The AI boom is not spreading value evenly across the digital economy.
It is making the largest platform companies larger.
An analysis by Prof. Dr. Holger Schmidt and Hamidreza Hosseini found that the world’s Top 100 platform companies reached a combined known market value and valuation of $22.7 trillion as of December 20, 2024. The same analysis shows a striking regional concentration: American platforms accounted for about 86% of that value, Asia-Pacific for 11%, Europe for 2% and Africa for 1%.
Image and data credit: Prof. Dr. Holger Schmidt / Hamidreza Hosseini, 2025.
The important point is not only that platform companies are valuable. That has been obvious for years.
The important point is that AI is strengthening the platform model instead of replacing it.
AI Rewards Platforms
Platform companies are powerful because they coordinate ecosystems.
Airbnb coordinates hosts and guests. Uber coordinates drivers and riders. Apple coordinates developers and iPhone users. Amazon coordinates sellers, buyers, logistics and cloud customers.
AI fits naturally into that model because the hardest parts of AI are not only model training. The hard parts are distribution, data access, developer adoption, infrastructure scale and customer trust.
Those are platform advantages.
The biggest AI winners have one or more of these assets:
- Existing global distribution
- Large proprietary or behavioral datasets
- Cloud or GPU infrastructure
- Developer ecosystems
- Strong balance sheets for capital expenditure
- Consumer or enterprise trust
- Marketplaces that can embed AI into transactions
That is why AI has lifted companies such as Microsoft, Amazon, Alphabet, Meta and Nvidia so aggressively. They already controlled the channels where AI can be deployed.
Why Nvidia Is a Platform
Nvidia is the clearest example of how the definition of a platform has expanded.
At first glance, Nvidia sells chips. But the company’s advantage is not only silicon.
Its real strength is the ecosystem around the chips: CUDA, libraries, frameworks, developer tooling, reference architectures, cloud integrations and partner systems. That ecosystem makes Nvidia hardware more valuable because external developers and infrastructure providers build on top of it.
This is platform logic.
Apple did something similar with the iPhone and the App Store. The device mattered, but the developer ecosystem made the device more valuable over time. Nvidia is applying that pattern to AI infrastructure.
For infrastructure teams, this matters because the AI platform economy is not abstract. It shows up in procurement decisions, cloud bills, accelerator availability and software compatibility.
America Extended Its Lead
The Schmidt and Hosseini analysis shows that American platform companies increased their share of global platform value to about 86%.
That is a remarkable level of concentration.
The five largest American platform companies in the analysis - Apple, Nvidia, Microsoft, Amazon and Alphabet - added trillions of dollars in market value over the prior period. AI was a major driver because investors priced in the strategic value of compute, cloud, search, operating systems, advertising, developer ecosystems and AI distribution.
This creates a feedback loop:
- More market value gives companies more capital.
- More capital funds data centers, GPUs, models and acquisitions.
- Better AI products increase customer lock-in.
- Larger ecosystems attract more developers and partners.
- More developers and partners make the platform stronger.
That loop is why AI can increase concentration even when the technology itself becomes more widely available.
Open models and open tooling matter, but distribution still matters more.
Europe Remains Weak in Platforms
Europe’s weak position in platform companies is not new, but AI makes it more visible.
The analysis puts Europe’s share of Top 100 platform value at roughly 2%. SAP and Spotify remain important European examples, but Europe does not have many global platform companies at comparable scale.
This is a problem for digital sovereignty.
Europe can regulate digital markets, fund research and build strong industrial technology. But without globally competitive platforms, it has less control over the places where digital value compounds: cloud infrastructure, AI developer ecosystems, marketplaces, consumer distribution and enterprise software networks.
That does not mean Europe has no path forward.
It means Europe has to be honest about where it is strong. Industrial AI, open infrastructure, privacy-preserving systems, robotics, sovereign cloud, open source and sector-specific platforms may be more realistic than trying to clone American consumer platforms after the fact.
Asia-Pacific Lost Relative Ground
Asia-Pacific still has major platform companies, especially in China, India, Singapore, Japan and South Korea.
Tencent, Alibaba, ByteDance, Meituan, Reliance, Sea, Zomato and others remain strategically important. But the region’s relative share is far below where it stood during the earlier platform boom.
China’s platform sector was hit by regulation after 2020, and many Chinese internet companies did not receive the same AI-driven market premium as their American peers.
At the same time, companies outside China are gaining attention. India’s Zomato and Singapore’s Sea are examples of platforms that benefited from regional growth and investor interest.
The geopolitical layer matters here. AI infrastructure, export controls, semiconductor access and cloud policy are now part of platform strategy.
AI Agents Will Change Marketplaces
The next platform shift will likely come from AI agents.
AI agents can search, compare, negotiate, generate product descriptions, translate listings, automate support and trigger transactions. That can reduce friction across marketplaces and B2B commerce.
Alibaba’s Accio is one example of this direction: AI can help suppliers create listings, translate descriptions, support product discovery and reduce transaction costs.
The strategic question is who captures the value.
In theory, AI agents could lower barriers for smaller sellers and buyers. In practice, the largest marketplaces may benefit most because they control the transaction layer, payments, trust systems, data and customer demand.
The same pattern applies in SaaS.
AI agents can automate routine workflows that used to require application screens, forms and human operators. That threatens some software categories, but it also gives large software platforms a chance to embed agents directly into their products.
Mobility Is a Warning
Not every platform is protected.
Mobility platforms such as Uber and Lyft show how AI can challenge a marketplace when the underlying supply changes. If autonomous vehicle fleets become the supply side, the marketplace model changes.
A ride-hailing platform built around independent human drivers is different from a mobility network built around robotaxi fleets, vehicle operations, mapping, maintenance and safety certification.
That is why companies such as Waymo matter. They are not just another supplier in an existing marketplace. They can reshape the operating model.
This is a useful reminder: AI strengthens platforms when it improves their existing network effects, but it can weaken them when it changes the core asset being coordinated.
The Bigger Lesson
The AI economy is not only a model race.
It is a platform race.
The companies best positioned for AI are the ones that can turn models into ecosystems: cloud platforms, developer platforms, marketplaces, operating systems, devices, enterprise workflows and infrastructure stacks.
That is why the concentration numbers matter.
If about 80% of total Top 100 platform value is held by the ten largest platforms, then AI is entering an already concentrated market. Without deliberate counterweights, it will likely make that concentration stronger.
For builders, investors and policy makers, the conclusion is straightforward:
- AI capability matters.
- Distribution matters more.
- Ecosystem control matters most.
The next decade of AI will not be decided only by who has the best model. It will be decided by who owns the platform where that model becomes useful.
