vCIO Insights is a thought leadership series from Mike Nazli, vCIO at i-Tech Support, sharing practical perspectives on AI, cybersecurity, compliance, and technology strategy for business leaders.
Two recent conversations made the AI economy real for me in a way no product announcement or market forecast could.
In the first, a community-serving organization needed eight dependable laptops. Its budget was approximately $600 per device. The least expensive business-grade option available through our normal procurement channels was close to $1,000, and that was before adding any meaningful margin.
In the second, someone close to my family had been offered work on a data-center project in Iowa. He described it as the kind of job that could carry him through the rest of his career and possibly create a path for his son.
Those two conversations seemed unrelated at first. The organization struggling to replace ordinary computers and the construction worker looking at years of steady work are on opposite ends of the same buildout, and for the organizations I advise, the practical result is an IT budget whose numbers stopped holding still this year.
Why Do We Need Data Centers?
We need data centers because AI is physical before it’s digital. Every model, every chatbot response, and every automated summary runs on processors, memory, storage, and cooling equipment sitting in a building somewhere, drawing power from a grid. There is no version of AI that exists only in software.
Most people experience AI through a browser, as a chatbot or a copilot inside software they already use, and never see the buildings behind it. Those buildings need land, power and water, and they need crews of electricians and engineers to build and run them.
That demand is creating enormous opportunity. It’s also placing pressure on the same technology supply chain that schools, nonprofits, and small businesses depend on.
An organization doesn’t need to purchase an AI subscription or train a model to feel the effects. It may encounter them when replacing a laptop, expanding storage, budgeting for electricity, or waiting for equipment that has become harder to source.
Organizations may already be paying for the AI race without realizing they entered it.
Why Are Laptop Prices Going Up?
Laptop prices are going up because the AI buildout is consuming the memory supply that consumer and business machines are built from.
Memory makers are shifting production toward servers, and the higher component costs are working their way through distributors to the laptops businesses buy.
Over the past several months, I’ve met with leaders from schools, nonprofits and small businesses, including several homeless-services networks. Most aren’t buying thousands of GPUs or building advanced AI systems. They’re trying to maintain dependable, secure technology environments while serving their employees and their communities.
For many of these organizations, technology funding comes from grants, donations, restricted funds, or budgets approved months in advance. Those budgets don’t increase because the market changed between the day a grant was submitted and the day the equipment had to be purchased.
That’s why the eight-laptop conversation was so difficult.
Someone involved in the project remembered buying comparable Dell Latitude systems for approximately $700 to $800 not long before. He was right. A year earlier, I could likely have sourced a similar business-grade system much closer to that range. In the situation in front of us, the available option cost close to $1,000, which is roughly 40% more than he remembered paying.
That doesn’t mean every laptop on the market has risen by the same amount. It’s one real example of what I’m seeing through the business technology channel, and the component market explains most of it.
Why Is RAM So Expensive Right Now?
RAM is expensive right now because AI data centers are buying memory faster than manufacturers can make it, and server demand pays better than consumer demand. When a manufacturer can sell the same production capacity into an AI buildout at a higher margin, laptop memory is what gets squeezed.
TrendForce reported that conventional DRAM contract prices rose approximately 93% to 98% quarter over quarter during the first quarter of 2026.
For the fourth quarter, the firm projects another 10% to 15% increase in conventional DRAM contract prices and a 15% to 20% increase for NAND flash, with suppliers still prioritizing server products and the market still undersupplied.
Memory is only one part of a computer, and contract pricing doesn’t translate dollar for dollar into the final price of every machine. It still shows the direction of the pressure, which is the same pressure we broke down in detail in why hardware prices are surging.
The cloud draws on the same supply chain that builds the computers used in classrooms, nonprofits and ordinary businesses.
How Long Is a Hardware Quote Good For?
A hardware quote is good for as long as the distributor’s stock and pricing hold, which in 2026 often means days rather than the 30 days most buyers still assume. Many distributor terms reserve the right to charge the price prevailing at shipment, not the price on the quote.
Managed service providers sit between manufacturers, distributors, and the organizations buying equipment. That position gives us a practical view of how quickly conditions change.
There was a time when I could build a proposal around distributor pricing and reasonably expect it to hold for about 30 days. Then the practical window started feeling closer to two weeks.
In practice, a quote now records the price on the day it was issued and commits the distributor to very little after that.
By the time an organization submits a request, receives approval, confirms its funding, and issues a purchase order, the original equipment may be unavailable or the price may have moved.
Ingram Micro, one of the largest distributors in the industry, puts this in writing: its U.S. sales terms state that products may be sold at the price prevailing when they ship, and that quoted prices can change without notice.
Not every quote will change, but a leadership team can no longer assume that a price captured during planning will still be there after a long approval cycle.
Do Data Centers Create Jobs?
Data centers create a great many jobs during construction and far fewer once they’re running. The construction phase draws electricians, HVAC professionals, welders, pipefitters, fiber technicians, and construction managers, often for years. The operating phase needs a comparatively small technical staff.
The other side of this story is opportunity. While these organizations absorb higher costs, the same construction is keeping electricians, pipefitters and fiber crews booked for years.
Iowa isn’t a random destination for that investment. The Iowa Economic Development Authority points to billions invested by Google, Meta, and Microsoft, alongside abundant wind energy, grid stability, telecommunications infrastructure, and construction costs below the national average.
The buildout has moved well beyond Silicon Valley, into farmland and small towns that few people would have associated with the technology industry a decade ago.
That demand is already reshaping parts of the skilled-trades economy. Building-trades officials cited in AP-syndicated reporting said data-center projects accounted for at least 40% of the labor hours worked by members of the Columbus, Ohio, building-trades council, and at least 50% of the hours worked by a major electrical union’s local in the Washington, D.C., area.
Unions reported larger apprentice classes, expanded training centers, and record membership during 2025.
Meta has committed $115 million to America’s Workforce Academy, a free training program for data-center technician and skilled-trades roles that guarantees participants a job on completion.
AI will disrupt some kinds of work, but building it is creating steady demand for the people who construct, power and maintain the facilities.
How Many Jobs Do Data Centers Actually Create?
A large data center typically creates thousands of construction jobs at peak and a few hundred permanent operational roles.
The opportunity is real, though it is limited and unevenly spread, because data centers require far more people to build than to operate.
Meta’s El Paso venture with BlackRock, a project now exceeding $10 billion, is expected to support more than 4,000 construction jobs at peak and 300 operational jobs once the campus is running.
A separate project in Tulsa, Oklahoma, is expected to create more than 1,000 construction jobs at peak and approximately 100 permanent operational positions.
The local effects are real but smaller than the headlines suggest. A 2026 Brookings analysis matched roughly 1,500 U.S. data-center facilities, along with 52 announced projects that were later canceled, against county employment and wage data. Counties that received their first large data center saw data-processing employment rise 56% over the first decade of operations, and telecommunications employment rise 43%. Wages were unchanged, and home prices rose a modest 2% to 5%. In a typical county, that works out to roughly 100 to 200 jobs.
The type of facility matters. Hyperscale campuses built by companies such as Google, Meta and Microsoft for their own workloads also produced telecommunications jobs, because they need local fiber and network operations staff. Colocation facilities, which lease space to remote tenants, did not.
When Budgets Do Not Move, Risk Does
The problem for small and community-serving organizations is that their budgets rarely move as fast as the market.
When an organization can’t afford the recommended business-grade equipment, it still has to operate. Leaders start comparing consumer computers against business systems. They consider refurbished devices, third-party warranties, online marketplaces, and manufacturers their technology team has never supported.
Those choices say more about the budget than about the people making them. Nobody stopped caring about security or reliability, and the budget simply hasn’t caught up with the market.
A properly evaluated refurbished business-grade system is sometimes a reasonable compromise. A low-cost purchase can also mean a thin warranty, weaker security features and higher support costs over the life of the machine. Cost pressure moves the risk from today’s purchase into next year’s support tickets.
That makes technology procurement a leadership decision, because it affects cybersecurity, cash flow and how long the organization can run on what it buys. It belongs on the same agenda as AI governance and the other technology decisions that start with leadership, not three levels below it.
A technology advisor should help leadership understand the tradeoffs, set standards it can defend and build a roadmap the organization can realistically fund.
How Do You Plan an IT Budget When Prices Keep Moving?
You plan an IT budget in a moving market by dating every assumption, funding a range instead of a single number, and preapproving a second choice before you need one. Most budget guidance assumes prices are known and the work is allocation. That assumption is what broke this year.
1. Date Every Assumption
Every number in your technology budget was true on a specific day.
Write that day next to it. A workstation figure sourced in the last 60 days is a price. The same figure carried forward from an 18-month-old quote is an estimate, and nobody reviewing the budget can tell the two apart once they’re in the same column.
2. Fund a Range
A single figure per line is a forecast presented as a fact. Replace it with a range and a stated contingency band, then say out loud what happens if the top of the range is exceeded.
For grant-funded organizations this matters most, because the funder is often willing to discuss technology inflation at the application stage and almost never willing to discuss it after the award.
3. Preapprove the Second Choice
Decide now what gets purchased when the first choice is unavailable or materially higher, and decide who is allowed to make that call without restarting the approval process.
This is the single cheapest change on the list and the one that recovers the most time. Most of what a stale quote costs is the weeks spent going back through an approval cycle that was already completed once.
The Boardroom Question
A nonprofit, a school or a small business may never train a model or buy an enterprise AI platform, yet when those organizations replace laptops, apply for technology grants, evaluate utility costs, or revisit their hardware standards, they’re already participating in the AI economy.
At the same time, a construction worker in Iowa looks at the same transformation and sees decades of opportunity for his family.
Both things are true. The question is whether your organization has accounted for either one.
“If the price we budgeted for is gone by the time the purchase order clears, what do we do?”
If your leadership team doesn’t have an answer to that, you’re in good company. Most of the organizations I meet with don’t have one either.
Usually the budget was built on numbers that were accurate when someone wrote them down, and answering that question is a planning job for leadership before the next purchase order goes out.
Continue the Conversation
The organizations handling this well treat the technology budget as a plan that has to survive a market moving faster than their approval cycle, and they revisit it more than once a year.
i-Tech Support works with schools, nonprofits, and growing businesses across Florida on exactly this: setting technology standards, building realistic multiyear roadmaps, and pricing them against what the market is actually doing.
