Who Is Building AGI? The Race Explained

Short answer: AGI is being pursued by fewer than ten organizations that can build frontier AI models, backed by big technology platforms, chipmakers, cloud providers and national governments, mainly in the United States and China. They are racing because whoever leads could gain enormous economic, scientific and strategic power. The race is real, it is expensive, and it is moving faster than the rules around it.

People love a race. Ask most people who are building AGI and they will name a few famous companies and a couple of famous CEOs. Then they will ask me who is winning.

I am going to do something different in this article. I am not going to give you a leaderboard.

Here is why. The leaderboard changes every few months. A model that tops the charts in spring is overtaken by summer. If you make strategic decisions based on who is ahead this quarter, you will be wrong next quarter.

What does not change as fast is the shape of the race: who the players are, what they need, why they are running and what it means for everyone else. That is what leaders actually need to understand. So let us map it.

If you need a quick primer first, read AGI vs AI: What's the Difference?

What Does It Take to Build AGI?

Before we talk about who, let us talk about what. Building frontier AI requires six ingredients, and very few organizations on Earth have all of them.

  1. Compute. Massive numbers of specialized AI chips. Stanford's 2026 AI Index estimates global AI computing capacity has been growing more than threefold every year since 2022.

  2. Energy. Those chips need enormous amounts of electricity. The newest AI data centers are measured in gigawatts, comparable to the power needs of a city.

  3. Talent. A relatively small pool of world-class researchers and engineers, paid like professional athletes.

  4. Data. Huge amounts of text, images, video, code and, increasingly, data generated by AI itself.

  5. Capital. According to the AI Index, global corporate AI investment reached about $582 billion in 2025, more than double the year before.

  6. Distribution. Products, platforms and customers that put the AI in front of millions of people, generating revenue and feedback.

Notice something about this list. Only the first ingredient is about technology in the narrow sense. The rest are about money, infrastructure, people and reach. That is why the AGI race looks less like a science fair and more like the space race.

The Six Groups Racing Toward AGI

1. The frontier labs

These are the organizations at the cutting edge, training the most capable general-purpose models. Some were founded with AGI as their stated mission. Others are research divisions inside larger companies.

How many are there? Fewer than you might think. Analysts generally estimate that fewer than ten organizations worldwide can currently build true frontier models. Industry now dominates: the AI Index found that more than 90% of notable frontier models in 2025 came from industry rather than universities.

The labs do not all bet the same way. I see four main strategies:

  • Scale first: build the biggest computer, train the biggest model and assume capability grows with size.

  • Efficiency first: get more intelligence per chip and per dollar through smarter design.

  • Safety first: treat alignment and safety research as the path to trustworthy, deployable AI, not as a brake on it.

  • Distribution first: use existing products and huge user bases to get models into the world fastest.

2. The big technology platforms

Large technology companies matter in two ways. Some build their own frontier models. Others invest billions in the labs and bundle their AI into products used by billions of people. They control search, operating systems, office software, social networks and cloud platforms. In other words, they control distribution.

3. The infrastructure layer: chips and cloud

This is the part of the race most people ignore, and it may be the most powerful. A small number of companies design the advanced chips that AI runs on, and an even smaller number can manufacture them at the cutting edge. According to the AI Index, one chip supplier accounts for more than 60% of global AI computing capacity.

There is a saying in every gold rush: sell shovels. In the AGI race, the shovel sellers have extraordinary leverage. Whoever controls the chips, the data centers and the power supply shapes who can compete.

4. National governments

AGI is now a national priority, not just a corporate one.

  • The United States leads in private investment by a wide margin. The AI Index reported about $286 billion in US private AI investment in 2025, compared with about $12 billion in China. The US has also used export controls on advanced chips to slow rivals.

  • China has responded with a focus on efficiency, domestic chips and a large ecosystem of competing developers. The AI Index found the performance gap between the top American and Chinese models had narrowed to just 2.7% by March 2026.

  • Everyone else is pursuing "sovereign AI." Europe, the Gulf states, India, Japan, South Korea, the UK and Canada are investing in their own compute, models and talent so they are not entirely dependent on someone else's AI.

5. The open-weight movement

Some developers release their models openly, so anyone can download, run and modify them. Open models spread capability widely, lower costs and put pressure on closed competitors. They also raise hard questions. Once a model's weights are released, they cannot be recalled, and safeguards can be stripped out.

The open versus closed debate is one of the biggest philosophical divides in the race. One side says openness democratizes power. The other says it hands powerful tools to anyone, including bad actors.

6. Universities and independent researchers

Academia no longer builds the biggest models. It simply cannot afford the compute. But universities and independent researchers still matter. They develop new ideas, test claims, measure risks and keep the companies honest. Many of the most important questions about AGI, including whether today's approach can get there at all, are being asked by academic researchers.

The AGI Race at a Glance

1. Frontier labs

  • Player: Frontier labs

  • What they bring: Models, research talent

  • What they want: To build AGI first

  • Main risk they create: Racing past safety

2. Big tech platforms

  • Player: Big tech platforms

  • What they bring: Distribution, cash, cloud

  • What they want: To own the AI customer relationship

  • Main risk they create: Concentration of power

3. Chips and cloud

  • Player: Chips and cloud

  • What they bring: Compute, energy, infrastructure

  • What they want: To supply everyone

  • Main risk they create: Bottlenecks and dependency

4. Governments

  • Player: Governments

  • What they bring: Policy, funding, security

  • What they want: National advantage

  • Main risk they create: Arms-race dynamics

5. Open-weight developers

  • Player: Open-weight developers

  • What they bring: Free, adaptable models

  • What they want: Widespread access

  • Main risk they create: Misuse that cannot be recalled

6. Universities

  • Player: Universities

  • What they bring: Ideas, independent testing

  • What they want: Understanding and accountability

  • Main risk they create: Being outspent and sidelined

Why Is Everyone Racing?

Follow the incentives and the race makes sense.

  • Economic power. An AI that can do most valuable knowledge work would be the most valuable product ever made. Some define AGI precisely that way: AI that outperforms humans at most economically valuable work.

  • Scientific discovery. AGI could speed up drug discovery, materials science and energy research by years or decades.

  • National security. Governments see advanced AI as a potential shift in military and intelligence power, comparable to past technological revolutions.

  • First-mover advantage. Many believe the first to reach AGI could use it to build even better AI, pulling further ahead. That belief alone fuels the speed.

  • Talent and prestige. The AGI mission attracts the best researchers. Nobody wants to be the lab that fell behind.

The Problem With Races

Competition usually makes products better and cheaper. I am a big believer in it. But races have a dark side.

When everyone believes the finish line is close and the prize is enormous, the pressure to cut corners grows. Safety testing takes time. Transparency gives competitors information. Caution feels like losing.

There is evidence this is already happening. Stanford's Foundation Model Transparency Index, which measures how much major AI developers disclose about their models, fell from an average score of 58 to 40 in the latest AI Index. The most capable models often disclose the least.

This is why governance matters so much. I cover who is setting the rules in Who Will Control AGI? Regulation and Governance Explained, and the risks in plain English in Is AGI Dangerous?

Is the Race Even Pointed in the Right Direction?

Here is a twist that most "AGI race" articles skip. Not everyone agrees the leading approach will get to AGI at all.

In a 2025 survey of 475 AI researchers by the Association for the Advancement of Artificial Intelligence, 76% said scaling up today's AI approaches was unlikely or very unlikely to produce AGI. Many believe new breakthroughs will be needed, such as AI that learns continuously or reasons in fundamentally different ways.

So the race might be a sprint toward a finish line that requires a different kind of runner. That possibility is one reason I tell leaders to be careful with headlines. More on that in AGI Hype vs Reality.

What the AGI Race Means for Your Business

You are not building AGI. But you are a customer, a competitor and a citizen in a world shaped by the people who are. Here is what I recommend.

  1. Do not marry one AI provider. The leader changes constantly. Build your systems so you can switch models without starting over.

  2. Expect falling prices and rising capability. The race is driving costs down quickly. Plans built on today's AI prices will be out of date within a year.

  3. Watch the infrastructure, not just the apps. Energy costs, chip supply and data center locations will shape what AI costs you and where your data lives.

  4. Think about sovereignty. If you operate across borders, ask where your AI runs, which laws apply and what happens if trade rules change.

  5. Focus on what the race cannot give you. Every company will have access to powerful AI. Your advantage will come from your data, your relationships, your judgment and how fast your people learn. I call this the deep generalist advantage.

For the full action plan, see How Should Companies Prepare for AGI?

My Take: The Race Is Not Yours to Win

In The Bold Ones, I write about looking outside your industry to recreate within. The AGI race is the ultimate example. Most organizations will never build a frontier model. That is fine. The winners in your industry will be the ones who learn to use whatever the race produces faster and more wisely than their competitors.

The question is not "Who will win the AGI race?" The question is "Will my organization be ready to run with whatever comes out of it?" If you want to see how I talk about this shift on stage, watch Why the Agentic AI Era Will Be Huge.

Conclusion: Understand the Shape, Not the Scoreboard

Who is building AGI? A small group of frontier labs, backed by the biggest technology platforms, a handful of chip and cloud providers and the world's most powerful governments, with open-source developers and researchers pushing and questioning from the sides.

They are racing because the prize looks enormous. They are spending at a scale the world has rarely seen. And they are moving faster than the rules designed to keep everyone safe.

You do not need to know who is ahead this month. You need to understand the forces shaping the race, because those forces will shape your costs, your competitors and your customers for the next decade.

If your leadership team or conference audience needs a clear-eyed view of where AI is heading, explore my AI keynotes or book me for your next event.

Frequently Asked Questions About Who Is Building AGI

Who is building AGI?

AGI is being pursued mainly by a small group of frontier AI labs, fewer than ten worldwide, supported by large technology platforms, chip and cloud companies, and national governments, especially in the United States and China. Open-source developers and university researchers also play important roles.

Which country is winning the AGI race?

The United States leads in private investment, with about $286 billion in 2025 compared with about $12 billion in China, according to Stanford's AI Index. But the performance gap between the best US and Chinese models had narrowed to about 2.7% by March 2026, so the lead is much smaller than the spending gap suggests.

Why are companies racing to build AGI?

Companies are racing because AGI could perform most valuable knowledge work, creating enormous economic value. Governments also see strategic and security advantages, and many believe the first to reach AGI could use it to accelerate further ahead.

How much money is being spent on AGI?

There is no separate figure for AGI, but global corporate AI investment reached about $582 billion in 2025, according to Stanford's AI Index, more than double the previous year. Much of that funds the chips, data centers and research behind frontier AI.

What do you need to build AGI?

Building frontier AI requires massive computing power, huge amounts of energy, elite research talent, large datasets, enormous capital and distribution to reach users. Very few organizations have all six.

Is the AGI race dangerous?

Racing can push developers to cut corners on safety and transparency. Stanford's Foundation Model Transparency Index fell from 58 to 40 in its latest edition, suggesting developers are disclosing less. This is why many experts call for stronger governance alongside competition.

Will today's AI lead to AGI?

Not everyone thinks so. In a 2025 survey of 475 AI researchers, 76% said simply scaling up current approaches was unlikely to produce AGI. Many believe new breakthroughs, such as continuous learning, will be needed.

About Shawn Kanungo

Shawn Kanungo is a globally recognized innovation strategist, keynote speaker and bestselling author of The Bold Ones. He spent 12 years advising leaders on disruptive innovation at a global professional services firm and has more than two decades of experience working with Fortune 500 companies, associations and governments. His keynotes help leaders understand where AI is heading and how to act on it. Watch more at watch.shawnkanungo.com.

Previous
Previous

How Should Companies Prepare for AGI?

Next
Next

How Will We Know When AGI Arrives?