Value Creation vs Value Capture

In brief

• Transformative technologies do not always produce transformative investment returns
• The critical question for AI is not how much value it creates, but where that value is captured
• Active management could matter more as investors distinguish between businesses that monetize AI from those that merely fund, enable, or give away its benefits.

One of the hardest things for investors to do is separate a technology’s usefulness from the durability of the profits earned by its producers. History is full of innovations that changed society while disappointing shareholders.

Electricity is perhaps the clearest example. It transformed factories, homes, transportation, and communications. Thomas Edison helped create enormous value. But much of that value flowed outward to households, manufacturers, cities, workers, and consumers. Society captured the surplus.

The internet was different, at least for a narrow group of companies. Amazon, Google, Netflix, and others built ways to capture value. They owned the customer relationship, controlled distribution, and built data advantages, but their costs fell as usage increased. The more people used these platforms, the stronger their economics became.

That distinction matters, because investors do not own productivity. They own cash flows.

AI will almost certainly create value. It can help people write, code, research, design, analyze, and make decisions faster. It could also lower the cost of many cognitive tasks and enable new products. The question is not whether AI creates value. The question is who keeps it.

For frontier AI models, the answer is not yet clear.

Tech is no longer asset light

The first challenge is cost. Unlike the software platforms of the 2010s, frontier AI is not asset light. It may live in the cloud, but that cloud lives on the ground. AI requires datacenters, semiconductors, memory, networking equipment, electricity, cooling, land, permitting, skilled labor, and enormous amounts of capital. Training AI models is expensive. Running them is expensive, too. Though every query is digital, they still consume physical resources.

This creates a different economic profile than many investors associate with technology. Internet 2.0 companies often benefited from powerful operating leverage: build the platform once, and each incremental user could be served at very low marginal cost. Some frontier AI models may face the opposite: more usage could mean more compute, more memory, more power, and more cost. Scale may grow revenue, but it may also grow the cost of goods sold.

When “good enough” is good enough

The second challenge is differentiation. Customers do not always pay for the best technology. They pay for the technology that is “good enough” for the job at an acceptable price.

If a legal summary, software test, marketing draft, or customer-service response can be handled by a cheaper model, many customers will not pay frontier-model prices for frontier-model performance. They will substitute. The token may look like a digital abstraction, but its price is an economic signal. It tells users that compute, memory bandwidth, power, and inference capacity are scarce. When prices rise or budgets become visible, low-return uses migrate toward cheaper or more efficient models. Expensive intelligence gets saved for problems where the benefits justify the cost.

That does not mean frontier AI will fail, but it does mean the market may split. The most advanced models may be used for complex, high-value work, while everyday usage might shift toward smaller, cheaper, and more specialized systems. Adoption can rise while pricing power falls. Usage can explode while value capture disappoints. This is the uncomfortable possibility investors need to take into account.

There is also a bargaining-power problem. If AI lowers costs for customers, competition may force those customers to pass savings on. If AI requires scarce chips, memory, and power, suppliers may capture the economics. If AI outputs are embedded inside someone else’s application, the owner of the workflow may capture the value. If models converge in quality, customers may pit providers against one another. AI creates value, but frontier model providers may not keep much of it.

This is why the distinction matters. A company can help change the world and still be a poor investment if it lacks pricing power, customer ownership, differentiated data, distribution control, or favorable unit economics.

So where might value be captured?

We think value is more likely to accrue to businesses that control scarce inputs, own trusted workflows, possess proprietary data, serve mission-critical use cases, or have the distribution to integrate AI within existing customer relationships.

The risk may be highest for businesses that spend heavily to stay competitive but cannot monetize that spending. These companies may adopt AI, fund AI, or sell AI-branded products yet still give away much of the benefit to others. That is not a technology problem. It is a business-model problem.

Conclusion

The market is still debating how large AI will become. We think that is the easier question. The harder and more important question is how the value surplus will be divided.

Benchmarks do not distinguish between companies that create value and companies that capture it. Indices own capital-intensive model builders, scarce suppliers, workflow owners, AI beneficiaries, and AI victims together. Active management can ask a different question: not simply who is exposed to AI, but whose economics may improve because of it.

AI may become one of the most important technologies of our lifetimes, but value creation belongs to society. Value capture belongs to owners. The gap between the two may define the next phase of equity returns.

The information included above as well as individual companies and/or securities mentioned should not be construed as investment advice, a recommendation to buy or sell or an indication of trading intent on behalf of any MFS product.

Keep in mind that all investments carry a certain amount of risk, including the possible loss of the principal amount invested.

The views expressed are those of the author(s) and are subject to change at any time. These views are for informational purposes only and should not be relied upon as a recommendation to purchase any security or as a solicitation or investment advice. No forecasts can be guaranteed. Past performance is no guarantee of future results.