Why is AI adoption so slow? 🐌

Turns out it starts with a €264 billion infrastructure situation.

Hello hellooo,

Let me start with a number: €264 billion.

That’s what European organisations — universities, research institutes, companies, public bodies — spend every year on cloud services provided by non-EU companies. Mostly Google, Microsoft, and AWS (Amazon).

Now, your first reaction might be: okay, but what does that have to do with me?

Fair. You didn’t sign that contract. But your employer did.

But here’s the thing — if you use Gmail, Microsoft 365, any HR software, any project management tool, basically any software at work — you are already sitting on top of that infrastructure. You’re not responsible for it. You didn’t choose it. But you’re using it every day.

A special welcome to new joiners of this newsletter. 🤗
My name is Pranoti and I am the founder of The Science Talk.
My mission is to empower and enable scientists to share their stories in digital spaces with confidence and ease. 🧡

With AI entering the picture, this gets even more interesting.

AI needs data. A lot of it. And that data needs to live somewhere — on a server, in a cloud. So as your employer starts integrating AI into workflows (or tries to), the demand for cloud capacity goes up. Which means the dependency on these non-EU providers deepens further.

This is actually what made me go down this rabbit hole in the first place. I’ve been using AI tools in my work for the past three years, and at some point I started wondering: why is it so difficult for organisations to just adopt AI? Even when the potential is obvious?

Turns out, the answer isn’t just about ethics or compliance or data security — though those matter too. It’s also about the infrastructure. About what your employer’s digital setup actually looks like underneath. And cloud dependency is a big piece of that puzzle.

So in the latest episode of The Science Talk podcast, I go into exactly this — what the dependency looks like, where it sits, and why it’s more layered than just “cloud services.” There are multiple moving parts across cybersecurity and software that are worth understanding. Listen to the latest episode on Spotify 👇

Listen on Spotify

It won’t change what software your employer buys. But it’ll give you a much clearer picture of why AI adoption in European organisations is moving the way it is.

Cheers,
Pranoti Kshirsagar Founder, The Science Talk

PS: If, apart from my email, you also need help strengthen impact strategy, I think you’ll be happy about:

1. Impact strategy for research proposalswhere I work with principal investigators who want professional science communication that matches the ambition and budget of their grant proposal. More details are on my website.
2. ERC Starting Grant Resource Kit 47 official resources, organised in 11 sections including core webpages, ERC classes, compliance & AI guidance, and examples of winner grants - all in one pdf. You can access the kit here.
3. ERC Consolidator Grant Resource Kit47 official resources, organised in 8 sections including core webpages, evaluation & panels, videos & webinars, results & examples, compliance & requirements, and examples of winner grants - all in one pdf. You can access the kit here.
4. ERC Advanced Grant Resource Kit 44 official resources, organised in 8 sections including core webpages, eligibility extensions, official documents, compliance requirements, and examples of winner grants - all in one pdf. You can access the kit here.