The answer is the vendor's business model. That is, how a business makes money out of its particular know-how. One of the most lucrative approaches is the model made famous by the Keurig single-cup coffee machine. First, you sell a technological wonder that promises a variety of gourmet coffee delivered conveniently; then, you sell the proprietary pods. The machine is intentionally cheap. The pods are where you live.
This is why the sticker price is the least reliable number in any quote. The cost that actually dictates the outcome arrives long after the signature, and the vendor's business model reveals its size in advance. Traditional technology vendors run this exact play. Because commoditization and simplicity drive prices and margins down, these vendors survive and thrive by warping your infrastructure in their favor to lock you into their ecosystem. Their complexity is not an accident. It is the pod, holding on to their share of your mind and your money.
Start with the bill you think you understand. You might sign a three-year Azure commitment to lock in the capacity your servers need, yet Microsoft's own documentation states you did no such thing. A reserved instance provides "a billing discount only" and does "not guarantee capacity." The product that actually holds capacity is entirely separate. It is billed whether or not the machine runs, and during a shortage, even that capacity can be refused. The discount was real, but the capacity you thought you purchased was merely a guess.1
Extracting your data carries a cost as well, and that price remains hidden until you try to leave. For years, major cloud providers charged almost nothing to move data in while metering every gigabyte out. This ensured the cheapest place to keep your data was exactly where it already sat. That exit toll was not defeated by market competition; it fell to legislation. The European Union's Data Act will ban switching charges outright starting in 2027. As this law advanced, the major clouds dropped their exit fees worldwide.2
The strongest lock is not the financial toll of leaving, but the part of you that cannot leave at all. In the realm of AI, this carries a steep price. The expertise you build into a fine-tuned model does not come with you. Once deployed, it runs exclusively on the vendor's platform, is accessed only through their tools, and incurs "an hourly hosting cost regardless of whether" you ever call it.3 Satya Nadella, whose company sells these very machines and meters their respective pods, defines the test himself. He states that being able to swap out a general model without losing the expertise your business taught it is "the key test of your control and sovereignty in the era ahead." If you apply that test to the fine-tuned model his own platform sells, it fails: the expertise you paid to cultivate cannot follow you off their platform.
Every one of these locks shares a fundamental weakness. When Keurig wired its later machines to reject unlicensed pods, customers revolted. A competitor shipped a workaround within months, and the company was forced to retreat. Complexity bought Keurig years of high margins, but it did not buy them the ultimate win. A lock is only ever as strong as the day you decide to leave.
The same model wears a service badge. Many managed IT service providers run your estate from their own commercial remote-management platforms, so the monthly fee buys the operation of a tool, not an estate that becomes yours: it acts as if it were your infrastructure while building nothing you keep. When the contract ends the endpoint agent comes off and nothing built remains. Building instead into a platform you own, such as Microsoft Intune, inverts the trade: the hardened estate lives in your own tenant and stays when the provider goes.4
Thankfully, there is another business model in IT, one that separates implementation from the technology itself and delivers unadulterated technology that is more robust and less costly. This is the model utilized by firms like IBM and Red Hat, which generate revenue by selling expertise and support for technology you are completely free to take elsewhere. It is also the model of the myriad other companies that contribute billions to Open Source Software development. On a humbler scale, it is our business model as well.* With unadulterated technology, you control the variables. If you control the variables, you control the results, and the cost.
Before you sign your next proposal, do the cost accounting required to predict the future. Ask how the vendor makes its money, and determine whether its margins grow in proportion to your dependence or your results. Turn this question on your incumbent provider, on the bright new platform, and on us. The standard industry answer relies on a markup on hardware, a rebate on software, and a commission on advice. Our answer relies on none of those. Seventy cents of every dollar you spend with us goes directly to senior engineering, meaning our margin only grows by building you something that truly works.
Concierge CIO Partners is a unified Guild of senior technologists providing dedicated, long-term fiduciary IT leadership to midmarket service firms. It offers a strategic alternative to fragile internal IT silos and to Managed Service Providers who advise you on what to buy and are paid on what you buy. With transparent unit pricing and an automated service catalog, the Guild eliminates administrative bloat and ensures every IT dollar spent and decision made directly drives your financial performance. The vendor's business model is your future invoice; the Guild wrote its own to be read in the open, and priced it by the engineering it delivers.
* 10 to 200 users and PE/VC backed startups. 1 Microsoft Learn, On-demand capacity reservation: "Reserved Instances provide a billing discount only and don't guarantee capacity." A capacity reservation is billed at the VM rate "even if the reservation isn't being used," and can still fail "if Azure doesn't have capacity available." 2 European Commission, Data Act explained: switching charges are to be removed entirely from 12 January 2027 (the Act has applied since 12 September 2025). Google, AWS and Microsoft waived exit-egress fees in 2024 as the Act advanced, and applied the waiver worldwide: AWS's is "available to all AWS customers around the world," Microsoft's to "all Azure customers globally." (AWS) The waiver covers a full departure; routine egress still meters. 3 Microsoft Learn, Deploy a fine-tuned model: once deployed, a customized model "incurs an hourly hosting cost regardless of whether chat completions or response API calls are made," and is served and used through the platform's own tools. Satya Nadella, sn scratchpad, 14 June 2026: "A company should be able to switch out a 'generalist' model without losing the 'company veteran' expertise built into their learning system. This is the key 'test' of your control and sovereignty in the era ahead." He is Microsoft's chief executive, writing on his own site, and the argument favours the platform his company sells. Quoted as a test worth applying, including to its author. 4 Concierge CIO, IT costs in context, and our published service terms: traditional providers take markups and rebates of 12 to 25% on software, 20 to 40% on cloud and 15 to 30% on hardware, with vendor recommendations "influenced by commissions," and bill meaningful improvement separately as out-of-scope "change order" projects (10% to 20% of their revenue), counting maintenance as included but evolution as extra. Concierge CIO takes no markup, rebate or kickback, directs about 70% of each engagement to senior engineering labor, and builds into platforms the client owns, such as Microsoft Intune, "so nothing is dependant on our internal systems."