OpenAI and Deepseek Drive AI Costs Up, Forcing MSPs to Rethink Pricing Models

[00:00:02] The real AI disruption for MSPs isn't mass job elimination. It's a margin squeeze driven by metered intelligence and shrinking seat counts, plus the inevitable rework of half-baked automation. So MSPs must shift to outcome-based pricing and become license and usage arbiters, that's standardizing, capping, and consolidating AI spend, with explicit re-review, rebuild clauses baked into every AI-enabled engagement.

[00:00:32] When AI becomes metered, every workflow becomes a cost surface, and when seat counts shrink, your old revenue proxy disappears. This is the Business of Tech. I'm Dave Sobel. We're seeing a specific set of signals that AI is moving from interesting capability to something that's actively reshaping how technology is packaged, priced, and adopted.

[00:00:58] Start with the money, because that's where behavior becomes visible. The information reports that enterprise software companies like Atlassian and HubSpot are moving away from flat, per-user AI add-ons and towards usage-based pricing, charging based on how much AI is actually consumed.

[00:01:17] The piece cites tracking that by the end of 2025, 79 of 500 software firms had introduced some form of usage-based AI fees, and it points to the growing concern that AI agents reduce the need for traditional seat licenses. In plain terms, vendors are building pricing models that treat AI less like a feature and more like a metered utility. We're also seeing AI spread through lower cost and open distribution models.

[00:01:47] Semaphore reports that DeepSeq has released another open-source model it says could compete with top-tier systems at lower cost. And the New York Times describes its latest preview as a meaningful jump in open-source code generation performance. That matters because cheaper, more available models increase adoption pressure even when premium licensing is not the entry point. And AI is moving into core workflows.

[00:02:14] Semaphore reports the CIA has produced an intelligence report generated by AI, and leadership says AI will be embedded across analytic platforms within two years. However you interpret that timeline, the operational signal is the same. AI is moving from experimentation into systems organizations expect to run real work. If you're listening to this and you haven't hit follow yet, on Apple Podcasts, search the business of tech.

[00:02:42] It takes five seconds, and you'll get tomorrow's show automatically. This episode is supported by Comet Backup. Not all heroes wear capes. Some live among us, quietly saving businesses one help desk ticket at a time. Whether you're battling ransomware, hardware failure, or human error, Comet's powerful backup and recovery solutions put you in control. Manage all your backups in Comet's simple, centralized platform. Protect computers, servers, virtual environments, emails, and databases.

[00:03:11] When disaster strikes, be the hero your business needs. With Comet Backup, you're not just saving the data, you're saving the day. Comet Backup, for the everyday IT heroes. Visit cometbackup.com to start your free 30-day trial today. Get $100 free credit when you sign up with the promo code MSPRADIO. Comet Backup, be the hero, save the day. Here's why this shift is happening.

[00:03:37] When AI agents reduce the number of human seats, vendors lose the clean per-user growth that software pricing has relied on for years. So they move AI from bundled feature to metered consumption. That restores the vendor's revenue curve, but it pushes cost volatility downstream to the customer, and eventually to the MSP. But metered pricing only works if AI gets embedded into real work.

[00:04:04] And that means orchestration, identity, approvals, ticketing, data access, review, and auditability. The model's not the bottleneck, the operating layer is. That's why OpenAI's symphony matters. It's a signal that the value is shifting from model access to the control plane that makes AI work legible, governable, and repeatable inside existing systems. The same pattern shows up in distribution.

[00:04:30] AppDirect's acquisition of partner stack is not just a channel story, it's a packaging story. When customers cannot assemble a coherent AI operating motion internally, value shifts to the platforms and partners that can provision, govern, route, and manage that motion for them. That is the mechanism. AI compresses seat-based revenue, expands variable consumption, and increases the amount of workflow governance required to make the system usable.

[00:04:56] So the MSP gets pulled toward the operating layer, standardizing providers, capping usage, documenting workflows, securing permissions, and reviewing outputs. If that layer is not explicitly scoped and priced, it shows up first as unbilled operating work, and then as rework. Let's put numbers on this, because this is the part that breaks MSP math.

[00:05:22] Imagine you've got a 120 user client at $150 per user. That's $18,000 a month. Now the client rolls out agents and automation, and they need fewer seats. They drop to 85 users. Your seat-based revenue just fell to $12,750. Same environment, fewer licenses. But your AI costs don't shrink with seat count. They move with activity.

[00:05:49] So now you've got usage fees, API calls, agent runs, and premium tiers that can easily turn into $1,500, $3,000, $5,000 a month in consumption. And it spikes when projects spike. That's the squeeze. The old revenue proxy shrinks while the new cost line becomes variable and harder to predict. OrgView survey data via tech bullion says 55% of companies that laid people off for AI later regretted it.

[00:06:18] Treating that as a warning flare, not a controlled study. The pattern is what matters. First-pass automation rarely lands cleanly. And the cleanup is exception handling, quality control, oversight, and fixing what breaks in real operations. In the mid-market, that second-pass work often lands on the MSP because the client team doesn't have the time or authority to keep revisiting it. When the failure mode crosses into security, cleanup stops being a quick fix.

[00:06:48] It becomes hands-on remediation, disruption, and documentation. The kind of work that destroys margin when it's bundled under support. So this automation layer becomes a standing source of variable, repeat work. Quality rework on the AI side and high friction remediation on the security side. The MSP either becomes the provider that simplifies and governs that automation layer, scoped, priced, and run as a managed operating motion.

[00:07:15] Or it becomes the sponge that absorbs the rework, the resets, and the exceptions under the word support for free. This episode is brought to you by Control Map. Growing MSPs are using Control Map to build recurring revenue by expanding their GRC services. Starting now, Control Map is offering a free plan for MSPs looking to get started with providing compliance as a service. Create a free account and run an assessment.

[00:07:44] Track key items like policies, risks, and evidence in one place. It's a practical way to prove value to a client before deciding to expand your compliance offering. Try Control Map for free today. Visit scalepad.com slash Dave to get started. That's scalepad.com slash Dave. A quick heads up, Acronis is hosting a live event on May 13th called The Pivotal Point of IT, building services for the AI-first era.

[00:08:14] Their CEO will be laying out Acronis' vision for AI-first service delivery for MSPs, including a new partner program and what they're calling a major platform announcement. If you want to hear directly from Acronis on where they're taking all of this, registration link is at go.acronis.com slash Dave Sobel AI era. No spaces. Why do we care?

[00:08:38] Because if an MSP treats broken AI workflows as support instead of scoped governance work, it accepts liability without pricing it. Fix the workflow without a change order and your fingerprints are now on the design, the remediation, and potentially the compliance exposure. The MSPs who win this transition will sell workflow governance as a documented, recurring service.

[00:09:02] The ones who lose will keep doing it as goodwill, which creates unbuilt labor, unclear responsibility, and lower quality revenue. And that matters beyond margin. If your business shifts from predictable per-seat MRR to variable usage pass-through plus surprise rebuild projects, lenders and acquirers see a riskier revenue stream and murkier contracts. This is not just about profitability. It's about staying financeable.

[00:09:32] So what to consider? Build a consumption monitoring capability before you need it. Even a basic spreadsheet-level tracking of per-client AI API usage, license tiers, and agent activity creates the visibility needed to act as a usage armature. Reframe your AI service offering around workflow governance, not tool deployment. The differentiator isn't we deploy AI tools.

[00:09:58] It's we design, document, and maintain the operating motion that makes AI outputs trustworthy and auditable. That framing commands higher margin and creates stickier engagements. Separate AI rework pricing from security incident response pricing. These are distinct labor categories with different skill requirements, different urgency profiles, and different liability exposures.

[00:10:23] Bundling them under support destroys pricing clarity and creates scope creep. If this trend continues, the mid-market MSP that does not implement AI usage caps, standardized model and provider choices, and contractual rebuild windows will be forced into quarterly repricing just to stay whole because AI consumption will become the fastest growing pass-through cost in the stack while seat-based revenue stagnates.

[00:10:54] This is the Business of Tech. Want more from the Business of Tech? Join Business of Tech Plus for ad-free episodes, early interviews, extended cuts, subscriber-only shows, and exclusive member perks and analysis. Sign up at businessof.tech slash plus. And follow this show on your podcast app, and if you're on YouTube, hit subscribe and the bell so you never miss a story. Reviews and comments help spread the word too.

[00:11:25] Interested in advertising? Head to mspradio.com slash engage. The Business of Tech is written and produced by me, Dave Sobel, under ethics guidelines posted at businessof.tech. Thanks for listening. I'll see you on the next episode.