Anthropic's recent $13 billion Series F round brought its valuation to $183 billion, the last sign of the escalating global AI arms race. The company plans to use the funding to scale enterprise products, expand globally, and deepen research on AI safety.
However, that international expansion was already well underway. Earlier this year, the company hired Guillaume Princen to be Head of EMEA. Princen will lead the company's expansion across Europe, the Middle East, and Africa.
Princen brings over 20 years of experience in high-growth technology environments, with expertise spanning revenue, product, operations, strategy, and international expansion.
I had a chance to speak with Princen back in July at the Raise Summit in Paris. The following is an edited transcript of our conversation.
Q: Can you tell us about your background and what led to your role at Anthropic?
GP: I joined Stripe in 2014, pretty early on the journey there, and built out the business in EMEA. I joined as the first employee outside of the US. That got me exposed to cutting-edge technology, innovation, B-to-B, and how companies use and consume innovative technologies. And in a regulated environment, right? Payments, in the same way as AI, is something that I think societies collectively want to see go well. So that's a topic I became pretty passionate about.
A few years later, I happened to know Daniela (Amodei, Anthropic president and co-founder, and former Stripe executive). We had a conversation. So here I am, joining Anthropic to build out the business in EMEA. We've got offices in London, Dublin, and we also announced a research office in Zurich. We're hiring, we're expanding tremendously in Europe. We're hiring over 100 people by the end of the year across all functions.
Q: Is that a combination of consumer and enterprise?
GP: It’s everything. The way Anthropic looks at this is that we are ultimately building products that businesses want. That is really our focus. So that ranges from native AI who are looking to build on AI, companies like Dust here in Paris. Another one is more of the enterprise, larger companies that are looking to transform. There are very different use cases. Companies like BMW, which are using Claude to streamline their R&D, or companies like Novo Nordisk, the creator of the Ozempic drug, which uses Claude to accelerate clinical trials and documentation of clinical trials that usually last months, are now done in weeks. And SNCF (the French railway system), which is building a customer agent to support its own customer agents with customer support.
If I were to focus on one, I would maybe mention agentic coding. This is the first real widespread agentic use case that we are seeing across the board. Claude Opus 4 reinforces Anthropic's lead in the coding space. I think what we're seeing on that end is really a sort of digital Renaissance because the building blocks of the digital economy have now become as simple as having a conversation.
What that means is, depending on who you are, you can do many different things. If you are an experienced developer, Claude Code or any agentic AI coding tool will help you to basically do much more. It will augment you, so it will help you to do 2-3-4-5 times more tests, more iterations, more products. If you're a big enterprise company and you've got a tech stack that's based on a 50-year-old coding language, you can suddenly rebase your entire code base on more recent languages that enable you to create much more velocity in the way you build products, which is not the case usually for big companies. And if you're not a coder at all, the whole vibe coding movement that we're starting to see, you can start building apps or products on your own by literally having a conversation with this agentic AI coding tool. So that coding, agentic kind of revolution is having an impact across the board from all the companies, whether you're a startup, a solopreneur, or an enterprise company.
Q: How do you see the competition with startups that say they are building coding tools?
GP: Claude 4.0 is the most powerful model in coding. That is what all the major coding agentic tools are built on. We at Anthropic are powering an entire ecosystem of startups, not only in coding, but also in the legal tech world, in the health tech world. We are powering hundreds of 1000s of startups in different verticals. Those startups are winning in their own verticals.
We are thought of as being the infrastructure for those startups winning in all sorts of verticals. We launched a powered-by-Claude repository, which is essentially a cross-tree of AI-native startups that are building on the Claude infrastructure for their own customers. All those companies are leading their sector, but building on Claude. So to answer your question, what we're seeing in the market is that Claude is powering many of those verticalized, expert solutions.
Q: In November 2024, Anthropic released the Model Context Protocol. MCP was, in retrospect, a secondary transformative movement for AI, with ChatGPT maybe being the first big one for the mainstream world. Is that fair, or is that hype?
GP: MCP is essentially solving what I call the "context crisis problem," especially for larger companies. Data is all over the place. We're working with 1000s of tools internally, and the data is disparate across teams. So the problem you're solving there is, "How do we enable those 10s of 1000s of people to access the right information for their day job, instantly in a structured manner?" MCP is solving that context problem.
Beyond that, I think the real question around agents is that an agent is seen as a virtual collaborator. That's usually how we define agents. So then the question is, "What is a virtual collaborator? What do you need a virtual collaborator to do?" You need a collaborator to a collaborator does two things. One, it needs to have the ability to work autonomously for long periods of time. Two, it needs to have the power to take action. When you combine those two things, you have a collaborator right now – especially with the launch of Claude 4. On the first point, Opus 4 can work for over seven hours autonomously. This is a full-day job. You've got the ability to take action with various integrations into your software tools. You can now also connect all this to your software tools to take action, so you now theoretically have the ability to have those virtual collaborators. Now, add on top MCP and you're adding context. So now you have it working autonomously for long periods of time. You have it taking action with all your software tools, and you have the context for it to make those decisions. We're really starting to have the right ecosystem and basic tools in place to do a lot of things. That's where we're seeing agents really take off right now, is the combination of those two things.
Q: What does MCP do for Anthropic's businesses?
GP: Anthropic has always been about B-to-B and solving problems for businesses. I think that's really important to understand. This is just more of a continuation. MCP is going to help solve those business problems. I don't have one customer conversation that does not talk about MCP. Every single large corporate today is wondering how to streamline customer service. How do I embed AI into my own products? How do I turbocharge my product offering with AI? Because ultimately, those companies are the ones who really know intimately what their customers want. We're not pretending to replace software companies. Software companies know what their customers want. We are enabling them to build whatever they want for their customers.
Q: Do you think people understand that shift from just cutting costs to rethinking the whole business?
GP: We're absolutely seeing that evolution from, "How do I solve these very well-defined, well-constrained, internal efficiency problems?" That's your internal chat, your internal marketing generation, your internal customer service. Now it's, "How do I think differently about my business? How do I think differently about creating value for my customers and embedding these tools into my products? How do I think differently about connecting through MCP, about connecting many parts of the company or my various software tools together to build more value internally?" It's become much less constrained to use cases and has moved to much more transformative thinking.
Q: You represent an American company that wants to succeed in a region where there is fear of losing its tech sovereignty to such companies. How do you address that?
GP: I talk to customers every day, and what I hear from my customers here in Europe is several things. One, I don't want you to be training your models on my data. Anthropic has the position that Anthropic does not train its models on customer data by default, which is different from most other providers. That's an important one. The second thing I hear a lot is, "How much can I trust your model? How safe is your model?" We’re talking about hallucinations, talking about safeguards, guardrails, etc. That ties back to the mission that Anthropic has. Anthropic is a safety-first AI company that builds safe, reliable, responsible models that are at the frontier.
What I'm seeing in the market today when I talk to customers, when I run them through what that means for us, what we do for that the trust element is extremely important for companies in this region, more than elsewhere in the world. We are in a business of trust here in EMEA, especially in AI, and that is something that customers value a lot.
