The search for 'tech for good'

I’ve a meeting with a ‘tech for good’ consultancy later in the week. I think we share a lot of similar ideas and values, and the discussion could be interesting. I hope they don’t mind the ‘tech for good’ label.

It made me think about this as a category, and how much gets spent on it probably being pitifully low.

‘Tech for good’ gets said the way sustainable used to get said, as though categorising does most of the work. It sits on conference agendas and LinkedIn bios and the About page of companies that would struggle to say which of their own decisions it describes. Ask people in the room how much of it there is, in money, in reach, and the straight answer is that few quite know, but many assume it's substantial and growing, because the alternative, that it barely exists, is uncomfortable enough that nobody tests it.

I tried testing it with some research.

• Tech Nation's one attempt to size the UK version, in 2018, found around 490 companies calling themselves tech for good, turning over £732 million between them, valued at over £2.3 billion.
• The Global Impact Investing Network puts the whole of impact investing, not just tech, at $1.57 trillion in assets under management.
• Climate tech pulled in something between $56 billion and $79 billion a year at its recent peak, according to PwC and CB Insights.
• Digital health raised $14.2 billion in the US in 2025, more than half of it going to AI companies, according to Rock Health.
• EdTech peaked near $20 billion in 2021, per HolonIQ, and has fallen since.

And of course on my doorstep there is the ESRC Digital Good Network, a £4 million research programme led by Sheffield with Cambridge, Oxford's OII, Sheffield Hallam, Lancaster, UMass Amherst and Minderoo, plus partners including the BBC and Birmingham Museums Trust. Its stated job is what does a good digital society look like and how do you tell.

Set any of that against Gartner's estimate of $6.15 trillion in global IT spending for 2026 and the sector looks tiny.

Except that comparison is wrong before you even get to the numbers, because it assumes both sides are counted in the same currency, but they aren't. Most of that $6 trillion is infrastructure and platform spend, cloud, data centres, enterprise software, identity systems, the plumbing everything else runs on.

That spend was never eligible to be ‘for good’ in the first place, and not because it fails some test. Infrastructure is built specifically to not have a purpose, so that anything can be built on top of it. Asking whether a cloud platform or a mobile operating system is tech for good is the same error as asking it of the electricity grid, which runs a dialysis machine and a taser without caring which.

The claim can only be made where a technology meets a defined consumer for a defined purpose, at the point of use, the specific app or service, not the infrastructure it sits on. Trevor Baylis's wind-up radio is at that point: one device built in 1991 after he watched a programme on AIDS in Africa, where health workers said radio education would help but much of the continent had no reliable power. It carries one purpose and answers for it. A smartphone carrying both a diagnostic app and an engagement-optimised feed on the same device is not, which is why the claim must attach to the app, not the phone.

Now, this is new territory for me and my plan was to then move this to how the architect can inform this, but I am still exploring and that will need to come later. I thought I would publish this to test the thoughts of others.

That doesn't let the infrastructure off entirely; it just changes what it can claim. A cloud provider can run its data centres on renewable power, treat its workforce fairly, and pay its taxes where it earns its revenue. That's worth having, and worth calling ethical provision but it's a claim about the provider's conduct, not about the technology being for good, and folding one into the other is probably the move that lets a virtuous-sounding company launder an ordinary product, or an unremarkable platform bask in the glow of whatever gets built on top of it.

The same two questions apply once you're at the point of use, and they can point in opposite directions. Does the technology itself do something different, expand who's served or what's possible, not just who's buying? A washing machine engineered to work without mains power or reliable water pressure is doing that. A washing machine that's unchanged, sold by a company that merely isn't destroying the planet, is not; that's a good company selling an ordinary product, worth having but a different claim.

Also, does the value it creates stay with the people it was meant to serve, or get extracted back out as margin? Housing associations in England have been criticised for launching profit-making subsidiaries, in facilities management and maintenance, that sit inside an otherwise mission-locked group and are reported to have top-sliced funding meant for tenants. The parent's mission is real; it doesn't reach the subsidiary.

C.K. Prahalad made the equivalent claim about multinationals and the ‘bottom of the pyramid’, that selling ordinary products to poor consumers was itself a form of development. Aneel Karnani's rebuttal, in 2007, was that this was mostly a mirage; the profit still had to come from somewhere, usually the customers the pitch said it was helping. Both are the same failure at different scales; a virtuous label checked against the structure it claims rather than the one it sits inside.

So the position on spend isn't a small percentage; it's that few have ever counted the right thing. The number that would matter, spend specifically at the point of use, against a defined consumer, tested against whether the value it creates stays with the people it serves, may not exist. It might be small, but it might not be, but my limited research suggests the data has not been cut that way.

Which is also connected to why this can't be fixed by moving money between categories that were never the same pot, a bank's marketing budget and a healthcare device company's R&D spend, say. Asking for that rebalancing is not going to happen. For example, recycling became normal because someone built a second bin before making the moral argument and made the good choice cost almost nothing once it existed. Tech doesn't have that ‘second bin’ yet, and there's no equivalent low-friction default sitting at the point of use, waiting the way a kerbside collection waited outside everyone's door.

Until something like it exists, whatever rebalancing happens will be decision by decision: an investor checking where the value lands before committing capital, a buyer asking what a vendor's premium pays for before signing, a designer asking whose discretion a design takes away before it ships. That’s going to be slow progress, but it's what's available while nobody's built the easier thing yet.