Key takeaways
- Most 2027 cost increases aren't optional: software renewals are up, AI is being folded into existing plans, and computers cost roughly 17% more.
- Buy computers with at least 16 GB of RAM (32 GB for intensive work). Under-buying now means replacing sooner, at higher prices.
- Put every renewal on one list and decide what to retire or consolidate before anything renews. That's where most of the savings are.
- Treat AI as one budget line, even though the charges live in different places, so overlapping subscriptions become visible.
- Any new tool, including ones staff build with AI, needs an owner, a clear answer to "What does this replace?", and a plan for retiring it.
Most technology budgets start the same way: take last year’s numbers, add a little, and move on to the next line item. For 2027, that approach is going to come up short.
Let’s start with the laptops, because that’s where the sticker shock is worst. If you haven’t priced a replacement computer lately, brace yourself. The machine you bought even two or three years ago now costs noticeably more, and the specs you settled for then won’t last as long now.
Computers are just the most visible piece. Platform renewals are increasing, AI charges are turning up in places nobody budgeted for, and staff are asking for new tools, including some they’ve built themselves with Claude or ChatGPT. Others want the AI apps they already use at home, like Grok, Meta AI, Gemini, or Copilot.
The short answer: most organizations should expect 2027 technology costs to rise, driven by software license increases, AI pricing folded into existing tools, and sharply higher hardware prices. Much of that increase isn’t optional. The planning work is deciding where to absorb it, what to consolidate or retire, and how to handle new AI requests before they turn into unplanned commitments.
The costs you didn’t choose
Computers cost much more, and they’re aging faster
The culprit is a global memory shortage, driven largely by AI data centers buying up every memory chip they can get. Gartner estimates average PC prices rising about 17%, and IDC doesn’t expect meaningful relief before late 2027. In other words, the AI boom is now showing up on your laptop invoice.
At the same time, everyday work needs more memory than it used to: dozens of browser tabs, back-to-back video calls, large spreadsheets, AI features running on the machine itself. Computers that once comfortably lasted three years or more are running out of breath sooner.
Our current guidance for new computers:
- 16 GB of RAM is the minimum.
- 32 GB where the budget allows, for anyone doing intensive work: large spreadsheets, data analysis, specialized scientific tools, and similar.
- Don’t save money by under-buying. A cheaper machine with 8 GB will likely need replacing much sooner, at whatever prices look like then.
- Tier devices by role and stagger replacements instead of refreshing everyone at once, and budget a higher per-unit cost than you’re used to.
- Don’t panic-buy. Replace machines on a plan, starting with the ones that are actually holding people back.
Software licenses are up almost everywhere
Microsoft is the clearest example, because so many organizations rely on it. As of July 1, 2026, Office 365 E3 went from $23 to $26 per user per month, a 13% increase, and Microsoft 365 E3 from $36 to $39. Microsoft 365 E5 rose from $57 to $60. (Those are list prices for the versions that include Teams.) Microsoft tied the increases to added AI, security, and management features, including email protections from Defender for Office 365 Plan 1 in the E3 plans. Existing customers see the new price at their first renewal after July 1, so many organizations will meet it for the first time in their 2027 budget. For a 75-person organization on Microsoft 365 E3, that’s an extra $2,700 a year before anyone adds a single new tool.
Google went down the same road a little earlier. When it folded Gemini into its business plans in 2025, Workspace Business Standard moved from $12 to about $14 per user per month, and customers on annual terms have been absorbing that at renewal since. Most of the other vendors you pay are doing some version of the same thing. Nobody’s renewal notice this year opens with good news.
What to do about it:
- Put every renewal on one list, with its renewal date and the notice window for changing or cancelling it. Auto-renewals are where budgets quietly leak.
- Check seat counts against actual people. Departed staff, shared accounts, and “temporary” licenses add up.
- Ask what each plan now includes. Price increases often arrive with bundled features (security tools, storage, AI) that may duplicate something you already pay for separately. Sometimes the increase is a chance to retire another tool.
AI is showing up in more places than you think
Many organizations are now paying for AI in several places at once: a Copilot or Gemini tier in the productivity suite, AI add-ons in the CRM, the project tool, and the meeting note-taker, plus individual ChatGPT, Claude, or Gemini subscriptions on staff expense reports. AI tools are also getting more expensive as vendors steer people toward higher tiers. Each charge looks modest. Together, they can add up to a meaningful amount that nobody has looked at as a whole.
Expect more requests, too, for the AI apps staff use in their personal lives: Grok, Meta AI, Gemini, the consumer version of Copilot. Familiarity is a reasonable starting point, but consumer versions often come with different data terms than business plans. Some may use your conversations to train their models by default, and none of them come with your organization’s admin controls. Before saying yes, check whether a business version exists, what it does with your data, and whether it duplicates something you already pay for. Often the better answer is to choose one or two approved AI tools and make them good enough that people don’t go looking elsewhere. (For the tools you do approve, our guide to AI privacy settings in ChatGPT, Claude, and Perplexity covers what to check.)
We recommend treating AI as a single budget line for 2027, even though the charges live in different places. That makes overlap visible. It also lets you move people from personal accounts to an organizational plan with proper data protections, and gives you a real number to weigh against what people are actually using. (We wrote more about what happens when approved software adds AI.)
The requests you’ll be asked to fund
Tools people prefer
Some of the fastest-growing costs come from preference, and task and project management tools are where we see it most. One team swears by one platform, another prefers a different one, and a few people keep their own to-do app on the side. Before long, the organization is paying for three or four tools that do roughly the same thing. Each request is reasonable on its own. Together they add license costs, scatter work across more places, multiply the accounts that need to be secured and shut off when people leave, and make it harder for anyone to see what’s actually in progress.
A light rule helps: any new tool needs an owner and a clear answer to “What does this replace?” If the honest answer is “Nothing, we’d just like it,” that’s worth a conversation before it becomes a renewal. Preferences are real, but they’re the most expensive kind of requirement.
Tools staff have built with AI
This is the newest category, and it comes up in nearly every planning conversation we’re having. Staff who are comfortable with AI are building things themselves: dashboards connected directly to Claude or ChatGPT, small apps, automations, tools published to a new website. Many are genuinely useful. Then comes the follow-up request for a web server to host it, an API account to run it, and a place for it to live. It’s a little like coming home with a puppy: the fun part happened first.
Whether to say yes depends entirely on the situation. Before the budget takes it on, someone should be able to answer a few questions:
- What data does it touch, and where does that data go?
- Whose account and API key is it running on, and whose card is paying for it?
- Where would it be hosted, and who keeps that secure and up to date?
- What happens when the person who built it leaves or gets too busy to maintain it?
- Would anyone notice, or be harmed, if it stopped working tomorrow?
Sometimes the right answer is to support it with a few guardrails. Sometimes it belongs with your IT partner or on an established platform. Sometimes a system you already pay for can do most of it. We walk through the full decision in Buy Versus Build in the Age of AI. For budgeting purposes, the point is simple: these tools are cheap to build and anything but free to own.
More vetting, and more security questions
Evaluating new products takes more time than it used to. Nearly every new tool now includes AI features that may send your data somewhere new, and platforms are attacked and breached with unsettling regularity. Before adopting something, it’s worth asking what would happen if someone gained unauthorized access to it, what its AI features can see, and how you would get your data back out if you needed to leave.
Those questions take staff or advisor time. Budget for that time, too, not only for the licenses.
Retire things properly
Replacing things means retiring things, and that’s the step that gets skipped. A laptop sitting in a supply closet with three years of email, client files, and saved passwords on it is a data breach waiting for someone to find it. Most organizations have a closet like that.
When you replace a computer:
- Wipe it properly before it’s reused, donated, recycled, or sold, using a secure erase or your device management tool. Deleting files and emptying the trash doesn’t count.
- Remove it from device management, security tools, and any licenses tied to it, so you stop paying for a machine nobody uses.
- Keep a simple record of what was retired, when, and how it was wiped. Auditors increasingly ask.
- Get a certificate of data destruction if a vendor handles disposal.
The same goes for cloud platforms. When you leave a tool, cancelling the subscription isn’t the end of it:
- Export what you need to keep, and confirm the export is complete and readable before you go.
- Delete your data from the platform, and check the vendor’s retention terms. Some keep data for months after you leave unless you ask.
- Disconnect integrations, revoke API keys, and remove accounts that still have access.
- Close the account itself, so it doesn’t quietly renew or linger as a forgotten login someone else could use.
A clean exit is part of the real cost of every tool. It’s also far cheaper than explaining, later, why old data turned up somewhere it shouldn’t have.
A practical planning sequence
If you’re building the 2027 budget now, this order tends to work. As a bonus, it leaves you with ready answers when your board asks why technology costs are going up:
- Pull every renewal into one list, with dates, costs, and notice windows.
- Decide what to retire or consolidate before anything renews. This is where most of the savings are.
- Set device standards by role (16 GB minimum, 32 GB for intensive work), and budget for 2027 prices, not 2024’s.
- Plan the exits as well as the purchases: how old computers get wiped, and how your data leaves any platform you drop.
- Create a single AI line item that pulls together suite tiers, add-ons, and individual subscriptions.
- Agree on an intake path for new tools and staff-built projects, so requests get the same questions every time.
- Name who owns these decisions. In many organizations without a CIO, nobody does, which is how auto-renewals and side projects end up deciding things by default.
The expensive decisions are the ones nobody makes
Most of 2027’s cost increases will happen whether or not anyone decides anything. The outcomes that hurt are the ones nobody chose: the renewal that went through unchanged, the five overlapping subscriptions, the side project that quietly became something the organization depends on. The work worth doing this budget season is making the decisions that are still yours.
If you’d like a second set of eyes on your 2027 technology priorities, talk with us.
Additional reading
Enjoyed this? Get the next one.
Signal, the FireOak newsletter: practical tips and thoughtful takes on technology, knowledge, and AI. No spam. Unsubscribe anytime.




