Stop Reselling AI: The Real Way MSPs Profit From Artificial Intelligence

Sir, please sell me AI...

Do I have your attention?

Lately, I’ve been thinking a lot about a piece of advice I heard on a podcast. When you're building a product, you want to draw a bead on the market and shoot for where they are going. Because if you shoot for where they are today, you're going to miss. If you shoot for where they are today, you're going to miss.

And that brings us to the elephant in the room: AI.

Everyone is talking about it. Your clients are talking about it.

In fact, for the first time in maybe 20 years, we as MSPs are blessed with a unique opportunity. We actually have a technology our customers actively want to buy before we even have to educate them on it.

Usually, we have to use fear-based education. "Buy this firewall, or else the bad guys will take your data."

With AI, they are running to us. They are driven by a mix of curiosity and pure FOMO. They see their competitors using it, and they are terrified of falling behind the times.

So they knock on your door and say, "Sir, please sell me AI."

And what is our default instinct as MSPs? We want to hit the easy button.

Bink. Just figure it out.

But if you want to know how MSPs profit from AI the right way, you have to look past the hype and realize that taking the lazy, easy-button route will completely kill your margins.

Let's look at the math.

THE RESELLING TRAP

The easiest thing to do is to grab a purple marker, act as a Value-Added Reseller (VAR), and start flipping software licenses.

"Ooh, you want this AI widget? Let me sell it to you and add a little markup."

It’s exactly what we did with Office 365 when it first launched. On paper, it feels like you've unlocked a beautiful, low-effort MSP revenue stream. But there is a massive problem with this motion: you don't actually make money.

Let's look at the raw math.

Say an AI license costs a cool $100 per user per month. And let's assume you have been blessed with a highly generous channel margin of 25%.

That means you are making a clean $25 in product profit per user, per month.

If you have a 100-user client, that scales to a nice, round $2,500 of monthly product margin. On paper, that looks like fantastic mailbox money.

But let's look at the support burden.

If you pay a helpdesk engineer a modest salary of $50,000 per year, they work roughly 1,880 hours annually after you subtract PTO and holidays. That means that technician costs your business anywhere between $30 and $31 per hour, fully burdened.

Now, think about generative AI. It is highly complex, constantly changing, and driven entirely by user prompt error.

How many support tickets does it take to wipe out your margin?

If that 100-user client submits just a couple of configuration tickets, or calls because "the AI is hallucinating," your team spends hours troubleshooting. Your $2,500 in product margin quickly becomes a deep labor deficit.

That "mailbox money" was an illusion, and your new MSP revenue stream just became a major drain on your helpdesk.

THE CYBERSECURITY MISTAKE

So if reselling licenses is a bust, what is the next logical step MSPs will try?

We will be tempted to just cram AI support into our existing flat-rate managed service packages, the standard $150 to $175 per user per month agreements. Why? Because it makes the sales process incredibly easy.

Friends, I am calling a lot of you out right now because I made this exact mistake myself.

We did this with cybersecurity. We went out and bought EDRs, SIEM tools, and pen-testing services, and we just bundled them into our flat-rate seat price.

And what happened?

We severely burdened our help desks. Our technicians went from an average reactive support burden of 0.7 to 1.0 hour per user per month, straight up to 1.5 to 2.0 hours.

Our margins evaporated.

Managed service agreements historically hover around a 35% to 40% margin structure. If you try to cram unstandardized, highly complex AI support into that flat-rate labor bucket, your financial model is going to snap.

"Well, Kyle, I'll just raise my seat price from $175 to $200 to cover the AI support."

No, you won't. Because that creates contract bloat.

Eventually, the client forgets why they are paying that premium. They stop using the tools, they look at the bloated invoice, they realize they don't care about the extra features, and they churn.

THE SOLUTION

If we can't resell it, and we can't bundle it, how do we actually make money?

We have to look at what the market is actually asking for.

Your clients want to learn, adopt, and get better. They don't care about tokens or software markups.

If you are chasing sustainable, long-term MSP revenue growth, you have to do two things that are often treated as bad words in our channel:

We have to teach, and we have to consult.

But here is the catch, and it's a massive talent gap.

This high-value consulting cannot be done by your Tier 1 or Tier 2 helpdesk technicians.

They can't read a P&L. They don't understand the supply chain of your customer's revenue. And if they can't understand how your client actually makes their money, they cannot help them apply AI to optimize those business workflows.

To capture this high-margin consulting revenue, you have to build that business-level consulting talent on your team, or you have to be that person yourself.

Consulting labor trades at an incredibly high value. It gets you a permanent seat at your client's leadership table. And most importantly, those high-margin consulting dollars provide the cash flow you need to fund the internal R&D to build future, highly scalable, productized AI services.

THE BOTTOMLINE

The market is moving fast, and your clients are looking for a guide.

Treating AI like a SaaS resale commodity is a direct path to a helpdesk support deficit. To protect your margins, you have to pivot to business-level consulting and education.

If you want to watch me map all of this math out live on a whiteboard, I put together a full video breakdown of this concept. You can watch the full session, Stop Reselling AI - The Real Way MSPs Profit From Artificial Intelligence, where I dive even deeper into the numbers behind these helper margins and support costs.

But once you watch that math, you have to look yourself in the mirror and ask a hard question: Are your Tier 1 and Tier 2 techs actually ready to deliver that level of consulting?

Because if they still don't know what a P&L is, they are going to keep hitting the "easy button" and default to fixing widgets instead of fixing business workflows.

Empath is a platform designed to turn technical executors into strategic, business-minded assets who actually understand how your clients and your own MSP make money.

Ready to get your team speaking the language of business?  Start your 14-day free trial of Empath today.

Like this? You'll love our newsletter.

Get our weekly emails packed with more expert insights, new courses, and community events you won't find anywhere else.