Thought Leadership

Something interesting is happening in the IT services industry. EPAM — approximately 50,000 engineers, roughly $3.8B in annual revenue — is pivoting hard toward consulting and AI orchestration with EPAM DIAL. Thoughtworks is repositioning from "engineering excellence" to "AI-native consulting." Globant is acquiring AI companies and building experience platforms.

These aren't random pivots. They're survival moves. And they tell you something important about where the industry is heading.

If you're a company currently buying engineering hours from an IT services firm — or considering it — what's happening inside these organisations should change how you evaluate your options.

The Billable Hour Problem

IT services firms sell engineering time. $150–$250/hour for experienced developers. The business model is straightforward: more developers × more hours = more revenue. It's a model that has generated hundreds of billions of dollars over the past two decades. And it's now facing a structural threat that no amount of rebranding can fix.

AI is disrupting this model from two directions simultaneously.

First, code generation tools — Claude, GPT-4, GitHub Copilot — are making individual developers 3–10x more productive. A senior engineer who used to write 200 lines of production-quality code per day can now generate, review, and refine significantly more — with AI handling the routine scaffolding while the engineer focuses on architecture and business logic.

Second, AI-native delivery platforms like Xamun's Software Factory use AI to generate 70–80% of boilerplate code — the data access layers, API endpoints, form validation, CRUD operations, and infrastructure scaffolding that have traditionally consumed the majority of engineering hours on any project.

The arithmetic is uncomfortable. If AI handles the routine 70%, what exactly are clients paying $200/hour for? The IT services firms know the answer: the 30% that requires genuine human judgment — architecture decisions, complex business logic, edge case handling, system integration, and quality assurance.

But billing $200/hour for 30% of the work doesn't sustain a $4B revenue model. The maths simply doesn't work. And the firms know it.

The Consulting Pivot

Their response is predictable and, in fairness, rational: move up the value chain. Stop selling engineering hours. Start selling consulting, strategy, and AI orchestration.

EPAM DIAL is an AI orchestration platform — essentially competing with Palantir and management consulting firms simultaneously. It's ambitious. It's also a tacit acknowledgment that pure engineering services are a shrinking market.

Thoughtworks is hiring strategy consultants and repositioning as "AI transformation advisors." The firm that built its reputation on engineering discipline and Agile methodology is now selling strategic advice. The shift is telling.

Globant is building a "digital transformation platform" that layers consulting on top of engineering, acquiring AI companies to fill capability gaps that their organic engineering talent can't address fast enough.

The problem with all three approaches is the same: they're now competing with McKinsey, BCG, and Bain — firms with decades of brand equity in strategy consulting. Boardrooms that wouldn't hesitate to hire EPAM for a Java migration will pause before hiring them for business strategy. The brand permission isn't there yet.

And there's a deeper structural issue. These firms are still fundamentally organised around billable hours. Pivoting to consulting changes what you bill for, but it doesn't change the billing model itself. The incentive misalignment remains: the firm makes more money when engagements take longer.

What This Means If You're Buying Engineering

If your company currently works with an IT services firm — or is evaluating one — these industry shifts have practical implications for your budget and your outcomes.

Expect rising prices. As firms pivot toward higher-margin consulting, they'll bundle strategy and advisory services into what used to be pure engineering engagements. The per-hour rate may stay flat, but the total engagement cost will increase as consulting hours are layered on top.

Expect less engineering focus. The best engineers at these firms are being redeployed to AI and consulting roles. The talent that made EPAM's engineering reputation is increasingly working on EPAM DIAL, not on your project. Senior talent moves to where the firm sees its future — and the firm's future is consulting, not staff augmentation.

Expect AI features at premium pricing. Every IT services firm is now adding "AI-powered" capabilities to their engagements. Some of these are genuinely valuable. Others are thin wrappers around the same LLM APIs that cost $20/month on the open market, marked up to consulting rates. Distinguishing between the two requires more AI literacy than most buyers currently have.

Key person risk increases. As top engineers leave for AI-native startups and product companies that offer equity and more interesting problems, the bench depth at traditional IT services firms thins. Your project's success becomes increasingly dependent on specific individuals who may not be there next quarter.

The fundamental incentive conflict remains. Whether a firm bills for engineering hours or consulting hours, the time-and-materials model means their financial incentive (bill more hours) still conflicts with your incentive (ship working software faster). Changing the label on the invoice doesn't change the underlying economics.

The AI-Native Alternative

What matters for companies buying engineering isn't "which firm should I hire?" It's a more fundamental question: should I be buying engineering hours at all?

AI-native delivery models flip the equation entirely:

AI generates the specification from requirements. Xamun's DesignStudio takes business requirements and produces detailed technical specifications — screen flows, data models, API contracts — in hours, not weeks. Stakeholders review and approve before a single line of code is written.

AI generates 70–80% of boilerplate and scaffolding code from the specification. Using Claude Code via MCP (Model Context Protocol), the Software Factory generates the boilerplate — data layers, endpoints, UI components, test scaffolding — that used to consume the bulk of engineering hours.

Expert humans handle what matters. Architecture decisions, complex business logic, edge cases, security review, and quality assurance. The work that actually requires a human brain. The 20–30% where $200/hour is genuinely justified.

You pay for outcomes, not hours. Working software per sprint. Features delivered. Business value created. Not timesheets.

This is what Xamun's Software Factory does. The cost comparison is stark: a $15,000/month Xamun subscription — covering 30 story points, 4–6 features per sprint — replaces $35,000–$50,000/month of equivalent agency output. Not because the engineers are cheaper. Because AI eliminates the hours that didn't need a human in the first place.

The Honest Comparison

This isn't a situation where one model is universally superior. Context matters. Here's an honest assessment of when each approach makes sense.

IT services firms still make sense when:

• You need staff augmentation for your existing engineering team — additional hands working within your architecture, your processes, your codebase.

• You have a specific technology stack requirement that demands niche expertise — legacy mainframe migration, specialised embedded systems, rare platform integrations.

• You want to manage the engineering directly, with your technical leadership making architecture and implementation decisions.

• The project is primarily maintenance and enhancement of existing systems where deep institutional knowledge of the codebase is the primary value driver.

AI-native delivery makes more sense when:

• You're building new systems from scratch, where AI code generation delivers the highest leverage.

• Speed matters — you need working software in 2–4 weeks, not 4–12 months.

• You want intelligence guiding what to build, not just execution of what you specify. Strategy and implementation in one loop.

• You want outcome accountability — paying for delivered features, not time-and-materials billing where scope creep is a revenue model.

The Window Is Now

The IT services industry isn't dying — it's transforming. The firms that adapt fastest will survive. EPAM, to its credit, appears to be moving most aggressively. Thoughtworks is making thoughtful if slower moves. Globant is acquiring its way forward. All three have talented people and real institutional knowledge.

But the transformation means their model is converging toward consulting. And consulting means higher prices for less engineering output. The $150/hour developer becomes a $300/hour "AI transformation advisor" — even when the underlying work hasn't fundamentally changed.

For mid-market companies — the $20M–$200M revenue businesses that can't absorb inefficiency the way Fortune 500 companies can — this convergence creates a window. AI-native delivery models are available now, at price points that make the traditional hourly billing model difficult to justify.