The cloud is not a chemical plant
There is a lot of recent discussion about digital autonomy and Europe’s “position in the cloud”.
Here, I want to break down and refute a commonly made argument: that the lead of American cloud providers is so great that we can never catch up. In the recent and excellent policy initiative “Clouds on the Horizon” (by Dutch political parties GL-PvdA and NSC), we can read on page 23 why this can’t be a reason to just give up. If things are as bad as people claim, that is all the MORE reason to get to work.
But I want to go a step further here. The US cloud providers (Amazon, Google, Microsoft) sell a phenomenal range of services—both in breadth (number of services), technical performance, and scale (billions of dollars per week). But even so, we are not hopelessly lost.
The European Chemical Factory
Many Dutch and European (chemical) factories have become completely outclassed and irrelevant due to cheaper (and better) suppliers from China. If you can produce para-aminophenol (needed for paracetamol/tylenol) here for a cost price of €7 per kilo, and someone in China offers it for $3.40/kilo, then your factory is simply done for.
Note that paracetamol sells in stores here for about €70 to €250 per kilo! We could afford our own factory. If we wanted to.
Photo by Crystal Kwok on Unsplash
A typical experience of being outclassed can also be read here: printed circuit boards for a few euros, produced within five days. We simply can’t do that in Europe, and we never could.
We often blame such situations on things like:
- China ignoring environmental regulations
- Use of Uyghurs in forced labor camps
- Shifting production to even lower-wage countries
- Illegal state subsidies to destroy our industry
- No zoning laws preventing new factories
- Lower energy costs
- They simply innovated harder and invented new things
But whatever the reason, if someone else can sell your product or chemical for far less than it costs you just to produce it, then it’s over in the short term.
Because no one buys your more expensive goods for that warm EU-feeling. We even let our entire medicine supply industry collapse over a few euros in price difference, part of why we now constantly face shortages. So we’re certainly not going to save your circuit board factory if it’s cheaper and easier to get stuff from China.
The European Cloud
There are people who believe we’re in the same situation when it comes to “the cloud.” But that’s simply not true. Essential in the Chinese example above is that those circuit boards and medicine precursors are much cheaper from China, cheaper than we could ever make them, and just as good or better.
The major US cloud providers offer a lot of services in one place, but many of those services aren’t cheap at all. In fact, if you want to rent servers and bandwidth today, you’re often much cheaper off with European providers, which is why companies like Leaseweb and Hetzner are doing excellent business with American clients.
In addition, with a few exceptions, all cloud services are also available in Europe—just not in one big “IKEA” store, which is admittedly a problem. But if you know what you’re doing, you can compose excellent services in Europe without relying on the US giants—and often for much cheaper. You just have to go against the grain a bit. You need to get on the phone (somehow) and you can’t assemble things with a few clicks (which is a damn shame).
It’s telling to ask cloud users how those promised cost savings after migrating to the cloud are going. For some reason, those savings never materialize, and everything even ends up being more expensive. This is also why people migrate back.
The main reason for this is that certain specific services in the cloud are outrageously expensive, and because of the legendary opacity and complexity of the pricing, you’ll end up consuming tons of those pricey services without even realizing it.
So, can we fix it here?
Even though our internet industry isn’t in the same dire state as the outclassed paracetamol factory, there’s still a lot of work to be done. China has indeed surpassed us in some areas and can now make things that are not only more expensive here, but where we don’t even know how to make them anymore—like building nuclear power plants within 20 years.
US cloud providers also offer a small number of services that simply don’t (yet) provably exist here, such as distributed multi-reader/writer databases, known as Google Spanner, Amazon Aurora, and Azure Cosmos.
But it turns out the vast majority of services consumed from big cloud providers aren’t that advanced at all. And we have all of those here too. You don’t need rooms full of special AI GPUs to send email, for instance. Yet we seem to think we do!
Where We Are
The scale of US providers is massive. Competing as a furniture business against IKEA is no fun, and that’s what it’s like being a European cloud provider going up against Amazon, Google, and Microsoft.
Right now, we still have an existing industry offering IT services here. And even though it’s not the same as what the US clouds offer, we’re still doing well in some areas. But a lot needs to happen now—otherwise we’ll lose it and be left with nothing.
Fortunately, the opportunity is still there, as long as we don’t let ourselves be discouraged by stories claiming the US lead can never be overcome. The bulk of cloud services are not out of our reach, and the opaque, complex, and enormous pricing of US providers leaves room for us to become relevant again.
In the piece But how to get to that European cloud? & followup posts, I describe a plausible path for Europe to catch up with US clouds, not by cloning what they did, but in a more European cooperative fashion.
But we do have to want it—if we don’t even find our own medicine production capacity important, the question is whether we’ll put in the effort to save our cloud industry.
Further reading
An overview of “the cloud” can be found in this Cloud overview article.