Kaseya surveyed over 1,000 MSPs for their 2026 State of the MSP Report. The number that should keep you up tonight: 48% of MSPs say AI and automation is the #1 thing their clients are asking for. Ahead of security. Ahead of backup. Ahead of everything.
Only 13% of MSPs generate real revenue from AI services.
That's not a gap. That's a canyon.
The triple squeeze
Here's what makes this worse than a normal capability mismatch. Three things are happening at the same time.
Your clients want AI. 48% is not a trend. It's the top line item. And if you're not delivering it, someone else will. Kaseya's own data shows 33% of new MSP clients are switchers -- they left their last provider. They're not shopping for the cheapest help desk. They're shopping for the MSP that can actually do the thing they need done.
You can't hire the people to build it. MSPs reporting difficulty hiring skilled technicians nearly doubled year over year, from 9% to 16%. Training gaps more than doubled, 7% to 15%. The global AI talent market runs at a 3.2-to-1 demand-to-supply ratio -- 1.6 million open roles, roughly 518,000 qualified candidates. ManpowerGroup's 2026 survey found 72% of employers report hiring difficulties, with AI skills now outpacing traditional engineering for the first time.
Your deal sizes are collapsing while you try to figure this out. MSPs reporting customers spending above $25K per year dropped from 75% to 41%. Nearly a quarter of MSP customers are actively cutting IT spend. You're being asked to do more, with less, for clients who are also paying less.
That's the squeeze. More demand. No talent. Shrinking revenue per client.
It's not even a skills problem
Here's the part most people miss. The instinct is to frame this as a hiring problem. "If we could just find the right AI engineer..." But AvePoint and Omdia surveyed 333 MSPs across North America, EMEA, and APAC, and 51% cited governance and compliance as the #1 barrier to AI adoption. Not skills. Not budget. Governance.
You can hire the engineer and still fail because you haven't solved the policy layer, the audit trail, the multi-tenant compliance scaffolding. 40% of MSPs blame multi-tenant operational complexity as the top barrier to automation. That tracks. Every client environment is different. Every exception requires another rule. Every rule introduces risk.
And here's the real kicker: 53% of MSPs already use AI internally for ticketing, patching, and monitoring. The problem isn't adopting AI. It's selling AI. Most MSPs consume AI as an operational tool but can't package, price, and deliver it as a service their clients can buy. The gap is commercial, not technical.
The math on building in-house
Let's run the numbers anyway. A minimum viable AI team in the US -- and I mean minimum, not good -- runs $755K to over $1 million per year in salary alone. That's before benefits, tooling, infrastructure, and the 90 to 120 days it takes to hire a senior AI engineer in this market.
Now factor in the failure rates. 70% of AI projects never reach production, according to Gartner. MIT researchers found 95% of generative AI pilots fail to produce measurable financial impact. S&P Global reported that 42% of businesses scrapped AI projects entirely in 2025, up from 17% the year before.
A failed senior AI hire costs 1.5x to 3x their annual salary when you account for recruiting, onboarding, lost productivity, and backfilling. At these pay grades, that's a $300K to $600K mistake.
GTIA panelists described this perfectly earlier this year. One MSP leader said it plain: "Everyone wants something that will magically solve the service desk. But you have to be a fundamentalist first. AI hyper-scales bad processes." Another rolled back an invoicing automation that generated angry client calls. A third abandoned automated security incident response because the stakes were too high for the error rate.
One MSP reportedly gained three clients from a competitor that deployed customer-facing chatbots poorly. 87% of MSPs say they need to improve their AI knowledge before they can meet customer needs. The gap between "we use AI internally" and "we sell AI services" is wider than it looks.
What actually works
The hybrid model is emerging as the 2026 playbook for a reason. Outsourcing AI capability runs 40-60% of the Year 1 cost of building in-house. But more importantly, it compresses the timeline from "maybe we hire someone by Q3" to "we have a deliverable next month."
The white-label AI market is projected at $99 billion globally by end of 2026. MSPs are charging 3x to 5x markup on white-labeled AI services, with $1,200 per month packages adding $14.4K in annual recurring revenue per client. That's real margin on someone else's R&D.
This is what we do at Kief Studio. We've been running white-label engagements for agencies and MSPs -- your brand, our engineering, we stay invisible. We've built 40+ custom internal tools, shipped production AI systems, and solved the governance and compliance scaffolding that most teams haven't even started thinking about. Between the two of us -- Brian and Meelie -- we cover what would typically require a 10 to 14 person team, augmented by custom AI tooling we built ourselves. That's not a pitch. That's just the math on how a two-person studio handles the workload.
The MSPs that figure out the partnership model will be the ones selling AI services by Q3. The ones that insist on building from scratch will still be interviewing candidates.
The window
Omdia projects the global MSP AI services opportunity at $276 billion by 2030. The M&A wave is already accelerating -- MSPs that can't offer AI services become acquisition targets, not acquirers.
You don't need to become an AI company. You need a partner who already is one.
First conversation is free. No commitment. kief.studio/contact