These are important questions, but they may not be the most important ones. As colleges and universities invest in AI and other emerging technologies, they face a more fundamental challenge: not every workload belongs in the same place.

Institutions can evaluate AI investments through both return on investment (ROI) and return on value (ROV).^Footnote1 Those measures can help leaders determine whether the capability is worth pursuing. If it is, a new question arises: How should the capability be supported in a way that aligns with institutional priorities, data responsibilities, financial realities, and future needs?

This may sound like a technology question, but it is increasingly a leadership question. Decisions about where workloads run, where data resides, and how capabilities are supported influence research competitiveness, student success, cybersecurity posture, institutional agility, and long-term financial sustainability.

The Broader Challenge Exposed by AI

AI did not create many of the challenges institutions face today. In many cases, it simply exposes them.

For decades, colleges and universities have made technology decisions in response to specific needs, acquiring new systems to support enrollment, improve student success, advance research, strengthen cybersecurity, or address operational inefficiencies. Those decisions often reflected the priorities, constraints, and information available at the time.

Over time, however, many institutions accumulated technology portfolios shaped more by individual decisions than by an overarching strategy. As a result, data became distributed across platforms. Departments adopted solutions to address immediate challenges, often without full visibility into how those choices might affect future integration, governance, cost, or flexibility. Today, AI is placing new demands on those environments because its workloads often depend on large amounts of data, substantial computing resources, and system integrations, while raising evolving governance considerations.

The question is no longer simply whether an institution should adopt AI. The emerging challenge is determining how AI-enabled capabilities should fit into the broader technology ecosystem without deepening fragmentation or creating the next generation of technology debt.

Different Requirements for Different Workloads

A student success platform has different requirements from those of a research computing environment. An AI assistant used for administrative work may have a different data profile and economic model from those of an institutional analytics environment. Cybersecurity operations may prioritize resilience and rapid response, while an academic innovation environment may prioritize experimentation and adaptability.

Despite these differences, discussions often begin with cloud versus on-premises infrastructure, capital versus operating expenses, or a preferred platform. Those considerations matter, but they do not answer the strategic question leaders should consider first: What role does this workload play in advancing the institutional mission?

Rather than beginning with deployment models, leaders should start by understanding the workload itself. What outcomes does it support? What obligations accompany its data? What is the sustainable economic model? How much flexibility will be needed as priorities and technologies evolve?

To support those discussions, I propose a Strategic Workload Placement Framework. The framework is built around four dimensions: strategic importance, data stewardship, economics, and future flexibility. The objective is not to prescribe a technology solution but to create a repeatable process that aligns technology decisions with institutional priorities.

The Strategic Workload Placement Framework

1. Strategic Importance

Strategic importance measures how critical the workload is to institutional mission, differentiation, and long-term success. Leaders should ask whether it directly supports teaching, research, student success, operations, or a strategic priority, and whether failure to support it would create meaningful institutional risk.

2. Data Stewardship

Data stewardship assesses responsibilities related to privacy, security, governance, intellectual property, compliance, and trust. Leaders should consider the data involved, the obligations attached to it, and the oversight required to use it responsibly.

3. Economics

Economics evaluates the complete economic picture, including life-cycle costs, integration, staffing, scalability, and ongoing support. The goal is not simply to minimize costs but to maximize sustainable institutional value. This is where ROI and ROV can help leaders balance measurable returns and outcomes such as student success, research capacity, resilience, and institutional competitiveness.

4. Future Flexibility

Future flexibility examines how well the support model can adapt as technologies, regulations, funding models, and institutional priorities change. Flexibility does not mean predicting every scenario. It means preserving reasonable options, interoperability, and the ability to evolve without repeatedly rebuilding the foundation.

Figure 1 shows an example scorecard for applying the four dimensions to an AI-powered student success analytics workload.

Figure 1. Strategic Workload Placement Scorecard

Dimension	Assessment	Guiding Question
Strategic Importance	Very High	How directly does the workload advance mission and priority outcomes?
Data Stewardship	Very High	What obligations accompany the data, users, and institutional trust?
Economics	High	What are the life-cycle costs, ROI, ROV, and sustainable funding requirements?
Future Flexibility	High	How much adaptability, interoperability, and scalability will be needed?

The objective is not to generate a precise score but to develop a workload profile that supports more informed decisions about governance, investment, placement, and long-term support.

The same framework can produce different profiles for other workloads. An AI research environment may rank exceptionally high in strategic importance and future flexibility because it supports grant-funded discovery, involves intellectual property, requires specialized computing, and accommodates rapidly changing research requirements. An AI assistant used for administrative work may enable meaningful efficiency gains but involve different strategic and stewardship considerations. A cybersecurity workload may emphasize resilience, integration, and institutional risk.

These differences matter because AI is not a single workload. It is a growing portfolio of capabilities, each with distinct objectives, data requirements, economic models, and support needs.

The Move from Workload Profiles to Placement Decisions

Once leaders establish a workload profile, they can evaluate the most appropriate support model. This reverses the order of many technology discussions. Instead of starting with infrastructure, cloud strategy, or a preferred platform, institutional leaders begin with mission, stewardship, economics, and flexibility.

The resulting conversation may include public cloud, private cloud, software as a service, campus infrastructure, research computing, edge environments, or a hybrid combination. The framework does not predetermine the answer. It helps to ensure that institutional leaders ask the right questions before making a decision.

The answers will appropriately vary among research universities, regional institutions, community colleges, and specialized institutions. The objective is not to create identical environments across higher education but to establish a consistent decision process that helps leaders align technology choices with institutional mission, resources, and responsibilities (see Figure 2).

Figure 2. Strategic Workload Portfolio Matrix

	Lower Flexibility Requirement	Higher Flexibility Requirement
Higher Strategic Importance	Protect and Sustain Core systems Sensitive institutional data	Invest and Orchestrate Research AI Student success analytics Cybersecurity operations
Lower Strategic Importance	Simplify or Retire Duplicative applications Low-use legacy tools	Experiment and Adapt Classroom AI pilots Administrative copilots Innovation sandboxes

The matrix is a discussion aid, not a placement calculator. Data stewardship and economics remain essential overlays. Leaders should use the scorecard to establish a workload profile and then use the matrix to compare priorities across the portfolio.

The Emergence of Hybrid AI as an Operating Model

One of the clearest implications of this approach is that no single deployment model is appropriate for every workload. Some capabilities may benefit from the scalability and rapid innovation of external platforms. Others may have distinct requirements for stewardship, performance, predictability, integration, or proximity to institutional data.

This is where hybrid AI enters the conversation. Hybrid AI is a strategic approach that orchestrates personal, enterprise, and public AI models and data across devices, the data center, edge environments, and the cloud, allowing organizations to run AI where it delivers the best balance of performance, security, governance, and cost. Hybrid AI is often described as a technology architecture, but in higher education it may be more useful to view it as an operating model that reflects the reality of diverse institutional workloads.

Hybrid AI should not be treated as a predetermined destination. It often emerges as the outcome of deliberate decision-making. When institutions evaluate workloads according to strategic importance, data stewardship, economics, and future flexibility, they may conclude that different capabilities require different environments and support models.

A Strategic Capability for What Comes Next

AI is reshaping higher education, but the most consequential decisions may have less to do with models and algorithms than with the foundations required to support them. The same questions apply to advanced analytics, research computing, cybersecurity, student success, and other emerging workloads.

The Strategic Workload Placement Framework is not intended to prescribe a single architecture. It offers a practical way to evaluate AI-era workloads through the lenses of strategic importance, data stewardship, economics, and future flexibility. By focusing first on workload characteristics rather than technology preferences, leaders can make decisions that better align with institutional mission, governance requirements, financial sustainability, and long-term goals.

Perhaps the most important lesson is also the simplest: not every workload belongs in the same place. Institutions that develop a repeatable process for determining which workloads matter most, what responsibilities accompany them, and how best to support them will be better positioned to maximize both ROI and ROV while preserving the flexibility to respond to whatever comes next.

Thank You

Note

1. See, Hernan Londono, "How AI Impacts Datacenter Spending in Higher Education: ROI and ROV Considerations in the Age of AI Growth," Lenovo Tech Today, June 16, 2026. Jump back to footnote 1 in the text.↩

---

Hernan Londono is Chief Technology and Innovation Strategist for Higher Education at Lenovo.