Your Offshore Team Was Priced by Headcount. AI Just Changed the Math.

This week Microsoft offered voluntary buyouts to roughly 60,000 employees in the US, India, the UK and Canada, about a quarter of its workforce, as it shifts spending toward AI. You don't need to run a company that size to take the signal seriously. The economics of technology work are changing, and a lot of mid-sized companies are still paying for their offshore and outsourced teams the old way.

I've built, acquired and integrated offshore engineering organizations, including leading the acquisition and integration of an offshore engineering group while at NTT Data. I'm also living this change in my own business. At Partner2Learn, I used two offshore firms to build our solutions. Today, with AI-assisted development, I've taken over the code base for three of those solutions myself. We are only seeing the beginning of this. Here's how I'd think about your team model heading into 2027.

The Old Model: More Heads, Lower Rates

For 20 years, companies used offshore firms as lower-cost labor for brute-force work: solutions that took a lot of manual building. The model ran on a simple formula:

• •Lower hourly rates than onshore.
• •A target ratio, for example 70% offshore and 30% onshore.
• •Contracts priced by full-time equivalents (FTEs), tickets or hours.
• •Success measured by how many people were on the team.

That formula assumed output grew with headcount. AI-assisted coding, testing and documentation break that assumption. A handful of people can now deliver what used to take a much larger team.

What's Changed

• •Productivity is moving faster than contracts. HFS Research reports that companies are reopening IT services contracts within 24 months of signing because AI gains have outrun the original pricing. Many deals signed in 2024 and early 2025 were priced before agentic AI changed delivery economics.
• •Pricing models are shifting. Negotiations are moving away from rate cards and offshore ratios toward productivity commitments, outcome-based pricing and gain-sharing.
• •The vendor promises are big. Providers often claim 40–70% productivity gains from AI-enabled services. In practice, the results are uneven.
• •Smaller teams can do more. One former CIO quoted by InformationWeek estimated that a team of 10–15 people using AI coding tools can now do work that once took 40–50 offshore developers, testers and analysts.

None of this means offshore is dead. For mid-sized companies, offshore and nearshore partners are still one of the best ways to access scarce skills. What's changing is what you should be buying.

Where Mid-Sized Companies Get This Wrong

• •They renew on autopilot. They roll over the same headcount and rate card for another year, and the vendor keeps the AI savings.
• •They cut blindly. They cut the team by a third because "AI will cover it," and lose the people who understand the systems.
• •They let the vendor bring AI tools with no rules. Nobody has agreed which tools are used, whether your code and data can train a model, or who owns AI-generated work.
• •They have no onshore technical owner. A smaller, AI-enabled team needs stronger architecture, code review and product direction, not less.

Questions to Ask Before You Renew or Add Heads

1. 1.What are we actually buying? Hours and people, or working features, resolved tickets and uptime?
2. 2.What has our partner done with AI so far, and where are those savings showing up on our invoice?
3. 3.Which roles change most? Testing, documentation and routine coding change the most. Architecture, integration and business analysis matter more.
4. 4.Does our contract cover AI? Tool disclosure, no training on our data, IP ownership of AI-generated output, liability for AI errors, and a share of productivity gains.
5. 5.Who owns the team on our side? Someone who can review the work, set standards and hold the partner accountable.

What to Do Next

• ☐List every offshore, nearshore and outsourced engineering or support contract, with renewal dates and pricing model.
• ☐Ask each partner for a written summary of the AI tools they use on your work and the measurable gains so far.
• ☐Choose two or three outcome measures (cycle time, defect rate, backlog burn-down) to replace "headcount" as the main metric.
• ☐Add AI clauses (data use, IP, liability, gain-sharing) before the next renewal, not after.
• ☐Re-balance the team: fewer routine roles, stronger architecture and onshore product ownership.
• ☐Build these changes into your 2027 budget now, while the numbers are still being set.

Frequently Asked Questions

Related Reading