Opinion: UK SMBs Are Funding AI's Energy Crisis and Nobody Asked Permission

Here's a question for your weekend: Did anyone ask if UK small businesses wanted to fund Microsoft's nuclear reactor restart?

Because that's what's happening.

While Microsoft spends $1.6 billion restarting Three Mile Island, Google partners with Kairos Power for 500MW of small modular reactors, and Amazon secures nuclear capacity across multiple projects, your cloud bills are climbing to pay for infrastructure you didn't vote for.

Nobody took a vote.

Nobody asked permission.

Tech giants made a collective decision that AI is worth unlimited energy consumption, and UK SMBs are involuntary investors in that bet.

Let's talk about that.

The Infrastructure Tax You Didn't Sign Up For

When you signed your AWS Enterprise Agreement or Microsoft 365 subscription, here's what you thought you were buying:

• Cloud computing resources

• Business productivity tools

• Scalable infrastructure

• Pay-as-you-go pricing

Here's what you're actually funding:

• Microsoft's Three Mile Island nuclear restart ($1.6 billion)

• Google's small modular reactor development with Kairos Power

• Amazon's nuclear capacity acquisitions across multiple sites

• The entire AI industry's exponential energy demands

Your monthly cloud bill isn't just infrastructure cost anymore. It's an energy infrastructure tax.

And nobody told you.

The AI Bet You're Forced to Take

Let me be crystal clear about what's happening:

Tech giants decided AI is the future. Maybe they're right. Maybe ChatGPT and Copilot and Gemini will revolutionize everything. Maybe generative AI will solve problems we haven't imagined yet.

But here's the thing: They made that bet with your money.

Every time your Azure bill increases 12% year-over-year, you're funding Microsoft's bet that AI workloads are worth restarting a nuclear reactor that's been dormant since 2019.

Every time AWS adds AI-enhanced services you didn't ask for at premium pricing, you're funding Amazon's bet that machine learning infrastructure justifies massive energy consumption.

Every time Google Workspace adds AI features to your subscription whether you want them or not, you're funding their small modular reactor development.

Did you agree to this?

Did anyone ask if UK small businesses thought AI was worth nuclear power?

Did your 15-person marketing agency in Manchester get a vote on whether Microsoft should restart Three Mile Island?

No.

Because you're not a customer. You're a funding source.

FinalSpark: The Path They're Not Taking

This week we covered FinalSpark's Neuroplatform: 16 brain organoids with 160,000 living neurons that compute using one million times less energy than silicon processors.

Let that sink in.

Biological computing exists. Today. Right now. It's not science fiction.

FinalSpark operates a real biocomputing platform in Switzerland where researchers can remotely access living neurons for computational tasks. The energy consumption is so low it's almost negligible compared to conventional processors.

If you can compute with neurons at one millionth the energy cost, why are tech giants doubling down on nuclear power?

Because they've already invested billions in silicon infrastructure, AI model architectures, and data centres designed for conventional computing. Biocomputing would require starting over.

So instead of pursuing radically more efficient technology, they're restarting nuclear reactors.

And passing the cost to you.

The Energy Hypocrisy

Remember when tech companies were falling over themselves to announce carbon neutrality commitments?

• Microsoft: Carbon negative by 2030

• Google: Carbon-free energy by 2030

• Amazon: Net-zero carbon by 2040

Then AI happened.

ChatGPT launched in November 2022. By 2024, those environmental commitments hit reality: AI workloads consume so much energy that renewable capacity can't possibly keep up.

The response? Nuclear power.

Not because nuclear is clean (it produces radioactive waste we still don't know how to dispose of safely). Not because nuclear is safe (Three Mile Island had a partial meltdown in 1979).

But because nuclear provides gigawatts of baseload power to keep AI training runs going 24/7.

Who pays for this pivot?

Customers. Specifically, the customers least able to absorb cost increases: small and medium businesses.

The UK SMB Perspective

Let's bring this home to UK small businesses specifically.

Your situation:

• Operating margins already squeezed by post-Brexit complexity

• Rising employment costs from National Insurance increases

• Energy bills that increased dramatically in 2022-2023

• Commercial rent increasing in most UK cities

• Competition from larger businesses with economies of scale

Your cloud spending:

• AWS/Azure costs up 12-15% year-over-year

• SaaS subscriptions adding AI features at 20% premiums

• Microsoft 365 pushing E3/E5 licenses for "AI capabilities"

• Vendor lock-in making it difficult to switch providers

Your leverage: None.

You can't negotiate better pricing because you're too small. You can't move to competitors because they're all doing the same thing. You can't opt out of AI features because they're bundled into base products.

You're a captive market funding an energy transition you didn't request.

The "Innovation Tax" Defence

I can already hear the rebuttals:

"AI will create tremendous value for businesses!"

Maybe. But UK SMBs aren't seeing that value yet. Most small businesses are using AI for basic content generation and customer service chatbots. They're not training large language models or doing advanced machine learning.

The value proposition doesn't match the cost increase.

"Cloud infrastructure has always had price fluctuations!"

True. But previous price increases were tied to capability improvements: faster processors, more storage, better networking. The current increases are infrastructure taxes for energy consumption your business doesn't directly benefit from.

"If you don't like the pricing, build your own infrastructure!"

This is laughable. The entire cloud value proposition was "don't build your own infrastructure, use ours." Now that UK SMBs are locked in, vendors are increasing prices because they can.

Switching costs are too high. They know it. You know it. They're using that leverage.

"Nuclear power is necessary for AI advancement!"

Maybe. But that's a societal decision, not one that should be made by three American tech companies and passed through as customer cost increases.

What This Actually Means for Your Business

Let's be practical. What are the actual implications?

Short Term (2025-2026)

Expect cloud costs to increase 10-15% annually regardless of your usage. This isn't market forces. This is infrastructure cost pass-through.

Budget accordingly. Don't assume your current cloud spending is stable. Build in annual increases.

Optimize aggressively. Every pound you waste on unused resources is a pound funding nuclear reactor restarts you don't need.

Medium Term (2027-2028)

More aggressive vendor lock-in. As tech giants invest billions in energy infrastructure, they'll work harder to keep customers captive to recover those costs.

AI feature bundling will intensify. Expect base products to include AI capabilities whether you want them or not, with pricing to match.

Alternative providers will struggle. Smaller cloud providers can't compete with the scale required for AI infrastructure, reducing your options.

Long Term (2029+)

Two possible futures:

Future 1 (Optimistic): FinalSpark-style biocomputing reaches commercial viability. Computing becomes radically more energy-efficient. Prices stabilize or drop. Nuclear reactors built for AI become stranded assets.

Future 2 (Realistic): Silicon Valley's nuclear bet pays off. AI becomes genuinely transformative. The energy infrastructure cost becomes accepted as "the price of progress." UK SMBs continue paying premium rates while large enterprises negotiate better deals.

Which future we get depends on factors entirely outside your control.

The Question Nobody's Asking

Here's what bothers me most about this whole situation:

When did we collectively decide that AI advancement was more important than affordable computing for small businesses?

Because that's the trade-off being made. Every gigawatt of power dedicated to training AI models is power that could run thousands of small business workloads at lower cost.

FinalSpark's biocomputing platform proves there's a more efficient path. But that path requires tech giants to admit their current architecture is fundamentally wasteful.

They won't do that.

They've invested too much in silicon. Built too many data centres. Trained too many engineers on conventional computing architectures.

So instead, they're restarting nuclear reactors and sending you the bill.

Your Realistic Options

I'm not going to pretend you have good choices here. You don't.

Option 1: Accept It and Optimize

This is what most UK SMBs will do:

• Accept that cloud costs will increase 10-15% annually

• Implement aggressive cost optimization (see Thursday's post)

• Budget accordingly

• Hope biocomputing eventually changes the economics

Pros: Realistic, achievable, doesn't require complete infrastructure overhaul

Cons: You're still funding nuclear reactors you don't want

Option 2: Minimize Cloud Usage

Bring workloads back on-premises where practical:

• Host simple web applications on dedicated servers

• Use local storage for non-critical data

• Reduce dependency on cloud-native services

Pros: More control over costs, less vendor lock-in

Cons: Higher upfront investment, requires IT expertise, loses cloud flexibility

Option 3: Vote With Your Wallet (Eventually)

Support alternative providers when they emerge:

• Watch for European cloud providers offering better pricing

• Consider open-source alternatives to proprietary platforms

• Participate in industry advocacy for fair pricing

Pros: Long-term market pressure might force change

Cons: Won't help your 2025-2026 budget, alternative providers are limited

Option 4: Get Angry

This is what I'm doing.

Write about it. Talk about it. Make noise about tech giants using small businesses as involuntary energy infrastructure investors.

Will it change anything? Probably not.

Does it matter? Yes.

Because the more UK SMBs realize they're funding AI's nuclear ambitions, the more pressure there is for regulatory intervention or market alternatives.

The Weekend Thought Experiment

Imagine Microsoft came to you directly and said:

"We want to restart a nuclear reactor to power our AI research. It'll cost $1.6 billion. We'd like your business to contribute £8,000 over the next five years through cloud price increases. In return, you'll get access to AI features you may or may not need. Sound good?"

What would you say?

I'm guessing: "Absolutely not."

But because the cost is hidden in cloud billing increases and AI feature bundling, you don't get that choice.

That's the problem.

The Uncomfortable Truth

Here's what I believe after 40+ years in technology:

AI will be transformative. Eventually.

Nuclear power might be necessary. For now.

But UK SMBs shouldn't be involuntary investors in that bet.

If tech giants want to restart nuclear reactors, they should fund it from profits, not customer price increases. If they want to bet billions on AI, they should take that risk themselves, not pass it through to captive small business customers.

The current model is extraction, not innovation.

FinalSpark's biocomputing platform exists. It works. It uses one millionth the energy. But it won't reach commercial viability for years because the industry is too invested in silicon to pivot.

In the meantime, your cloud bill pays for nuclear reactors.

And nobody asked permission.

What I'm Doing About It

For what it's worth, here's my personal response:

1. Writing about it. Obviously. This post is part of that.

2. Implementing aggressive cost optimization for every client I advise. Every pound saved is a pound not funding nuclear restarts.

3. Supporting alternative technologies. I'm watching biocomputing development closely and will advocate loudly when it becomes commercially viable.

4. Calling out vendor behaviour. When Microsoft, AWS, or Google implement pricing that unfairly targets SMBs, I'll say so publicly.

5. Educating UK businesses on what's actually happening with their cloud bills so they can make informed decisions.

Will this change the industry? No.

Does it matter? I think so.

Because the more people understand what's happening, the more pressure there is for change.

Your Weekend Action

Don't spend your weekend angry about cloud pricing. That's not productive.

Instead, spend 30 minutes understanding where your money goes:

1. Pull your last three months of cloud bills

2. Identify which costs increased most significantly

3. Check if those increases correlate with AI feature additions

4. Calculate your annual trend

Then on Monday, implement one optimization from Thursday's post.

You can't control whether Microsoft restarts nuclear reactors.

You can control how much of your budget they get to do it.

Source	Reference
Microsoft	Three Mile Island Nuclear Power Partnership Announcement, September 2024
Google	Kairos Power Small Modular Reactor Partnership Announcement
Amazon Web Services	Nuclear Energy Investment Announcements 2024-2025
FinalSpark	Neuroplatform Biocomputing Research Platform Documentation
Flexera	State of the Cloud Report 2025 - Price Increase Analysis
Gartner	Cloud Pricing Trends and Infrastructure Investment Report