The largest AI companies and the real estate of tomorrow

The largest technology companies in the world are shifting their center of gravity toward artificial intelligence. They are investing in chips, computing capacity, data centers, models, networks, energy and talent. AI is thereby evolving from a software trend into a new infrastructure layer of the world economy, a layer that is becoming as fundamental as electricity or telecom.

This article starts with ten of these AI giants. But the companies are not the real subject. They are evidence of a larger movement. Because if billions are flowing into the digital infrastructure behind AI, how much attention are we paying to the digital infrastructure of the buildings where AI companies, and soon virtually all companies, do their work?

We look at data centers, GPUs, energy and cloud capacity. But we rarely look at the last meters of that digital chain: the building. That is precisely the blind spot that IT-Label wants to make visible.

What does "the largest" mean?

Before reading a list, it is fair to state that "the largest" is not a straightforward concept. A ranking by market capitalization puts publicly listed giants at the top, while private AI companies fall out of view. Those are instead judged on private valuations, revenue, funding, user numbers and their position in the AI chain.

That is why we present ten of the largest and most influential AI companies below, not an absolute ranking. Market capitalization of listed companies and private valuations are different metrics and cannot be compared one to one. The selection looks at the role in the chain, from chip to user.

Ten AI giants, different roles

What stands out about the companies below is that "AI company" is a broad term. The AI economy does not consist only of organizations that build chatbots, but of a complete chain: chips, compute, cloud, models, software, data and the user.

• NVIDIA (Santa Clara, USA): designs the GPUs that power virtually all heavy AI training. Position in the chain: chips and compute. One of the highest-valued publicly listed companies in the world and thereby the foundation under AI hardware.
• Microsoft (Redmond, USA): combines cloud (Azure), AI models and widely deployed software. Position: cloud to software. Relevant because of the integration of AI into work tools used by hundreds of millions of people.
• Alphabet / Google DeepMind (Mountain View and London): own models, search engine, cloud and leading research. Position: research, models, cloud and data.
• Amazon / AWS (Seattle, USA): the largest cloud provider in the world and thereby the computing base for countless AI companies. Position: cloud infrastructure.
• Meta (Menlo Park, USA): develops large, partly open models and invests heavily in AI infrastructure. Position: models, data and user.
• OpenAI (San Francisco, USA): private company with one of the highest private valuations in the sector, known for widely used generative models. Position: models and user.
• Anthropic (San Francisco, USA): private AI lab focused on safe, reliable models. Position: frontier models. Judged on funding, revenue and impact.
• xAI (USA): young AI lab with rapid growth and its own model development. Position: models and compute.
• TSMC (Hsinchu, Taiwan): produces the most advanced AI chips in the world on behalf of, among others, NVIDIA. Position: chip production, the literal foundation of the chain.
• Broadcom (Palo Alto, USA): supplies network and custom chips that make AI data centers possible. Position: chips and network infrastructure.

From chip to user: CHIPS → COMPUTE → CLOUD → MODELS → SOFTWARE → DATA → USER. Each link is different, but they share one characteristic. They all depend on physical infrastructure.

We know exactly which companies are driving the AI revolution. The question no one asks: which buildings are ready to house that revolution?

AI has a physical reality

AI feels digital, but is surprisingly physical. Behind every model sit chips, servers, data centers, fiber optics, networks, electricity, cooling and hardware. And at the end of that chain stand people, at a workplace, in a building. As we argued earlier, a building without IT is like a body without a nervous system.

There is nothing virtual about the infrastructure behind artificial intelligence. The cloud still needs a cable. Even the most advanced AI organization remains dependent on a physical connection that enters a building somewhere.

From data center to workplace

A large part of heavy AI compute takes place in specialized data centers and cloud environments. But the people who develop, implement, sell and use AI work from offices, campuses, labs and flexible locations. That is where the final link in the chain begins.

That chain runs from workplace → building infrastructure → connectivity → network → cloud/data center → AI. We invest billions in the middle of that chain. But how much do we actually know about the starting point, the place where the person enters the digital economy?

Can a billion-dollar company work from a digitally mediocre building?

Suppose an international AI company opens a new location with hundreds of developers, engineers, data scientists and researchers. They use AI agents, cloud development, video communication, large datasets, real-time collaboration, security tools and countless connected devices.

The broker presents the building: ten thousand square meters, energy label A++++, three hundred parking spaces, BREEAM Excellent, excellent location. Then the question comes up: what about the digital infrastructure? And the answer too often is: "There is internet." That is exactly where the blind spot lies. In many rental brochures, the digital infrastructure is completely absent, even though it is precisely that infrastructure that determines usability.

From energy-intensive to data-intensive real estate

The real estate sector has learned to look at energy consumption, grid capacity, sustainability and grid congestion. Rightly so. But organizations are simultaneously becoming increasingly data-intensive. If we measure the energy consumption of buildings, why do we barely look at the digital needs of their users? Energy and IT are in fact two equal foundations of modern real estate.

That is why the concept of data consumption is useful. This does not simply mean "how much internet someone uses". It concerns the complete digital usage profile: connectivity, cloud use, real-time applications, devices, security, redundancy and continuity combined.

Not every AI company has the same needs

It is tempting to think: AI company means extremely fast connection. The reality is more nuanced. A frontier lab training models has different requirements than a chip designer, a cloud provider, an AI software company, a consulting firm or an organization that mainly uses external AI APIs.

That is why IT-Label asks a more relevant question: what is the digital usage profile of the user, and does the digital delivery level of the building match it? That is far more useful than simply asking whether fiber optic cable is present. You can read more about this mechanism in what IT-Label actually assesses.

What is an AI-ready workplace?

The term AI-ready workplace is not an official certification, but a way of thinking about the workplace of the future. A digitally future-proof building requires attention to a number of layers.

• Connectivity: fiber optics, capacity, multiple providers and network redundancy.
• Internal infrastructure: data cabling, enterprise WiFi, technical rooms and network facilities.
• Continuity: backup connections, emergency power, monitoring and outage procedures.
• Cybersecurity: physical access, segmentation, clear responsibilities and management.
• Smart building: IoT, sensors, building data, access control and smart climate and lighting systems.
• Scalability: can the infrastructure grow along with the company?

AI-first company, digital-first building?

An organization can be fully AI-first, cloud-first and data-driven. But if the physical workplace provides no insight into the digital infrastructure, a strange contradiction arises. AI-first companies call for digital-first real estate.

Yet not every building needs maximum specifications. The building should match the needs of the user. That is why transparency matters more than simply adding more technology. Not every building needs to be IT1+ PREMIUM, but every user should know what they are digitally getting.

The role of IT-Label

IT-Label makes the digital delivery level of real estate visible and understandable through a classification: IT1+ PREMIUM, IT1 HIGH PERFORMANCE, IT2 PLUG & PLAY, IT3 READY, IT4 CORE and IT5 SHELL. The label does not indicate how much AI a company can run and does not automatically certify cybersecurity or suitability for every organization.

What it does do is make visible what is present, what is delivered, who is responsible and what the user needs to arrange themselves. The user determines the digital need. IT-Label makes visible what the building delivers digitally. You can read more about this in the explanation of what the IT-label is.

What if these companies wanted to rent your building tomorrow?

Suppose one of these AI giants tomorrow looks for five thousand square meters of office space in your building. Could you answer these questions within a single day?

1. Which fiber optic connections and providers are present, and with what capacity?
2. Is there redundancy and what is the mobile coverage like?
3. What data cabling is in place and how is the technical room set up?
4. What smart building systems are there and who manages the infrastructure?
5. Which certificates and inspections are available, and what does the tenant receive digitally upon delivery?

And the most important question: can you prove it? That is the shift IT-Label aims to bring about. From "I believe it is well arranged" to "this is the digital delivery level of the building". For owners who want to get started with this, the page on IT-Label for office buildings is a good starting point.

AI companies are merely the vanguard

Ultimately, this is not about ten companies. They are the vanguard. Banks, law firms, real estate agents, accountants, healthcare organizations, logistics companies and governments already all use AI. Today we ask what an AI company needs digitally. Tomorrow that may simply be what every company needs.

The best-known real estate saying, location, location, location, still holds true. But in the digital economy, an extra dimension is added: location, connectivity, capacity. Location determines where you work. Connectivity determines whether you can work. Digital quality does not replace location, it becomes an additional quality dimension of it.

The digital value of real estate

As the economy becomes more digital, does digital quality become part of real estate value? That is an exploratory question, not a proven claim. It is conceivable that digital infrastructure will influence rentability, that transparency will speed up a rental process, and that data-intensive users will come to prefer certain buildings. It is also conceivable that digital infrastructure will become part of due diligence and that appraisers will look at it more explicitly. We explore this connection further in digital infrastructure as a value driver of real estate.

Take the first step toward insight

The largest AI companies show where the economy is moving. The next question is which buildings are ready to house that movement. AI is changing how we work, how we work is changing our workplace, and our workplace is ultimately changing our real estate.

Do you want to know what your building can handle digitally? Start by mapping out the current situation: what connectivity, capacity and responsibilities are already in place. Take a look at how the IT-label works to see how that situation is translated into a clear delivery level. Is your building ready for the companies of tomorrow?