[Emmanuel M] : Hello Yann, and thank you for agreeing to this brief interview. To set the scene, could you introduce Scaleway in a few salient features, before getting to the heart of the matter?
[Yann Lechelle] : Thank you for the opportunity of this exchange. Scaleway is simply a cloud player, indeed probably not well known to the general public. However, Scaleway’s parent company is known as Iliad or Free. It is a fully owned subsidiary of the Iliad group which offers a complete service from the Data Center (Dedibox) to the public cloud (Elements) and which is what everyone is talking about. It’s the equivalent of what AWS (Amazon), Google or Microsoft do with Azure. Of course, Scaleway is much smaller than these players. We are a regional player, rather barycentric around Paris but with a presence in Amsterdam (1 data center) and Warsaw (1 data center).
[EM] : The reproach that is made to the French players is that they are not up to the level of the proposed functionalities. However, as the cloud is mainly a commodity, from what I understand, are you not in fact an equivalent player?
[YL] : This is a statement that must be qualified. The Americans obviously know how to do things well. They have the capital to do it well, but also to say what they do, and to shout it loud and clear. Then you have to distinguish between world-class companies, which have invested to go to and in each territory, and a regional player who is only known in its own area. It is clear that when we talk about Microsoft, it is an eminently respectable player that is about 40 years old and has a rich and complete offer. It goes from the Operating System to the cloud, it’s quite sprawling. It’s quite easy for an IT department to say to itself, we already work with Microsoft for the Office suite, and it’s easy to be seduced by the pitch that says we’re already doing that, and so if you take the whole package we’ll give you a small discount… This is what we call bundling from a commercial point of view. From a technical point of view, if we take the market leader, AWS, they were pioneers in 2014, on the public cloud. Scaleway, which used to be called online, was a pioneer in its time on bare metal servers, and the public cloud bare metal server, but we didn’t have the catalog depth that allowed us to say that we were an alternative to AWS 7 years ago. But today things are very different. On the one hand, France and Europe have not completely migrated to the public cloud; we are in an intermediate situation. On the other hand, when you migrate to the public cloud, you don’t need everything right away.
What is the cloud, you were rightly talking about convenience, it is someone else’s computer. In this computer we put either calculation or storage and transit of information. This computer is assembled in China with South Korean chips for the memory and American chips like Intel for the cpu. This computer is the same for everyone, the same for AWZS, for Google, for Microsoft and Scaleway, it is already a convenience. The advantage is that once you buy this computer, you own it, then you need a lot of software to share and consume this resource in the form of modern architecture. Well, that’s the complex part. Scaleway today employs 400 people, 300 of whom do just that, software to develop the means to consume this resource in a very elastic way with high value-added services.
The Americans have an extremely rich functional catalog, backed by their tens of billions of investments, and Scaleway has the common base to respond to at least 80% of use cases. According to Pareto’s law, with 20% of the products, 80% of the needs are met. If 2-3 years ago Scaleway was not in a position to meet this level, now we are!
Le cloud c’est une commodité pour preuve il y a 3 Américains, et même un quatrième peu connu qui est arrivé il y a 10 ans, Digital Ocean. Nous ressemblons beaucoup plus à ce dernier. On a aussi les Chinois qui arrivent. La bonne nouvelle c’est qu’en France nous avons trois acteurs du cloud.
The question is not to ban American players, but for many use cases, the French meet the specifications perfectly.
[EM] : For example, how do you position yourself in the choice of Microsoft Azure to host French health data (with all the issues related to the Cloud Act) ?
[YL] : To say that French actors cannot meet this need is indeed a borderline lie. It would be necessary, one, to tell us in what we are not capable openly. We are able to hear it. Are there missing functionalities, probably, but we have to push the intellectual reasoning to the end, why not decouple the storage part from the processing part and/or Artificial Intelligence for example. If so, in this case, couldn’t the storage be with French players because the notion of multicloud is a simple one, we don’t have to do everything with a single provider. Moreover, IT departments are used to working with multiple suppliers. For the Health Data Hub, the prototype was launched at a time when perhaps the right partner was Microsoft. But then it was promised that a call for tenders would be issued, and we are still waiting for it. This brings us back to the question and the lie, the French don’t have what it takes.
The French have what it takes! The proof Scaleway has been around for 15 years and has tens of thousands of customers who trust it. Scaleway is a trusted cloud.
[EM] : I particularly liked your proposal that rather than talking about and defining “trusted cloud”, rather define what is not trusted. What would you consider to be the three major conditions that would define this lack of trust?
[YL] : So, when I analyze this French doctrine of the cloud at the center, I was shocked by the announcement of May 17, 2021 because structures that do not yet exist to date were put forward. Moreover, it does not seem to me to be trustworthy with respect to extraterritorial laws.
For me, a cloud that is not trusted is:
- A cloud whose data is elsewhere where sovereignty is important. If the data in our case is not in France, this is not a guarantee of confidence
- The operator of the entire stack (from the data center to the software) is subject to extraterritorial laws (Cloud Act or FISA), so there is no reason to trust.
- Software control is a subject, you have to be able to audit the code, to make sure there is no backdoor or data leakage. By definition, the cloud is always connected.
These three elements are red flags to have confidence, then it is up to the market to organize itself, not to the government to say what is the cloud of confidence. In France we are lucky to have three, which is not the case for many other European countries.
[EM] : Can we envisage a purely private initiative on this subject?
[YL] : We have launched an initiative within Euclidia, an association with an educational purpose, to make people understand that within Europe we have this software expertise on the subject of the cloud. We also set up pairs of cloud + European SaaS solutions, sometimes open source and sometimes not. So there are proposals for a sovereign cloud. Scaleway is a sovereign cloud, we own our data centers, our software and we are French owned. Why should Scaleway have to justify being sovereign, are we questioning the sovereignty of Microsoft in the United States?
[EM] : The reproach often made to the French Cloud players is that they do not have the necessary and performing tools for Big Data and Deep Learning. How to fight against these prejudices, which I believe to be unfounded?
[YL] : The Data Hub Santé may be doing “Rocket Science”, I don’t know, but we have start-ups and scale-ups that require large computing capacities that we are able to meet. So yes, I think we are capable of meeting the needs of the Health Data Hub. But they would still have to tell us what they need.
[EM] : Another highlight of the last few weeks was the announcement by Thales of a collaboration with Google. Couldn’t we have had an equally fruitful collaboration between Thales and Scaleway?
[YL] : I think that France suffers from a “CAC Quarantien” tropism, which is a bit of a problem for the country, with a lot of dependence on the ESNs, which do things instead of. We have lost this fundamental engineering capacity in France, it’s a slippery slope, we have them do it… Thales is still an entity that can do it and is an important player in our national defense sovereignty.
Moreover, I am not a sovereignist, but rather a liberal protectionist. In order to be in the race, we must develop and protect our SMEs. And France does not know how to do that. We talk about the Smal Business Act, it doesn’t work in Europe, it doesn’t work in France, at least not in the digital sector. We may have a network of efficient SMEs in agriculture, but in tech it doesn’t work.
So, I don’t have any particular opinion on the Thales-Google alliance, it doesn’t shock me because Google can very well propose a particular brick. We should dig deeper into this partnership. On the other hand, this announcement is in line with previous ones, such as Capgemini – Orange, where the model of the moment is transposed, the licensed version. For me, this model is not a trusted cloud, it is precisely an offer that does not exist and if it does exist, it is not trusted because the software is not sovereign.
[EM] : Don’t you think that large institutional or private actors such as CCIs, communication companies, and large French ESNs are harming our digital sovereignty by advocating exclusively #MAGAF solutions?
[YL] : I think it’s a dynamic, I don’t blame them. However, there is a global awareness. The notion of playing collectively cannot work alone, “business is business”. If the consultants of these large ESNs are trained at Microsoft, and the prices have been negotiated, and Microsoft charges a lot of money for a lot of elements, there is a whole engine that will maintain this rather inefficient machine. Because there are a lot of intermediaries, so it can’t come from them.
On the other hand, I’m having very interesting conversations with some of these NSEs, not necessarily those that have already made these major alliances with the Americans, but there again it’s a question of affinity and size. We don’t claim to be Microsoft, Scaleway is a regional player with a commodity base that is quite respectable, but we don’t claim to be Microsoft. So there are NSEs that correspond to us in terms of size, that have customers who want a sovereign cloud. The notion of playing collectively comes from the demand, if the French and European demand asks for more diversity and if we have been clear about the geopolitical dangers, things can move. If these famous clouds, which are announced with great fanfare, are in place and used, if the United States, which has a doctrine of America first, decides to use all of its resources to weaken Europe so that we don’t bother them and attack a superpower, overnight they can cut everything off.
To conclude on Gaia-X, this initiative does not address the issue of sovereignty at all. My rather sudden decision to leave the consortium was the result of a long analysis and the realization that Gaia-X does not address the issue of sovereignty. Moreover, the announcement of the Gaia-X management even shocked me, which is to promote the growth of the cloud but the market is self-sustaining at 30-40%, finally if it is to reinforce the status quo and allow the dominant players to continue to dominate (ESN – Big Tech) it is useless.
[EM] : Having followed your various interventions, leaving Gaia-X allows Scaleway to refocus on its development and to benefit from a good dynamic, can you enlighten us with some figures?
[YL] : We are very focused, but we need to be even more focused, and the market expects only that from us. That is, to provide solutions on the commodity base, and to provide innovation as well. But a structure like Gaia-X wants everything but innovation. They want a common base, but it will not be common. This common denominator will be as high as possible to be a barrier to entry, which the dominant players want. Small and medium-sized players are welcome, but in the passenger seat…
[EM] : In France, all of your data centers are located in the Paris region. Do you have any plans to set up in the regions? If so, please specify which ones?
[YL] : Scaleway has four data centers in the Paris region and three in AZ, a zone of availability, we have two data centers (DC) that are twinned. Scaleway is the only European player to have a complete region in France. A complete region in the sense of the modern cloud, as defined by AWS, which consists of having 3 zones of availability with total redundancy in a given region. The other two players are AWS and Microsoft. Scaleway also has a presence in Amsterdam and Warsaw, and there too we intend to offer complete regions. And that goes without saying, because that’s what a public cloud is all about.
[EM] : I was thinking more of the provinces, to reduce the risk of a presence in the Paris region?
Having locations in the regions is quite relevant. There are several notions, our DCs are in regions with varying risks, but this notion of region (DC AZ) allows us to reduce the impact of an incident on a DC. However, if our DCs are in a flood zone, they are elevated, limiting the risk. One of our DC is 45mn from Paris, it is our flagship of technological innovation. It is the most responsible one using an adiabatic cooling system.
In absolute terms, the consumption of DCs is enormous, but in relative terms, the more elements you put in a DC, the more efficient you become, especially in terms of energy. In particular compared to a poorly managed and distributed computer system.
On the question of locating DCs in the provinces, one of the elements to take into account is that France is extremely well meshed in terms of communications infrastructure (Telecom, fiber). Smaller DCs spread throughout France could make sense, especially in terms of edge computing. It’s a complex question about the granularity of the computing unit, the storage unit, how much we put in and where? To get closer to the user. Edge computing is the subject in the making, and it is part of our objectives and reflections. The challenge is the 5ms between the mobile application and the DC.
[EM] : We are coming to the end of our interview. What perspectives would you like to give to conclude?
[YL] : The outlook is quite good. Scaleway was not known, and yet it is a company that is 15-20 years old. It’s up to us to make some noise, to make ourselves known. But above all, it’s up to us to show that we have the right products for the right use case. We have many customers in France and in Europe. We also have customers in 160 countries. We probably have interesting things for everyone. France has the capacity to meet these needs. Let’s be crazy, let’s imagine that France and Europe provide up to 30% of these cloud needs.