Executive Summary

As corporations grapple with the rapid advance of artificial intelligence (AI), upskilling employees in AI is now a strategic imperative. Global spending on corporate learning already exceeds $340–400+ billion annually (^[1]) (^[2]), yet firms struggle to meet growing AI skill demands. Industry surveys illustrate this gap: in one study 89% of companies saw an AI skills shortfall among their workforce, but only 6% had implemented any substantial upskilling program (^[3]). Similarly, two-thirds of workers reported they would avoid employers lacking sufficient AI training programs (^[4]), and over 60% of employees want their companies to provide AI training (^[5]).

Organizations face a fundamental choice in addressing this need: should they “build” AI training programs internally, “buy” them from external providers, or adopt a “hybrid” of both approaches? Each option entails distinct tradeoffs in cost, speed, customization, and control. The decision depends on factors such as budget, timeframe, content specificity, and scale. Our review finds:

• Building In-House allows maximum customization and alignment with corporate strategy and culture (^[6]) (^[7]), but requires substantial upfront investment and expertise (^[6]) (^[8]). For example, Lloyds Banking Group launched its own AI Academy to train all 67,000 employees, aiming for 100% upskilling by end of 2026 (^[9]) (^[10]). This “build” strategy yields tailored content and intellectual property ownership (^[11]), but demands lengthy development (often 12–24 months) (^[12]) and high initial costs (potentially millions of dollars) (^[12]). In-house training can also struggle with scalability and instructional design gaps (^[6]).

• Buying Third-Party training (via external courses, MOOCs, or consultants) offers speed and access to expertise (^[13]) (^[8]). Large eLearning platforms and associations (e.g. Coursera, Udemy, CompTIA) and industry consortia provide turnkey AI curricula. For instance, CompTIA’s “AI Essentials” course (Dec 2025) targets desk workers, addressing practical AI tool use (^[14]). Cisco “Learn with Cisco” pledged to train 1 million Americans in AI/digital skills over four years (^[15]). Buying can scale quickly and leverage best-practice content (^[13]) (^[8]), often at lower upfront cost. However, off-the-shelf programs may lack company-specific relevance and control (^[6]) (^[8]). Vendor lock-in, licensing fees, and potentially higher long-term cost at scale (e.g. subscription fees rising with headcount (^[16]) (^[17])) are concerns.

• Hybrid Approaches attempt to capture the benefits of both. Companies often build core training modules in-house (for proprietary processes or strategic topics) while buying or outsourcing specialized content (e.g. compliance, general AI literacy) (^[18]) (^[19]). In a published case study on St. George Bank (Australia), a blended model of internal core modules plus external regulatory courses “yielded stronger learner commitment” at scalable cost (^[18]). Hybrid models can maintain brand alignment and flexibility (^[18]), yet require careful management of multiple vendors and consistent learner experience.

To inform strategy, we analyze these approaches in depth. Our report synthesizes current data on corporate learning and AI, evidence from recent industry research, and illustrative case studies. We include detailed comparisons (with tables) of build/buy/hybrid strategies, examining costs (including total cost of ownership), timelines, and outcomes. Over 60 citations from industry, academic, and practitioner sources are embedded throughout. Finally, we discuss implications for organizational outcomes and future trends (e.g. generative-AI-powered learning, immersive training), to guide decision-makers exploring corporate AI training programs in 2026 and beyond.

Introduction

Background and Context. Corporate learning and development (L&D) has evolved dramatically over the past decade. Once seen as ancillary, L&D is now recognized as a strategic driver of business performance. Globally, organizations spend on the order of hundreds of billions of dollars annually on training and upskilling (^[1]) (^[2]). Even during downturns, companies invest roughly $1,200–$1,500 per employee per year on L&D (^[1]). A COVID-era pivot to remote work and eLearning saw nearly 98% of firms adopt online training by 2023 (^[20]).

The AI Imperative. The rise of AI has intensified this focus. Widespread generative AI tools (like ChatGPT since late 2022) and increasing AI adoption in business have created an urgent “AI skills gap”. For example, the Adecco/A.I. World Economic Forum reported that in late 2023 about 70% of workers were already using AI tools at work (^[21]), and 62% expected a positive impact on their jobs. Employees want training: Adecco’s CEO noted “a majority [of employees] want their organizations to train them” in AI (^[22]). Meanwhile, corporate leaders see demand outpacing supply: A February 2026 Josh Bersin study found companies spend $400 billion on learning and development, yet 74% admit they cannot meet their demand for new skills (^[23]). Bersin concludes that “our entire approach, philosophies, tech stack and operating models for learning are out of date” in the AI era (^[24]).

This training gap has real consequences. Surveys show a critical disconnect: 81% of employees think employers should provide AI upskilling, but 83% of CEOs think workers should fend for themselves (^[25]). Notably, two in three workers reported they would avoid applying for a job if an employer offered inadequate AI training (^[26]). Meanwhile, a Boston Consulting Group analysis notes only about one-third of employees have been properly trained in AI, even though training sharply raises usage and confidence (^[27]). Companies that invest in AI training reap productivity gains: BCG reports that early adopters of AI workflows see employees save significantly more time and focus on strategic work (^[28]). Mercifully, AI offers solutions: generative AI can speed content creation by orders of magnitude, creating personalized learning in minutes (^[29]) (^[30]).

Given this context, corporations face a critical decision: how to build AI capability among their workforces. Traditional L&D must adapt: are new AI training programs built from scratch by internal teams, outsourced or licensed from vendors, or some combination? This “build vs. buy vs. hybrid” debate is well known in technology strategy (e.g. build an internal system vs. buy software), but it is only now emerging in the specialized context of corporate AI training.

The goal of this report is a thorough analysis of these approaches to corporate AI training in 2026. We review market trends and data on AI-skills needs, outline the characteristics of each approach, present evidence and case examples of successes and challenges, and compare the trade-offs with quantitative and qualitative arguments. We also consider the broader organizational implications, including costs and cultural fit. Finally, we look ahead to future directions (e.g. AI-driven adaptive platforms, immersive simulations, skill taxonomies) that will shape how companies educate their employees in AI over the next decade.

Build, Buy, and Hybrid: Defining the Options

We begin by defining the three archetypes of corporate AI training programs:

• Build (In-House Development): The company designs and develops its AI training program internally. This might involve a dedicated team of instructional designers, data scientists, and subject-matter experts creating courses, workshops, or learning platforms tailored to the organization’s context. Role-based learning paths can be created for different employee categories (e.g. AI Users, Builders, Leaders as in Lloyds’s program (^[31])). Building in-house often means high initial investment in curriculum design, learning management systems, and staff time, but it promises maximum relevance and control.

• Buy (Third-Party Solutions): The company purchases or licenses training content and services from external providers. This can include subscribing to online learning platforms (e.g. Coursera, Udemy for Business, LinkedIn Learning, Pluralsight, etc.), purchasing specialized AI courses or certifications (like CompTIA’s new AI Essentials (^[32])), or hiring L&D consultants/vendors to deliver training. Buying leverages existing expertise and content, and enables rapid deployment, but can lack customization to the company’s unique needs.

• Hybrid (Blend of Build and Buy): A mixed approach combining internal and external elements. For example, a company may build core, proprietary modules in-house while outsourcing standard modules (e.g. basic AI literacy, compliance training) to vendors (^[18]). Often the “build” component focuses on company-specific use cases and strategy, while the “buy” component covers generic skills or supplements internal efforts. Hybrid aims to balance flexibility and speed, but requires managing multiple sources of content.

Each approach has distinct pros and cons (see comparison in Table 1). In practice, companies may lean toward one approach as a baseline and then incorporate elements of the others. The remainder of this report examines each approach in depth, presenting multiple perspectives and evidence.

Building In-House AI Training Programs

Developing an internal AI training program is resource-intensive but can yield high fidelity to business needs and culture. We examine what building in-house entails, its benefits and pitfalls, and illustrative examples.

Characteristics of In-House Development

Building an AI training program internally typically involves:

• Establishing an AI Training Team: This team may include instructional designers, AI experts/data scientists, developers (for any custom platform), and project managers. Roles might span content creators, course developers, and AI-savvy trainers. For example, a technology firm might repurpose data scientists to co-create tutorials on their tools.

• Defining Curriculum and Learning Paths: The program must specify what skills to teach. Common topics include AI fundamentals (ML concepts, LLM usage), responsible AI ethics, prompt engineering, and role-based applications (e.g. marketer’s use of AI tools vs. engineer’s). Companies may develop tiered “credentials” (e.g. AI User, AI Builder, AI Leader (^[31])). A Harvard-style taxonomy of skills may power modular lessons.

• Using or Building a Learning Platform: Often, the same Learning Management System (LMS) or enterprise portal used for general training is extended. Some companies create dedicated portals (e.g. Lloyds’s AI Academy platform (^[33])). Content delivery could be through a custom-built platform or an existing LMS.

• Content Development: Internal SMEs produce the actual training materials: videos, interactive modules, quizzes, documentation. Generative AI tools can accelerate this (e.g. AI can draft an initial course outline or generate quizzes (^[30])). Some organizations also “curate” third-party content, integrating publicly available tutorials into internal programs.

• Integration and Updates: As AI tools evolve, content must be updated. In-house teams can react quickly to new models or tools without vendor delays (^[7]). For example, Lloyds quickly incorporated an AI-responsibility module for its entire workforce (^[10]).

Case Study: Lloyds Banking Group (Build Approach)

A well-documented example of the build approach is Lloyds Banking Group (UK). In January 2026, Lloyds announced an internal AI Academy to train all 67,000 employees (^[9]). Key features of Lloyds’s build strategy include:

• Company-wide Rollout: Every colleague (“no matter their role or level”) will participate, starting with a mandatory brief module on “Working with AI Responsibly” (^[10]). This ensures foundational knowledge across the board.

• Tailored Learning Paths: The Academy offers content for different cohorts: AI Users, Leaders, Builders and Enablers (^[31]) (^[34]). This role-based framing allows customizing depth – e.g. executives take strategic courses while developers get technical modules.

• Mixed Media: The program uses interactive eLearning, podcasts, articles, and community forums (^[35]) (^[34]). This follows best practice (microlearning + varied formats) and leverages an internal platform.

• Partnerships for Specialized Courses: For senior leaders, Lloyds cooperated with Cambridge Spark to create an 80-hour “Leading with AI” course (^[36]). This indicates Lloyds did not build everything internally, but partnered for higher-level leadership training.

Outcomes: Lloyds anticipates a major skills transformation: by 2030 it expects ~60% of roles will need reskilling due to AI (^[37]). The bank’s executives emphasize that by investing in its people’s AI skills, operations can scale responsibly and innovate faster (^[38]) (^[39]). Notably, Lloyds set an aggressive goal: “100% of colleagues upskilled by end of 2026” (^[40]), highlighting its commitment to the build approach.

This case exemplifies the pros of building: complete control and strategic alignment. Lloyds retained IP over its curriculum (e.g. its AI ethics framework (^[38])), ensured content matched its culture and systems, and could embed training into broader transformation efforts (^[39]). However, it also illustrates common challenges: such an undertaking requires major coordination and resources over months to years. Lloyds had to staff up instructional design and partner for content, and carve out employee time for training across the organization.

Advantages of Building In-House

Building allows organizations to tailor programs precisely. Key advantages include:

• Customization and Relevance: In-house courses can incorporate proprietary data, tools, and workflows (^[6]). For example, a software firm can teach AI via its own codebase and data pipelines. This alignment often boosts learner engagement; when employees see that “training is closely related to their job,” learning sticks better (^[41]).

• Cultural Alignment: Training can reinforce the company’s values and language. The LMSPortal analysis notes that in-house content ensures messaging aligns with company culture (^[42]). Employees may take training more seriously if it clearly ties into internal goals.

• Control Over Content and Updates: An internal L&D team can rapidly update materials. For instance, when a new AI regulation or tool emerges, the team can revise modules immediately (^[7]). There’s no waiting for external publishers to refresh content.

• Intellectual Property (IP) Ownership: All created materials (slides, frameworks, case studies) remain company IP (^[11]). This allows reuse, adaptation, or even external licensing in the future. IP ownership avoids recurring licensing costs.

• Deep Engagement: Employees involved in building or curating content can become “AI champions” internally, generating momentum. The process of involving cross-functional teams in content creation can itself be a learning exercise.

These benefits come with evidence: Learning Elements’ case on St George Bank shows blending in-house core modules (“internal ‘core’ modules”) with outsourced content improved learner commitment (^[18]). This suggests high relevance (from in-house topics) drives engagement.

Challenges and Risks of Building

However, building also has serious challenges (^[6]) (^[8]):

• Resource Intensiveness: Creating a full training program “requires time, money, and expertise” (^[6]). Companies must hire or repurpose instructional designers, graphic designers, and subject experts. According to LMSPortal, lacking these, in-house courses often result in “glorified PowerPoint” with low engagement. Typical problems include limited design skills, poor visuals, and minimal user testing (^[43]).

• High Upfront Costs: As an analysis shows, building proprietary AI training content and platform can cost in the millions in Year 1 (^[12]) (e.g. $2.5–$4.8M just for initial setup in an AI solution context). Even for training, expenses accrue from staff time, possibly IT infrastructure, and development tools.

• Longer Time-to-Deploy: Developing internally often takes much longer than buying. An enterprise AI TCO study found building an AI solution takes ~12–24 months (^[12]), whereas buying (licensing) is faster. For training programs, similar lags occur as courses must be scripted, reviewed, piloted, and integrated.

• Scalability Issues: In-house programs may struggle to scale across large or global organizations (^[6]). Extending specialized content to thousands of employees (possibly in multiple languages) requires ongoing effort.

• Expertise Gaps: Not all companies have strong L&D departments. Internal teams may lack the latest instructional design methods or technology skills (^[8]). Fresh pedagogical approaches (e.g. microlearning, mobile learning) may be underutilized. Stagnation in methodology leads to low knowledge retention (only ~8–10% for passive training vs ~25–40% for interactive learning (^[44])).

• Maintenance and Refresh: AI is a fast-moving field; content becomes stale quickly. If in-house resources are diverted elsewhere, courses may not be updated, eroding ROI.

These risks suggest that in-house build is most viable when an organization has sufficient budget and expects the content will need regular updates and tight integration with business. It favors companies viewing AI skill development as a long-term strategic capability (their “competitive moat” (^[45])). Smaller firms or those with limited L&D staff may find pure build daunting.

Cost and Resource Considerations

Analyzing costs is complex. A broad framework from enterprise AI shows pure “build” incurs heavy initial investment (^[12]). Translated to training, one can think of:

• Year-1 Investment: Building may require hiring or assigning several instructional designers/specialists (their salaries), acquiring authoring tools (e.g. Articulate (^[46])), and perhaps setting up IT infrastructure (LMS licenses, video recording equipment). Even if off-the-shelf tools are used, project management and SMEs’ time cost thousands per week. For example, a hypothetical in-house AI lab might spend $2–3M in the first year on people and systems (^[47]).

• Ongoing Costs: Unlike a subscription, most cost is front-loaded. After launch, maintenance might drop, involving occasional updates and hosting. However, hidden costs persist: talent retention premiums (keeping skilled trainers may require salary bumps of 15–25%) (^[48]), and continuous training for the L&D team itself (e.g. learning new AI tools).

• Time Costs: The long development cycle can delay impact. If business needs change in the interim, the program may have to restart design. In contrast, buying starts showing results immediately.

One rule of thumb from similar build vs. buy analyses is the breakeven point usually comes after a couple of years (^[17]). While a subscription model scales costs linearly with usage, an internal build’s cost curve flattens out. For large-scale programs, building can become more cost-efficient over time. For instance, one estimate suggests that for 10,000 users, a 5-year subscription can cost ~$1.6M versus ~$250K for a custom platform (^[16]). (Admittedly, training content differs from platform, but the principle holds: SaaS content fees accumulate with user count, whereas owned content is mostly fixed cost.)

In summary, building in-house offers high customization and control, at the price of greater complexity, cost, and lead time (^[6]) (^[6]). It works best where content truly needs to be bespoke or proprietary, and where an organization can commit a major investment to get it right.

Buying Third-Party AI Training Solutions

The “buy” approach means leveraging external providers for AI training. This spans vendor courses, online platforms, certifications, and partnerships.

Forms of External Training

Companies can purchase training content or services in several ways:

• Online Course Marketplaces: Platforms like Coursera for Business, Udemy Business, LinkedIn Learning, Pluralsight, etc., offer libraries of courses. Businesses subscribe and assign modules to employees. Many have scrambled to add AI tracks: for example, in late 2025 Coursera and Udemy merged into a $2.5B platform aimed at addressing an anticipated AI reskilling wave (^[49]). These services often cover AI fundamentals, Python for ML, and tool-specific courses.

• Industry Certifications: Organizations can have employees pursue certifications (e.g. AWS or Azure AI certifications, Google Cloud ML Engineer). They might reimburse these or integrate them into development plans.

• Professional Training Firms: Companies like Udacity, Firebrand, or general L&D consultancies provide cohort-based AI bootcamps or customized workshops. For example, Udacity’s “AI for Business” nanodegree or specialized institutes (e.g. OpenAI’s own training programs) could be engaged.

• Vendor Partnerships: Technology vendors sometimes offer training as part of enterprise deals. For instance, Microsoft has education initiatives (Project Amplify, 365 AI skilling) and AWS Train (free online labs).

• Public/Nonprofit Initiatives: Some large-scale programs (e.g. Cisco’s pledge) are open to corporate workers, effectively free “buy” resources. Corporations can encourage staff to use these resources.

• AI Tools Training: Separately, vendors that sell AI tools often include training. A company buying a data science tool might also purchase a training package.

Data on adoption of such solutions is growing. A Deloitte survey (2024) found 34% of companies had implemented AI in their training programs, with another 32% planning to within two years (^[50]). The VirtualSpeech statistics (Jan 2026) highlight this momentum: adoption of AI-powered training is substantial, with companies planning to increase AI investments in training over 90% (^[51]). These figures imply a strong market pull for buy-side solutions.

Examples of Buy-Side Initiatives

• CompTIA (Dec 2025): The IT industry association launched an “AI Essentials” course for non-technical workers (^[32]). This reflects a buy approach: a vendor providing upskilling content aimed at bridging the skill gap. CompTIA cites its own research showing only 34% of companies currently require AI skills training (^[52]), motivating their offering.

• Cisco’s AI Pledge (Sep 2025): Cisco committed to train one million Americans in AI/digital skills via its Networking Academy (^[15]). Though public, this program is effectively a third-party resource companies can tap (e.g. through hiring trained talent, or encouraging employees to sign up). It shows how large vendors are positioning themselves as education providers for AI.

• Databricks (Oct 2025): Databricks funded a plan to train 100,000 people in the UK & Ireland in data/AI skills (^[53]). Their Free Education portal and $10M funding illustrate an enterprise “build-buy hybrid” strategy: an AI company building its own platform but offering it to external learners, partly supplementing third-party coordination (with government and universities) (^[53]) (^[54]).

• Coursera & Udemy Merger (Dec 2025): As noted, the merger of Coursera and Udemy underscores the scale of the reskilling need (^[49]). It is a buildup of external training capacity; for companies, this merged platform represents an attractive “one-stop shop” for both academic-style courses and marketplace content.

• Other Corporate Programs: Many multinationals have publicly announced AI upskilling initiatives, though the exact approach often blends build and buy. For instance, Google’s AI Works for America (2025, with U.S. government) is a partnership to train workers in AI, largely via Google’s channels (^[55]). Walmart announced plans for a new AI training collaboration (with OpenAI) for its workforce (^[56]), which would use external expertise. These highlight “buy” elements: leveraging external platforms and content to train employees, rather than each firm building the course from scratch.

In all these examples, companies turn to outside expertise or platforms to rapidly reach large audiences. Third-party solutions are especially appealing for standardized skills (e.g. general AI literacy, prompt engineering, ethics) or compliance topics (e.g. data privacy) that are similar across firms (^[13]) (^[8]).

Advantages of Buying Third-Party

Key benefits of the buy approach include:

• Speed and Convenience: Ready-made content means organizations can deploy training quickly (^[13]). There is no need to create materials from scratch, reducing lag time. For example, companies can assign an external AI course to employees within days. This is critical when skill gaps are urgent.

• Access to Expertise: External vendors often employ experienced instructional designers and subject-matter experts. A third-party course may include insights from multiple clients or industries, raising overall quality. For instance, learning firm content is typically vetted and tested across different users. As LMSPortals notes, vendors have “extensive expertise” and can deliver high-quality, best-practice content (^[57]).

• Scalability: Many vendor platforms are built to handle thousands of users concurrently. A SaaS LMS allows easy enrollment of large global teams. In contrast to in-house efforts that may bottleneck, buy-side solutions often “offer seamless scalability” for large or distributed workforces (^[8]).

• Ongoing Support and Updates: Vendors usually maintain and update courses. For example, compliance modules are refreshed for new regulations. This means buyers need not continuously revise material themselves. Third-party providers may also supply user analytics, certifications, and customer support.

• Lower Initial Cost: Especially for smaller companies or limited budgets, buying sidesteps large upfront build costs. Subscription or per-seat pricing means companies pay as they go. Providers can spread development costs across many customers.

• Focus on Core Business: By outsourcing training creation, the organization can focus L&D resources on strategy and implementation rather than content development.

From a strategic viewpoint, buying is often the “safe” default for general skills. LMSPortals notes that for common topics (e.g. leadership, basic AI tools, compliance), third-party solutions usually suffice (^[8]). Many corporations already have relationships with vendors (e.g. roll-out of industry certifications or popular platforms). These can be extended into AI curricula.

Challenges and Risks of Buying Third-Party

Despite these advantages, the buy path has drawbacks:

• Limited Customization: Off-the-shelf content is designed to be generic. It may miss nuances of a company’s products, data, or processes. LMSPortal warns that third-party courses “may not fully align” with an organization’s unique workflows (^[6]), potentially leaving employees less prepared for company-specific scenarios. In some cases, customization is offered by vendors at extra cost, but that reduces the simplicity.

• Loss of Control: When outsourcing, the organization cedes control over content quality, depth, and updates (^[58]). A vendor might change or remove content (such as dropping a module), leaving the company with outdated training. This dependence can be problematic if a vendor deprecates a product or goes out of business.

• Dependency and Contract Risk: There is vendor lock-in risk. If the provider fails to meet service level requirements, the company may have difficulty migrating content elsewhere. In the AI boom, consolidation (e.g. Coursera/Udemy merger) can alter the competitive landscape. Additionally, license renewals can be a budget risk: many SaaS products increase fees yearly. As one TCO analysis notes, depending on usage, SaaS costs can spiral (with usage overage fees or price hikes) (^[59]).

• Recurring Costs at Scale: While initially cheaper, vendor costs can add up. For example, many platforms charge per user. With thousands of learners, the subscription model can surpass build costs over time (^[16]) (^[17]). Also, some fees like certification exam costs or platform integration charges may apply.

• User Engagement: Without internal advocacy, external content can feel impersonal. Employees might not engage as fully if a course feels “bought in” without context. Engagement may suffer if learners see the training as generic.

• Quality Variability: Not all content providers are equal. The burden is on the buyer to vet the quality and pedagogy of courses. Poorly designed eLearning modules (e.g. “click-next” courses) are common and result in low retention (^[43]).

In summary, buying third-party courses is fast and scalable, but may not address all specific needs and incurs ongoing licensing commitments. It is well-suited to covering baseline AI literacy and general topics, especially when time-to-market is critical (^[13]) (^[57]).

Cost and Considerations for Buy

Typical cost factors for buying include:

• Subscription/Licensing: Platforms often charge per user or per year. For example, an LMS might be $50–$100 per seat annually; with 10,000 employees this could be $500K–$1M per year. As noted earlier, vendor TCO analyses show that SaaS pricing scales with headcount, potentially reaching $1.6M over 5 years for 10k users (^[16]).

• Onboarding/Integration: Initial setup fees may include customizing the learning portal, integrating with internal systems (HRIS, identity management, etc.), and migrating users. Though mostly one-time, these can be substantial (e.g. tens of thousands of dollars).

• Content Bundling: Some providers offer enterprise bundles. The company should ensure the bundle includes relevant AI content. If not, there may be per-course purchases.

• Support and Training: The organization may need to train its trainers or HR team to administer the vendor platform and create learner engagement strategies.

• Cost Efficiency: Buying can be cost-effective in the short term, but organizations should project multi-year costs. For example, if a startup buys an AI curriculum for $100K and scales to 5,000 employees quickly, switching to an internal solution later might pay off.

Given these points, financial analysis often compares first-year costs (low for buy) versus 3–5 year totals (higher if usage grows). The decision often incorporates a Total Cost of Ownership perspective: third-party solutions avoid heavy capital spend but create sustained operational expenses (^[60]) (^[16]).

Hybrid Strategies: Blending In-House and External Training

Recognizing the upsides and downsides of both extremes, many organizations adopt hybrid models that combine build and buy. A hybrid approach can take various forms, but the core idea is to use internal resources where differentiation matters, and external solutions for general needs.

What Hybrid Looks Like

A hybrid AI training program may involve:

• Core In-House Modules: The company builds certain modules or learning tracks internally. These might cover proprietary tools, internal data practices, company-specific AI applications, or strategic objectives. For example, a retailer might internally develop a training module on its AI inventory management system.

• Outsourced Specialized Content: At the same time, the company sources vendor content for standard topics. This could include courses on general AI fundamentals, instructor-led seminars from a training vendor, or ready-made eLearning on data privacy compliance. By outsourcing these parts, the company saves time and leverages external breadth.

• Partner-Developed Programs: Sometimes hybrid means partnering with external experts selectively. For instance, Lloyds used Cambridge Spark for its executive course (^[36]) (an external partnership), even as it ran its own Academy for the wider workforce.

• Mixing Modalities: Hybrid can also refer to blending formats. A company might host an internal “AI bootcamp” build with its HR team, but supplement it with weekly vendor webinars or access to an external MOOC. The training path is stitched together from in-house events and bought content.

The St. George Bank case study from Australia is illustrative (^[61]). In their hybrid model:

• Internal “Core” Modules: The bank created fundamental training aligning with its business needs.
• Outsourced Content for Special Topics: They purchased external content for highly specialized or regulatory topics. This blend yielded a “robust business case” that met both cost and engagement goals (^[61]). It maintained brand consistency (internal materials) and leveraged vendor expertise where needed.

Advantages of Hybrid

A hybrid strategy can capture the advantages of both building and buying:

• Balanced Customization: Organizations can prioritize which content must be custom and which can be generic. This leverages in-house strengths while avoiding unnecessary work on standard topics. For example, a tech firm might build all content related to its bespoke AI platform internally, but buy vendor training for open-source ML tools.

• Faster Implementation on Non-Core Topics: By not building every course, companies can roll out parts of the program quickly. Critical modules that require speed to market (e.g. basic AI literacy for all staff) can be implemented via off-the-shelf resources.

• Cost Efficiency in Scope: Hybrid can reduce costs: the company only invests in developing core materials (rather than all training). Outsourcing commoditized portions (like compliance training) avoids duplication. In the St. George example, combining internal and external content “helped scale e-learning effectively across teams” (^[62]).

• Continuous Improvement: Internal and external content can complement each other. Feedback from vendor courses can inform future internal development. Conversely, successful internal modules might be open to outside audiences.

• Risk Mitigation: If one approach falters, the other can fill in. For instance, if internal content is delayed, vendor modules cover immediate needs. If a vendor is acquired or discontinues a course, the company still has internal core materials.

Empirical practice suggests many organizations naturally evolve hybrid strategies. LMSPortal notes that a hybrid approach “offers the best of both worlds” (^[63]). Indeed, Learning Elements’ analysis concluded hybrid architectures (like St. George’s) can achieve “stronger learner commitment” while keeping costs in check (^[61]).

Challenges of Hybrid

Hybrid requires coordination and strategic oversight. Key challenges include:

• Complex Management: Juggling both in-house development and vendor relations can strain resources. It demands clear governance: who owns the learning roadmap, how modules interconnect, and how to ensure consistency in quality.

• Integration and Interoperability: If internal and external materials are delivered on different platforms, ensuring a seamless learner experience can be tricky. Learners might have to log into multiple systems. Data integration (e.g. tracking completion across systems) is a technical challenge.

• Content Gaps or Overlaps: A hybrid program must carefully avoid duplicating training or leaving gaps. For example, if a vendor course covers AI ethics and the company also builds one, the redundancy can waste effort; conversely, if each assumes the other covers certain basics, knowledge gaps can occur.

• Alignment of Quality and Style: The tone and pedagogy of internal vs. external content may differ. This can confuse learners. Ensuring a consistent “look and feel” requires oversight.

• Budget Splitting: Hybrid budgets need to account for dual channels. If not well planned, cost savings from vendor use could be lost in maintaining two streams of development.

These issues can be managed with clear planning. Decision frameworks (see next section) emphasize careful needs assessment to decide which elements to build or buy (^[8]). In practice, hybrid models often phase in over time: start with a core content area to build internally, while subscribing to existing platforms for general skills.

Examples of Hybrid in Practice

Beyond St. George Bank, many organizations effectively use hybrid strategies (even if not explicitly labeled). For example:

• Lloyds Banking Group: Although Lloyds built its AI Academy, it partnered externally for specialized courses (^[36]), pairing in-house training with vendor expertise. This is hybrid in spirit.

• Large Consultancies: Firms like Accenture or Deloitte often develop proprietary AI training videos and also license third-party certifications.

• Technology Corporations: Tech firms frequently appoint internal “AI ambassadors” (build) while also providing access to vendor training (Microsoft Learn, Google Skillshop). Many companies encourage employees to earn external AI badges.

The common thread: hybrid programs tailor the mix to organizational needs. Research supports this: many organizations find “the most practical path is not choosing strictly between buying and building, but combining both” (^[64]). (Indeed, Gartner also advises that a mix will likely make sense (^[65]).)

Comparing Approaches: Pros and Cons

The choice between build, buy, and hybrid hinges on multiple factors. Table 1 below summarizes the key trade-offs across dimensions commonly considered by L&D leaders. (Subsequent sections and references delve into each factor in detail.)

Each organization will weigh these factors according to its needs. Often, short-term urgency or limited budget push for buying, while strategic differentiation or long-term cost optimization justify building. Hybrid offers a compromise by splitting the difference. The following section outlines a decision framework to guide this choice.

Decision Factors and Frameworks

Choosing the right approach requires careful assessment of context, resources, and goals. Key factors to consider include:

• Budget and Resources: If funds and skilled personnel are plentiful, building in-house becomes feasible (^[6]). Conversely, with constrained budgets, buying may be the only viable option initially. An organization with deep pockets might amortize CDN experts over time by building its own training, whereas a lean startup may rely fully on purchasing courses.

• Time Sensitivity: How fast is training needed? When the timeline is tight (e.g. a critical AI rollout next quarter), off-the-shelf solutions or third-party workshops can be implemented quickly (^[13]) (^[57]). In contrast, if leadership can afford a longer ramp-up and wants highly tailored training, they may choose to invest time in building a program (^[66]).

• Content Scope & Complexity: If the AI skills required are commonplace (e.g. using AI chatbots or basic data analysis), external courses likely exist to cover them. However, for specialized domains (proprietary ML models, internal algorithms, or industry-specific regulations), in-house development ensures the curriculum matches unique needs (^[8]). For highly regulated industries, custom compliance training may be necessary.

• Company Size & Structure: Large, distributed organizations often favor scalable external programs (^[67]). Smaller companies or those with tight-knit teams may find it easier to train in-person or with internal seminars. International firms must consider language/localization – working with global vendors might be easier than translating all content themselves.

• Existing Infrastructure: Companies already using a learning platform (LMS/LXP) should assess whether it can host AI training content. If an LMS is in place, building content for it (and supplementing with purchased modules) is straightforward. Otherwise, buying might entail also procuring a new platform.

• Desired Outcomes: If the goal is broad AI literacy across all staff, external programs can efficiently reach thousands. If the goal is developing internal AI leaders or innovators, custom “academy” style programs might align with that outcome.

• Risk Tolerance: Buying introduces vendor risk (e.g. what if a provider obtains new knowledge unavailable in bought content?). Building carries execution risk (e.g. lack of expertise leads to poor quality training). Hybrid can mitigate some risk by diversifying sources.

As one Gartner analysis advises, many organizations “implement the best AI strategy for [their] needs — and know that some combination of sources will likely make sense” (^[65]). In practice, a stepwise approach often works best: start with a clear needs assessment, possibly pilot a small in-house program for core skills, while immediately adopting proven external content for general skills. Over time, adjust the balance based on feedback and metrics.

Illustrative Example: An organization might ask: “Do we have internal experts to create courses on our AI tools? If not, buying those makes sense. Do we need training on widely-used platforms like ChatGPT? If so, a vendor course could be fine.” This thought process is supported by experts: “if training needs are standard, buy; if highly specialized, build” (^[8]).

Data Analysis and Market Trends

A thorough decision must also consider broader L&D and AI trends. Below we examine relevant data on market size, technology adoption, and training outcomes.

Corporate Learning Market and AI Adoption

• Market Scale and Growth: The corporate L&D market is large and growing. Bersin (Dec 2023) estimated it at $340 billion (^[1]). By early 2026 that estimate was up to $400+ billion (^[2]). The eLearning industry projected to reach $463B by 2027 (^[20]). AI is a key driver: virtualspeech (2026) notes the global AI in education market was $254.5B in 2025 and forecast to reach $1.68T by 2031 (^[68]).

• E-Learning Penetration: Nearly all large companies now use digital training: surveys found ~98% of firms had eLearning programs by 2023 (^[20]). Training budgets are robust, with $1.2k–$1.5k spent per employee (^[1]). These figures imply companies are primed to invest in advanced training like AI, provided it aligns with ROI.

• AI Training Adoption: Corporate adoption of AI in L&D is accelerating. According to a Deloitte report, 34% of companies already use AI-enhanced training, and another 32% plan to do so soon (^[50]). Continu.com (2025) lists that by 2023 virtually all companies planned eLearning, and many are now adding AI elements. In practice, early data suggests companies that leverage AI in training see measurable benefits: IBM cited a ~20% increase in productivity from AI-driven upskilling programs (^[51]). Generative AI specifically is driving change – Bersin describes a “learning revolution” where AI platforms can auto-generate courses and content on demand (^[69]) (^[30]).

Effectiveness and Outcomes

Several studies quantify the impact of effective training / AI training:

• engagement and retention: Trusted sources report large improvements: AI-driven personalization can boost learning engagement by ~30% and learning outcomes by ~25% (^[70]). Companies using AI training saw similar 20% productivity gains (^[51]). One analysis projects that by 2025, AI-powered learning will increase engagement 20% and retention 15% (^[71]).

• Retention Rates: More broadly, corporate eLearning shows higher retention than traditional: Digital courses can achieve 25-60% recall vs just 8-10% in classrooms (^[44]). Personalized AI tutors may push this even higher.

• Upskilling outcomes: Frontline employees with 5+ hours of AI training are significantly more likely to use AI effectively (^[27]). But only ~1/3 of workers report having that training, indicating significant unmet potential.

• Employee attitudes: Studies show people value AI skills. For example, 74% of workers in a Davos/Adecco study expected positive impacts from AI on their jobs (^[72]). Lack of training can threaten retention – Emergn found 66% of workers would avoid a job without AI training (^[26]).

• Leadership Buy-In: BCG finds only ~25% of employees feel strong leadership support for AI usage (^[73]). Yet companies where leaders actively backtraining see much higher adoption. This underscores that any training program (build or buy) must have executive support to be effective.

Future Skills Lifecycles

AI’s fast pace means the target skills are moving. WEF/Adecco reports that by 2030 “nearly 60% of the workforce will need to reskill” as their skills evolve (^[74]). This implies AI training programs must be ongoing, not one-off. The skills taxonomy itself must adapt (e.g. new roles like “prompt engineer” emerged in late 2023). Successful programs will therefore include mechanisms for continuous learning (AI assistants, microlearning updates, community platforms) rather than static courses.

Summary of Data Insights

The data underscores both urgency and opportunity. The corporate training market is gargantuan and increasingly digitized (^[20]). AI adoption is already widespread and set to rise (^[50]) (^[71]). Employees expect training (70% using AI already (^[21])), and evidence shows AI training works (boosting engagement, productivity (^[75])). Conversely, failure to train leads to skill gaps that impede adoption (only 6% making progress (^[3])).

In this environment, designing an effective AI training strategy—whether built, bought, or hybrid—becomes a critical leadership task. The remainder of this report applies these trends to real choices, comparing scenarios and drawing lessons from corporate practice.

Case Studies and Real-World Examples

We now survey real organizations’ approaches, beyond the Lloyds and St. George examples already discussed, to ground our analysis in evidence.

Lloyds Bank (Build with Hybrid Elements)

• Approach: Primarily build. Established AI Academy (in-house learning platform) for all 67,000 staff (^[9]). Also partnered externally for leadership modules (Cambridge Spark’s “Leading with AI”) (^[36]).

• Scale & Outcome: Aiming for 100% of colleagues upskilled by end-2026 (^[40]). Early stages: 110 senior leaders completed the 80h leadership course by late 2025 (^[76]). The bank projects that ~60% of roles will change by 2030, and training is central to handling this (^[37]). Attesting to impact, Lloyds has already deployed 50+ generative AI use-cases internally, partly thanks to upskilled teams (^[77]).

• Considerations: Lloyds’s case demonstrates that with sufficient scale and commitment, a build strategy can create broad upskilling. The bank’s “responsible AI” principles and existing AI project pipeline required tight alignment of training (e.g. mandatory ethics module (^[31])). The high internal control allowed Lloyds to integrate training with its broader digital transformation strategy.

Walmart (Buy/Hybrid with Vendor Partnership)

• Approach: Hybrid/Buy. Walmart has existing internal training programs for frontline roles, but for AI specifically it is turning to external partners. In September 2025, Walmart announced it would offer an AI skills program via collaboration with OpenAI (^[56]). This suggests a buy-oriented approach (using OpenAI resources and potentially ChatGPT-like tech) rather than building its own curriculum from scratch.

• Scale & Context: Walmart has 2.1 million employees worldwide. It plans to focus on making AI improve everyday tasks (e.g. using ChatGPT to save store associates time) (^[78]). CEO Doug McMillon indicated interest in “plussing up” every role with AI tools (^[79]). The exact model is to offer training next year (2026) presumably through OpenAI’s channels, likely covering how to use generative AI safely and effectively in retail contexts.

• Considerations: Walmart’s path shows a pragmatic hybrid: leveraging external tech and expertise to train a huge, diverse workforce. By partnering with OpenAI, Walmart avoids the heavy lift of content creation and gains access to cutting-edge AI developers. However, it must manage the challenge of tailoring such training to shop-floor employees (who may have low time for courses) and integrating it into its culture. The approach reflects that very large employers may lean on external vendors to quickly reach many people (echoing buy benefits (^[13])), while internal training for critical roles (like leadership) may still be developed.

Cisco Networking Academy (Buy/External)

• Approach: Buy/Public Program. Cisco pledged (Sep 2025) to train 1 million Americans in AI and digital skills over 4 years (^[15]). Its “Learn with Cisco” platform is an evolution of its long-running Networking Academy. This program is free and open to the public (especially youth and educators), but corporates can encourage employees to participate.

• Scale & Context: This is purely external, (similar to Walmart’s but aimed at education/public workforce). It indicates corporate social responsibility, but it also builds a pipeline of trained potential employees. Other tech giants (Google, Amazon, Microsoft) are also investing hundreds of millions in AI training initiatives (^[80]).

• Considerations: For corporations, such public programs serve as an external talent-developing pipeline (effectively “hiring ready” staff). While not an internal build, Cisco’s commit highlights that buying into large coalition efforts is an option for companies wanting workforce upskilling without building a program themselves.

Databricks Free Training (Buy)

• Approach: Buy/Own (Vendor). Databricks, a data and AI platform vendor, announced (Oct 2025) it would train 100,000 people in UK/Ireland in AI/data via its Databricks Free Edition portal, backed by $10 million funding (^[81]). Although run by Databricks, this qualifies as third-party content from the perspective of a corporation: a bank or retailer could send employees to this free program.

• Scale & Context: The curriculum covers generative AI, data engineering, ML and analytics (^[81]). Databricks partners with universities and government programs to roll it out (including a UK public training scheme). The program complements, rather than replaces, internal training.

• Considerations: Databricks’s move exemplifies vendors expanding into education (selling not just tools, but related training). It is a multi-stakeholder “buy” resource that companies can utilize. It underscores that content vendors (like Databricks) will increasingly serve as training providers, offering self-paced courses aligned to their products. Corporations buying its platform might leverage this free training as part of an upskill strategy.

Udemy/Coursera Merger (Market Context)

• While not a training program, the Dec 2025 Coursera-Udemy merger (^[49]) is a market signal. Valued at $2.5B, it consolidates two of the largest corporate learning platforms under one roof, explicitly “anticipating a flood of workers requiring reskilling because of AI” (^[49]). For corporations, this means the buy-side market is consolidating, potentially simplifying vendor relationships but also reducing competition.

• This illustrates the scale of anticipated AI reskilling demand: even the big learning platforms expect a “talent transformation opportunity” of unprecedented size (^[82]). It also suggests that for many companies, buying from this merged entity could be an attractive way to tap into a vast library of AI courses.

Summary of Case Insights

These examples highlight several lessons:

• Big Companies Build: Organizations with the resources and strategic focus (Lloyds) tend to build comprehensive internal programs. They create training that is tightly integrated with their transformation goals, often on a massive scale (67k employees). The in-house build is feasible when training is core to business strategy.

• Partnerships and Hybridization: Even builders often partner for specialized content (Lloyds with Cambridge). Companies like Walmart partner with top AI labs rather than build from scratch, blending internal and external resources.

• Vendors as Trainers: Platform and tech vendors (Databricks, Cisco, others) are moving into the training space, effectively offering ready-made content that companies can adopt. This pushes the boundary of “buy” to include vendor-run academies.

• Forge vs. Forge Aid: The choice can reflect competitive advantage: Lloyds may see its AI upskilling as creating proprietary know-how (build), whereas Walmart peers feel more comfortable outsourcing given their retail focus (buy/hybrid).

Overall, the landscape is dynamic. What suits one company (e.g. build for banking) may differ for another (e.g. buy for retail). Decision-makers should weigh these real-world cases alongside their own organizational profile.

Discussion: Implications and Future Directions

The choice of AI training strategy has broad implications. We discuss organizational impacts and project future trends in corporate AI education.

Talent, Culture, and Organizational Impact

• Warp Speed Reskilling: As Bersin notes, AI readiness is the #1 issue for CHROs (^[83]). Companies that excel at training will better retain and attract talent. Conversely, laggards risk losing high performers to employers who offer robust AI upskilling (the “war for talent” (^[84])). In fact, employees expect their employer to invest in their AI skills (^[25]), and lacking that can drive turnover.

• Leadership and Change Management: Successful training requires leadership sponsorship. BCG found that organizational support is key: employees with strong leadership backing for AI are far more optimistic and engaged (^[85]). Integrating AI training into business strategy ensures it is taken seriously. Training programs should be aligned with real projects to avoid the “training in a vacuum” risk.

• New Roles and Skills Taxonomy: The World Economic Forum and Gartner highlight that AI will alter roles. Besides technical skills, complementary skills (ethics, creativity, adaptability) become important (^[86]). Companies may instate new positions (e.g. AI learning architects, prompt engineering trainers) to maintain training programs and manage AI tools.

• Data-Driven Learning: Future training will leverage data analytics. Knowledge graphs and skill taxonomies (using AI) will map employee skills to learning paths (^[86]). Training outcomes (time saved, productivity gains) can be measured and tied to business metrics. This raises the possibility of continuously “closing the loop” where training iterates based on performance data.

• Ethics and Responsibility: Training programs must include AI ethics and safety, as the Lloyds and others emphasize (^[87]) (^[10]). Ensuring responsible use of AI is itself a training need. Companies often institute AI ethics courses upfront to prevent misuse and build trust.

• Remote/Hybrid Learning: COVID-19 shifted training online; going forward, AI training will rarely be fully in-person. Even internal programs will use digital platforms mixed with virtual labs and peer collaboration (hybrid learning). For instance, an Airmeet analysis notes that AI enables virtual co-learning at scale (^[88]) (^[89]). The hybrid training media itself (videos, simulations, live sessions) is increasingly chosen based on learning efficacy.

Future Trends in AI Training

Looking ahead to the mid/long-term, several emerging developments will shape how corporate AI training is done:

• Generative AI-Driven Content Creation: Generative models will continue to accelerate content generation. As Airmeet and Bersin suggest, AI will author courses, simulations, and even dialogues on the fly (^[30]). This reduces build time enormously. For the build vs buy decision, generative tools might shift the balance: even “build” may become feasible more cheaply, as AI can write initial drafts of courses that humans then refine.

• Adaptive Personalized Learning: The promise of AI is individualized learning. Learners will receive hyper-personalized modules tailored to their history and performance (^[88]). Build vs buy may be less of a binary if both internal and external training systems incorporate AI tutors that adapt in real-time.

• Immersive and Experiential Training: As virtual and augmented reality mature, immersive simulations will be used for critical training scenarios (^[90]). Consider safety or high-stakes training (e.g. reactor operators, pilots) using VR/AR with AI-driven scenarios. Companies may find they build or integrate such experiences with vendors specializing in simulations.

• AI Tutors and Learning Assistants: Chatbot tutors and virtual assistants (agentic AI) will provide 24/7 help to learners (^[91]). Companies may “buy” licenses to enterprise AI tutors (or use open versions). This blurs the line between training content and tool: if employees can query an AI coach at work, formal training becomes part of an ecosystem.

• Microlearning and Workflow Integration: The concept of training embedded in work will grow. AI will push micro-lessons to employees within the flow of tasks (^[92]). For strategy, this implies measuring build vs buy not just by courses delivered, but by ongoing digital experiences integrated into daily software (e.g. a CRM that suggests a one-minute lesson triggered by an action).

• Continuous Skill Upgrading: Instead of one-time training, companies will operate on continuous upskilling models. Subscription to AI tutor services (buy) or on-demand internal content updates (build) will be normal. Hybrid platforms that seamlessly serve both internal and external content libraries might emerge.

• Regulatory and Ethical Imperatives: With AI regulation expanding, training content will also need frequent legal updates. Buying from specialized compliance training firms may be attractive for evolving legal content, while culture-specific aspects still come from in-house.

• Collaborative Ecosystems: Companies may form consortiums or learning collaboratives (e.g. industry alliances) to co-create training content (a form of “hybrid” scaling). Especially for non-competitive training (like general AI readiness), shared development can lower costs. We already see some coalition examples (Cisco’s national pledges).

In summary, the lines between build and buy will blur as AI itself becomes part of the training process. Organizations should stay flexible: what is built today might be partially automated tomorrow. The key will be balancing innovation with strategic alignment – building where true differentiation and intellectual capital lie, and buying where scalability and expertise are paramount.

Conclusion

Corporate AI training programs are at a pivotal juncture in 2026. Organizations acknowledge that AI capability is essential for future competitiveness and employee retention (^[26]) (^[3]). Yet they face resource constraints and strategic questions in designing training initiatives. Our analysis highlights that no one-size-fits-all answer exists to the build vs. buy dilemma.

Build strategies yield tailor-made training closely aligned with business needs (^[6]) (^[41]), but require heavy investment and time. Buy strategies leverage external expertise for speed and scale (^[13]) (^[50]), though at the cost of customization and long-term licensing fees. Hybrid models promise a balance (^[18]) (^[13]), but demand disciplined management of multiple content sources.

The optimal choice depends on factors like urgency, budget, company culture, and the uniqueness of the skills required (^[6]) (^[8]). Companies should analyze their own context using decision frameworks and possibly pilot small programs. Crucially, successful AI upskilling also requires leadership commitment, change management, and integration with real business projects (^[85]) (^[39]).

Data strongly supports the payoffs of effective training. Instances of increased productivity, engagement, and retention (^[75]) (^[71]) indicate that appropriate investment in training yields tangible ROI. Indeed, Josh Bersin warns that struggling to meet AI skill demand renders much training spend wasteful (^[23]). Companies must act now to rectify that gap.

Looking forward, corporate AI training will become increasingly AI-driven. Generative AI, immersive simulations, and intelligent tutoring will reshape how we design and deliver training (^[30]) (^[90]). We anticipate that future programs will merge human expertise with AI tools — for example, an AI engine crafting personalized modules, refined by in-house experts.

In conclusion, whether building, buying, or blending, corporations must thoughtfully align their AI training approach with strategy. The decision is not merely pedagogical but transformational, impacting talent, productivity, and innovation. By leveraging the insights and evidence presented here — and continuously monitoring evolving trends — organizations can build training programs that truly equip their workforces for an AI-driven future.

---

Sources:

A breadth of industry reports, news articles, and studies were cited, including research by Josh Bersin (^[23]) (^[1]), the World Economic Forum / Adecco (^[21]), Emergn/TechRadar (^[25]) (^[26]), Deloitte (^[50]), CompTIA (^[52]), U.K. ITPro/CIO news (^[93]) (^[49]), AP News (^[56]), vendor press releases (^[9]) (^[76]), and various training industry analyses (^[18]) (^[6]). Citations are provided inline by section. These sources collectively elucidate the state of corporate learning and the build-buy tradeoff in AI training as of 2026.