AI talent demand is outpacing supply

Enterprise AI adoption is constrained by skills shortages

Companies are spending aggressively on the infrastructure required for AI. Hyperscalers have invested billions of dollars in data centers, computing capacity, and related infrastructure. Gartner expects global IT spending to exceed $6.3 trillion this year, with data center and infrastructure investment contributing to that growth.

But compute alone does not create business value. Enterprises still need people who understand where AI can deliver measurable results, how to integrate it into existing workflows, and how to operate it safely at scale. For many organizations, this capability is developing more slowly than the underlying technology.

This changes the AI investment equation for executives. Buying more AI tools or computing capacity can be relatively fast. Building the organizational capability to use them effectively takes longer. Companies need employees who can redesign workflows, evaluate AI outputs, manage data and security requirements, and connect AI projects to actual business objectives.

The key issue is therefore not simply whether an organization has access to AI. Access is becoming increasingly broad. The harder question is whether the organization has enough capable people to convert that access into productivity, revenue growth, better customer experiences, or lower costs.

For C-suite leaders, workforce capability should be treated as part of AI infrastructure itself. Investment plans that heavily fund technology while underfunding training, process redesign, and adoption risk producing disappointing returns. The companies that close both gaps at the same time will be better positioned to move from experimentation to scaled deployment.

Internal skill development can be more effective than relying on external recruitment

Hiring more AI specialists sounds straightforward, but the labor market creates a basic constraint: demand for AI expertise exceeds the available supply. Expanding a candidate search does not solve the problem when many companies are competing for the same limited group of experienced workers.

Kyle M.K., Senior Talent Strategy Adviser at Indeed, describes the alternative clearly: “The more useful move for tech leaders is to build these skills inside the organization they already have, rather than treat it purely as a recruiting problem.”

That does not mean companies should stop hiring specialists. Experienced AI engineers, data scientists, product leaders, and governance experts can remain critical. The more scalable approach, however, is to combine selective external recruitment with systematic development of existing employees.

Existing employees also bring something new hires often need time to acquire: knowledge of the company. They understand customers, processes, internal systems, regulatory requirements, and operational constraints. Giving these employees practical AI skills can help organizations apply the technology to relevant problems rather than pursuing AI projects disconnected from business needs.

Executives should also distinguish between employees who need deep technical expertise and those who need practical AI proficiency. Most workers do not need to become machine-learning engineers. They need to understand which tools are approved, where AI can improve their work, how to verify results, what information should not be entered into a model, and when human judgment remains essential.

The opportunity is significant. Instead of competing indefinitely for scarce talent, companies can expand the effective supply of AI skills from within. External experts can fill specialized roles, while broader internal training creates the capacity needed to make AI useful across the enterprise.

Self-directed learning is driving AI skills, but companies need more structure

Employees are not waiting for formal AI programs. Many are already experimenting with generative AI, learning new tools, and finding ways to improve their work. That initiative is valuable. It also exposes a weakness: companies are relying heavily on employees to develop important skills on their own.

Indeed data shows that only 32% of workers say they have received proper training for the job they currently perform. Kyle M.K., Senior Talent Strategy Adviser at Indeed, explains the situation directly: “They’re motivated, but the structure doesn’t exist everywhere.” He adds that because formal training remains limited, “most of the progress happening today is happening on individual initiative.”

For executives, unmanaged self-learning creates several challenges. Employees will develop different levels of expertise, use different tools, and follow different standards. Some may become highly capable, while others struggle to understand where AI is useful. Employees may also use public AI services without fully understanding company policies for confidential information, customer data, intellectual property, or regulatory compliance.

Formal training can reduce this inconsistency, but generic AI courses are not enough. Training should reflect actual roles and business processes. Finance teams need different capabilities from software developers, marketers, legal teams, or customer service employees. Companies should define what effective AI use means for each function and then train employees against those requirements.

There is also an opportunity to use existing employee enthusiasm more effectively. Organizations can identify workers who are already developing strong AI skills, give them structured support, and use their experience to accelerate adoption across teams. The objective is not to eliminate individual experimentation. It is to connect that experimentation with company priorities, governance, and measurable results.

For the C-suite, training should therefore be viewed as part of the AI deployment strategy rather than a secondary HR initiative. Technology changes quickly, so capability development will need to be continuous. Companies that create this structure can turn employee interest into repeatable business performance while reducing unnecessary operational and governance risks.

AI use remains concentrated on personal tasks rather than business applications

AI adoption is growing fast, but usage does not automatically equal enterprise value. CompTIA data indicates that business tasks account for only one-quarter of professionals’ total AI usage. The remaining three-quarters aligns with personal tasks as tools such as ChatGPT become more widely available.

This distinction matters. Employees can become familiar with AI through personal use without developing the skills required to apply it effectively inside a company. Business applications often involve additional requirements, including security, data privacy, accuracy, integration with existing systems, management approval, and measurable performance standards.

The relatively low share of business use also suggests substantial room for enterprise adoption to expand. Companies can move beyond basic experimentation by identifying specific workflows where AI can generate measurable gains. Potential areas include drafting and summarizing documents, software development, customer support, research, data analysis, internal knowledge retrieval, and routine administrative work.

Executives should avoid measuring progress simply by counting AI users or purchased licenses. A stronger measure is whether AI changes business outcomes. That could mean reducing the time required to complete a process, improving service quality, increasing employee output, lowering operating costs, or generating additional revenue. Usage without measurable impact should not be confused with successful adoption.

This also reinforces the importance of structured training. Employees already show an appetite for AI outside formal business applications. Companies have an opportunity to convert that familiarity into productive workplace use by providing approved tools, clear policies, relevant training, and well-defined use cases.

The next stage of adoption is therefore less about convincing people that AI exists and more about making it useful inside real business processes. Organizations that can make that transition systematically will be better positioned to convert growing AI familiarity into measurable returns.

Trust and transparency concerns are slowing AI adoption

The AI skills shortage is only part of the adoption problem. Employees also need confidence that the technology they are being asked to use is accurate, secure, and governed responsibly. Without that confidence, providing access to more AI tools or additional training may not translate into sustained workplace use.

An Indeed survey found that 25% of workers identify AI accuracy, ethics, and privacy concerns as the leading barriers preventing them from becoming more familiar with AI. These concerns have practical business implications. Generative AI can produce incorrect information, employees may enter sensitive data into systems without appropriate safeguards, and organizations can face questions about how AI-generated decisions or content are reviewed and used.

Kyle M.K., Senior Talent Strategy Adviser at Indeed, argues that employee hesitation should not simply be interpreted as resistance to new technology. “People aren’t avoiding these tools out of stubbornness,” he said. “They’re avoiding them because they don’t have visibility into how the tool is being used or what happens with what it produces.”

That visibility is a management responsibility. Leaders need to tell employees which AI systems are approved, what data can be entered, how generated material should be verified, and when human review is mandatory. Employees should also understand how their own AI usage is monitored and how the company handles information submitted to third-party platforms.

For the C-suite, this requires coordination across technology, security, legal, compliance, HR, and individual business units. Governance needs to be strong enough to control material risks without making legitimate AI use unnecessarily difficult. The appropriate controls will vary according to the task: using AI to draft an internal document does not carry the same risk profile as using it to support a regulated decision involving customers.

Clear governance can also support faster adoption. When employees know what they can do, which systems they can trust, and where the boundaries are, they can use AI with greater confidence. Companies can then focus training on approved, high-value applications rather than leaving workers to determine appropriate practices independently.

The executive priority is to address skills and trust together. Companies need employees who know how to use AI, but they also need rules and systems that make responsible use practical. Organizations that establish both can scale AI more confidently while managing privacy, accuracy, ethical, and compliance risks.

Key executive takeaways

• Skills are the real AI constraint: Infrastructure investment is accelerating, but workforce capability is not keeping pace. Leaders should fund skills development alongside AI tools and compute to turn spending into measurable business results.
• Build AI talent from within: External hiring alone cannot meet demand for scarce AI expertise. Combine selective recruitment with targeted training for employees who already understand the company’s customers, systems, and processes.
• Turn self-learning into structured capability: Employees are already learning AI independently, while only 32% report receiving proper training for their current roles. Create role-specific AI training that aligns experimentation with business priorities, security requirements, and governance.
• Convert AI usage into business value: CompTIA data indicates business tasks represent only 25% of professionals’ AI usage. Measure adoption through productivity, cost, revenue, and service outcomes rather than licenses or user counts.
• Build trust into AI deployment: One-quarter of workers cite accuracy, ethics, and privacy as leading barriers to greater AI familiarity, according to Indeed. Establish clear policies for approved tools, data handling, output verification, and human oversight to support confident adoption.