How companies can capture profits as AI reshapes industries

AI will shift $4.7 trillion in profits. who will capture them?

AI could reshape $4.7 trillion in corporate profits between 2025 and 2035, equivalent to roughly one-fifth of projected global corporate profits in 2035. Companies that adopt AI successfully could improve productivity and still lose ground to competitors that capture more of the value the technology creates.

For executives evaluating AI investments, this creates a problem that conventional performance measures cannot solve. Automating workflows, reducing operating costs, and deploying AI tools can all produce measurable results. Yet these improvements offer little competitive protection when rivals have access to similar technology. A company might become substantially more efficient while a competitor develops new services, builds stronger customer relationships, or changes how an entire market operates.

AI adoption, productivity, and profit capture are therefore three distinct outcomes. Adoption gives companies the ability to use the technology, while productivity improves the economics of existing activities. Profit capture depends on whether a company can turn those improvements into higher earnings, new revenue, or a stronger competitive position.

The strategic question is where AI will change an industry’s economics, how those changes will affect existing profits, and which companies will be positioned to retain the resulting value.

Productivity accounts for only a quarter of AI’s projected profit shift

The forecast in AI Will Shift $4.7 Trillion in Profits. What’s Your Stake? identifies three ways AI could reshape corporate earnings: productivity, innovation, and competitive redistribution. Each affects profits differently and calls for a different investment approach.

The component estimates total $4.6 trillion because of rounding. Productivity accounts for approximately 24% of the projected shift. AI can reduce labor requirements in professional services, shorten manufacturing processes, and improve logistics and other operations. Companies can retain some benefits through higher margins or greater output, while customers may benefit from lower prices or better products.

A manufacturer using AI to improve production scheduling, for example, may increase output without expanding its facilities. A professional services company that automates routine analysis can complete more assignments with the same workforce. Both gain an economic advantage by using fewer resources to produce the same result.

But that advantage can shrink as competitors adopt similar systems. If several companies reduce the cost of delivering the same service, competition may eventually pass some savings to customers through lower prices. Productivity can therefore make an entire industry more efficient while giving individual companies only a temporary advantage.

Innovation offers a larger opportunity because it changes what companies can sell. At $2.2 trillion, it represents the largest individual component of the projected profit shift. AI makes some products and services commercially practical by lowering production costs or enabling capabilities that previously required too much human effort.

Continuous health monitoring shows how this works. AI systems could analyze patient information continuously, identify developing conditions before symptoms appear, and support earlier medical intervention. These capabilities would create demand for monitoring services, new healthcare systems, and different relationships between providers and patients. The commercial opportunity extends well beyond making existing diagnostic appointments more efficient.

Autonomous systems create another opportunity by changing the economics of physical work. Equipment operating with limited human intervention can alter how companies move goods, run facilities, and perform production tasks. AI-assisted radiology and automated contract drafting can similarly expand the volume and range of services that professional organizations deliver. These applications also increase demand for the computing infrastructure and AI models needed to operate them.

Competitive redistribution accounts for another $1.3 trillion, approximately 29% of the projected changes. While innovation creates opportunities for additional economic activity, redistribution changes how existing industry profits are divided. A company that cuts delivery costs faster than competitors gains pricing flexibility, while another offering a better AI-enabled product may attract customers from established providers. Both can increase earnings partly at their rivals’ expense.

Together, innovation and competitive redistribution represent roughly 75% of the projected opportunity. That balance changes how executives should think about investment. Internal efficiency remains valuable, but a strategy focused on cost reduction addresses the smallest of the three mechanisms while leaving larger changes in customer demand and competitive leadership to others.

AI changes the economics of production

The scale of AI’s projected impact becomes clearer when compared with the Internet. Both technologies change how companies operate, but they affect different costs within the economy. This helps explain why AI could transform industries that experienced relatively little structural change during earlier digital transitions.

Between 1995 and 2015, the Internet affected an estimated $1.4 trillion in global corporate profits. Approximately $1.1 trillion came from additional industry profits generated through new markets and productivity improvements, while $0.3 trillion involved profits moving between competitors. The net increase associated with the Internet represented roughly one-fifth of total corporate profit growth during those 20 years. Demographic changes, China’s industrialization, commodity cycles, and other economic forces accounted for much of the remaining growth.

The Internet’s main economic effect was to reduce the cost of distributing information and reaching customers digitally to almost zero. Businesses that depended heavily on customer access, information distribution, or intermediaries consequently experienced major changes. Travel booking and news distribution developed new delivery models, while cybersecurity, financial market infrastructure, and enterprise software expanded to meet emerging digital needs. Payments and retail banking incorporated Internet capabilities into established services.

Google, Meta, and Amazon became major Internet-era businesses by building products and commercial models around these changes. Their growth showed how lower distribution costs could support new ways to reach customers and operate at scale. Yet the same shift had a much smaller effect on industries where digital information distribution accounted for little of the underlying economic activity.

Mining illustrates that limit. Industry profits doubled during the comparison period, but that growth was unrelated to the Internet. Pharmaceutical research and development, construction, farming, and other physical activities also experienced relatively limited structural changes from Internet adoption. Their main costs and operating constraints remained tied to work that digital distribution did little to change.

AI reaches a different part of those economics because it reduces the cost of producing analysis, decisions, software, professional work, and certain physical outputs. Companies can therefore face fundamental changes in how they create products and services even when their customer relationships and distribution channels remain familiar.

Two forms of AI explain this broader reach. Cognitive AI performs or supports knowledge-intensive tasks involving analysis and judgment, such as interpreting radiology images, drafting legal contracts, producing software code, and analyzing complex information. Embodied AI connects intelligence to physical systems such as robots and autonomous vehicles. Autonomous tractors can perform agricultural operations, while industrial robots can carry out factory welding and other production tasks with increasing independence.

Most immediate changes are expected to come from cognitive AI because companies can apply it to existing digital information and workflows. Embodied systems will extend AI’s effects on physical operations over time. Together, these technologies bring pharmaceuticals and biotechnology, healthcare delivery, industrial manufacturing, and professional services within reach of substantial change. The forecast estimates that AI will structurally transform 71% of economic sectors, compared with 41% during the Internet era.

The projected AI impact exceeds the Internet’s estimated effect by more than three times over half the time. Structural transformation is also expected to reach approximately 1.7 times as many sectors. These differences reflect the wider range of economic activities AI can influence, particularly the work involved in producing goods and services.

Earlier technology transitions involving PCs, mobile devices, and cloud computing changed what businesses could do and how markets operated. AI extends that progression into activities that have historically depended on specialized human expertise or physical labor.

For companies in previously stable industries, this creates a new competitive problem. Their products may continue meeting familiar customer needs, but the costs, capabilities, and business models involved in delivering those products could change substantially.

Companies using AI may capture more value than those creating it

If AI changes production across much of the economy, access to the technology alone will offer little protection against competitors. As advanced systems become more widely available, profit distribution will increasingly depend on who controls the assets needed to turn those systems into commercially useful products and services.

Those assets include proprietary data, customer relationships, regulatory permissions, distribution channels, and operating processes. They affect how quickly a company can deploy AI, which services it can deliver, and how much economic benefit it can retain. Their importance varies by industry, creating different opportunities for established companies and new entrants.

The forecast divides industries into four categories based on how AI is expected to change their economics.

These categories describe four ways industry economics can change. They differ from the earlier productivity, innovation, and redistribution breakdown, which describes how profits are affected. The industry-category estimates total $4.6 trillion because of rounding.

This distinction matters because the same AI capability can produce different competitive outcomes across industries. In some markets, established assets become more valuable. Elsewhere, those assets can make it harder for incumbents to respond when new competitors change how services are delivered.

Healthcare imaging shows how existing assets gain value

Healthcare imaging shows how companies controlling the commercial use of AI can capture substantial value from technology developed elsewhere. AI systems can now analyze radiology scans faster and more accurately than most clinicians, according to the forecast’s assessment. This creates opportunities to improve diagnostic services, increase capacity, and change how medical imaging is delivered.

Turning those opportunities into healthcare services requires more than image-analysis software. Health systems control clinical workflows, patient relationships, and valuable medical data. Medical-device manufacturers have established commercial relationships, regulatory approvals, and products already integrated into healthcare operations. These assets determine how AI enters clinical practice and influence which organizations retain the economic benefits.

A health system can incorporate automated image analysis into existing diagnostic services because it already operates within the relevant regulatory environment. It also has the clinical processes needed to turn analytical results into patient care. The AI developer supplies an important technical component, while the health system remains responsible for delivering the service to patients.

The forecast identifies healthcare imaging as an area where value is concentrating among health systems and device manufacturers with these established advantages. Their position shows how proprietary information, regulatory requirements, operating scale, and specialized processes can shape AI competition.

Similar conditions exist in pharmaceuticals and biotechnology, healthcare delivery, healthcare equipment, aerospace and defense, payments, and life sciences. Established leaders in these sectors may be especially well positioned to strengthen their advantages by combining AI with assets built through years of operating in their markets.

Those assets become economically useful when companies apply them to improve products, services, and customer outcomes. A business may control valuable data or distribution relationships, but capturing additional profits still depends on how effectively it puts those resources to work.

Enterprise software shows how incumbency can become a disadvantage

Enterprise software presents a more contested version of the same problem. Established providers have valuable customer relationships and operational knowledge, while AI-native competitors can design products around automated workflows from the start. Competition depends on whether existing advantages help incumbents improve faster than new entrants can establish new ways of working.

An enterprise workflow software company may serve thousands of customers through extensive integrations with business systems. Those integrations, along with proprietary information and high switching costs, support customer retention. Replacing an established platform often means changing workflows, retraining employees, reorganizing data management, and reconnecting applications.

AI can make those relationships more valuable. A provider already supporting a customer’s operations can add new capabilities to familiar workflows, making adoption easier. Information gathered through existing products can also improve the relevance of AI-assisted decisions. These advantages give incumbents opportunities to become more deeply involved in customer operations.

But established operating models can slow change. Existing cost structures, workforce composition, and distribution relationships may make rapid product redesign difficult. Older software architectures can restrict how new functions are built and integrated. An AI-native competitor faces fewer of these constraints when designing its first product.

That flexibility lets a new entrant rethink how work gets done. It can decide which tasks still need traditional software interfaces, how users should interact with automated systems, and how services can require less manual effort. If the resulting product solves customer problems substantially better, established integrations and switching costs may become weaker barriers.

Incumbents therefore face a choice about AI’s role in product strategy. Treating it as a peripheral research and development activity may produce demonstrations and incremental features with little commercial impact. Making AI a board-level priority connects investment decisions to product development, operating changes, and capabilities customers will pay for.

For an established provider, success depends on using customer access and operational knowledge to improve workflows before alternatives gain traction. A new entrant must create enough additional customer value to overcome the incumbent’s integration and switching advantages. Both face the same test: whether their approach delivers a service customers prefer strongly enough to support a profitable business.

This uncertainty extends across industries classified as rewired, where AI can fundamentally change operating economics and competitive positions. The category includes automotive original equipment manufacturers, advertising, management consulting, corporate law, cybersecurity, ground transportation, freight and logistics, machinery, automobile components, consumer finance, and research and data services. Publishing, including books and academic publications, and interactive media and platforms also belong to this group.

Across these industries, AI can change how products are created, how services are delivered, and which capabilities customers value. Established companies enter with resources and relationships that new competitors must develop or overcome. Yet those advantages retain their value only if incumbents can adapt their products and operations as customer expectations change.

Hospitality shows how leadership can change within an established business model

Other industries will retain much of their existing structure while AI changes which companies operate most effectively. The forecast calls this category augmentation and assigns it $1.3 trillion in profits at stake. Hospitality shows how competitive positions can shift while the core customer experience remains familiar.

Hotels and hospitality businesses have developed digital capabilities for demand forecasting, pricing, and customer analysis over the past decade. AI can improve these activities by processing more information and supporting faster decisions. A company can use it to anticipate demand, adjust prices dynamically, identify customer preferences, and improve communication before and after a visit.

These applications affect revenue and operating costs in different ways. Better forecasts support staffing and capacity decisions, reducing the risk of having too many or too few resources available. More responsive pricing helps companies adjust to demand changes, while better consumer insights and post-visit engagement can strengthen customer relationships and encourage repeat business.

Customers will still purchase accommodation and related experiences. Yet two hospitality companies offering similar services may achieve different financial results because they use AI with different levels of effectiveness. One may operate at lower cost, respond faster to demand, and maintain stronger relationships with returning customers.

Over time, those differences can shift margins and market positions toward more effective operators. Slower competitors may face growing pressure even though the industry’s basic business model remains intact. Hospitality shows how AI can redistribute profits through better execution without requiring an entirely new product category.

Customer support shows how AI can replace a delivery model

Customer support faces a more fundamental change. Customers will continue needing assistance, but AI can alter the resources and processes required to provide it. The forecast places this type of change in the revolution category, which accounts for approximately $0.3 trillion in projected profits at stake.

The category includes customer support, IT services, online tutoring, test preparation, and similar activities. In each case, AI can perform work traditionally handled by people. That shift can change an industry’s cost structure and which companies are best positioned to compete.

Consider an illustrative customer-support operation employing hundreds of representatives to handle routine inquiries. As AI systems become capable of resolving common requests, the company can move much of that volume to automated services. Customers may receive faster responses, while the cost of handling each interaction falls substantially.

Lower handling costs change how providers compete. A traditional company may have competed through staffing capacity, labor costs, employee training, and service quality. An AI-native provider can instead build its service around software, automated resolution, and continuous system improvement.

This creates opportunities for companies developing AI-native software and autonomous service-delivery systems. Incumbents can participate by redesigning their services and finding additional ways to create customer value. Providers whose economics depend heavily on people handling routine work, however, face pressure as customers adopt automated alternatives.

The contrast with hospitality explains why the industry categories matter. In hospitality, AI improves activities that support an established service. In customer support, AI can replace much of the process used to deliver the service. Both affect profits, but companies competing in those markets need different responses.

AI infrastructure captures value from demand across industries

The fourth category covers companies supplying the technology and physical capacity required for AI adoption throughout the economy. The technology-foundation category represents $1.5 trillion in projected profits at stake. Its economics depend on growing AI activity across many industries.

As organizations deploy more applications, they increase demand for semiconductors, data centers, cloud infrastructure, foundation models, PCs and servers, and the physical infrastructure needed to run computing systems. Foundation models are general-purpose AI systems that support many applications and tasks. Training and operating them requires computing resources, connecting demand for AI services to infrastructure investment.

Executives already see part of this relationship in rising cloud and computing expenses. As companies expand AI use, suppliers benefit from the additional capacity required to train, deploy, and operate their systems. Infrastructure businesses therefore participate in economic activity across several sectors without delivering the final products or services themselves.

This creates a different profit mechanism from healthcare or enterprise software. A healthcare provider captures value by delivering better medical services, while an enterprise software company earns more by improving customer workflows. Infrastructure suppliers benefit from the computing requirements generated by both.

For adopting companies, however, that demand represents an operating expense. Greater AI use can create revenue opportunities for suppliers while raising costs for businesses buying their technology. Those customers must generate enough commercial returns from their applications to justify the resources consumed.

These industry situations show why competitive advantage depends on more than technical performance. Some companies benefit from assets they already control, while others can change how services are produced. As AI improves and becomes more accessible, companies must build advantages that survive successive generations of the technology.

Durable AI advantages come from what companies learn and control

Widely available AI technology presents a challenge for companies investing today. Specific implementations can quickly become outdated as more advanced systems emerge. A company that spends heavily on one model or application may soon discover that competitors can buy technology with better performance.

Yet early deployment can still create a lasting advantage if it produces assets and knowledge that remain useful after the original technology is replaced. The mechanism is recursive learning: an organization uses information from each deployment to improve subsequent deployments. Over time, repeated improvements can create capabilities that competitors cannot immediately reproduce by purchasing similar AI systems.

Consider a company implementing AI in a core operational workflow. Its initial system produces results that employees and customers can evaluate. Those interactions reveal errors, customer preferences, process limitations, and opportunities for improvement. The company can use that information to adjust workflows, refine how people interact with the system, and improve future results.

Each new deployment then generates more information. As the organization incorporates that feedback, it accumulates proprietary data, operational knowledge, and processes embedded in daily work. The value of early adoption therefore depends partly on the experience it creates and how effectively the company uses what it learns.

A competitor entering later may be able to purchase a more advanced model. But reproducing another company’s operational experience, customer understanding, data infrastructure, and workflow improvements takes time. The earlier adopter may retain an advantage even after the underlying technology changes.

That advantage becomes commercially meaningful when accumulated learning helps the company serve customers better, operate more efficiently, or develop products competitors struggle to match. Deployment starts the process; continued improvement determines how much lasting value the organization builds.

Learning must become part of the operating model

Recursive learning requires an operating process that turns experience into better decisions. Collecting large amounts of data creates little advantage if the organization cannot identify useful information and apply it. Companies therefore need clear connections between deployment, evaluation, engineering, product development, and operational change.

The process begins when AI systems generate customer and operational signals through everyday use. Teams interpret those signals to identify errors, unmet needs, and possible improvements. Engineering and product functions incorporate the findings into systems and workflows. Subsequent use then reveals whether those changes work.

Repeated over time, this process improves organizational judgment as well as technical performance. Teams learn which customer needs matter most, which tasks benefit from automation, and where AI produces measurable economic value. Those insights can guide future investments, product design, and decisions about where to direct employees and capital.

The knowledge becomes especially valuable when it changes how the company operates. Employees make better decisions because they understand their systems’ strengths and limits. Product teams can prioritize improvements based on actual customer behavior, while executives direct resources toward applications producing stronger commercial results.

A company that repeatedly completes this cycle builds an advantage through its ability to learn and adapt. Competitors may obtain similar AI technology, but they must still develop the processes and experience needed to use it as effectively.

Durable assets outlast individual implementations

Because AI technology is advancing rapidly, organizations must also replace systems when better alternatives emerge. This creates a distinction between temporary implementations and assets worth preserving. A model, application, or workflow may become outdated, while proprietary data, customer relationships, technical architecture, and organizational knowledge remain useful.

Technical architecture matters because it determines how easily a company can adopt new technology. Systems designed to preserve useful data and connect applications make replacement less disruptive. Poor architectural choices have the opposite effect, forcing teams to rebuild integrations or lose accumulated information whenever an important component changes.

The strategic objective is to preserve assets that support future improvement while remaining willing to retire outdated implementations. Companies that manage this balance can benefit from AI advances without repeatedly rebuilding the organizational knowledge gained through earlier deployments.

The potential effect of accumulated learning can be substantial. One illustrative scenario describes a company developing AI capabilities for 10 months. During that period, repeated feedback and improvement could produce more than 10 months of effective development progress.

By the time a competitor matches the initial investment, the first company could theoretically have developed a tenfold advantage rather than a 10% advantage. The tenfold difference illustrates the possible effect of repeated learning; it is not presented as an observed industry result. Achieving anything comparable would depend on whether accumulated information and experience produce capabilities competitors find difficult to reproduce.

This mechanism can benefit established businesses and new entrants in different ways. Incumbents may start with customer data, operational infrastructure, and existing commercial relationships. New companies may build advantages through faster feedback, better workflows, different operating models, or proprietary information generated through their products.

Neither position guarantees success. Competitive gaps can emerge within years when one organization repeatedly improves commercially important capabilities while another focuses on isolated deployments. The relevant measure of progress is how much useful knowledge and defensible capability each investment creates.

CEOs must plan for where profits move

For executives, the work begins with a prediction about how their industry’s economics could change over the next 10 years. That prediction should identify which customer needs are likely to remain stable, which preferences may evolve, and how AI could change the cost or quality of meeting those needs. It should also account for how competitors might use the same technology to challenge established positions.

Management can then separate opportunities into three areas: productivity improvements, new products and markets, and transfers of market share between competitors. Each demands different investments. Reducing operating costs may require workflow automation, while creating new services may involve product development, customer research, and unfamiliar commercial models.

The four industry categories help executives understand the broader competitive environment. Yet investment priorities depend on the company’s assets, vulnerabilities, and market position. A healthcare business with valuable clinical data faces different opportunities from an enterprise software provider whose customers may adopt AI-native alternatives.

Executives should therefore assess which existing assets AI could make more valuable. They also need to identify where established operating assumptions create vulnerabilities and what new capabilities the organization must develop. Those judgments provide the basis for deciding which opportunities deserve capital and management attention.

The resulting priorities must connect the board, executive leadership, product teams, engineering functions, and frontline operations. Without that connection, strategic ambitions can remain separate from everyday investment and development decisions. Linking them helps direct resources toward deployments that support the company’s intended competitive position.

Those deployments should generate information that improves subsequent decisions. Companies can monitor customer behavior, operational results, and competitor activity while gathering proprietary information through actual use. Product and engineering teams can apply those findings to improve systems, refine workflows, and develop services that create more customer value.

As AI advances, management must also reassess existing implementations. Outdated components may need replacing, while useful data, architecture, customer relationships, and organizational knowledge should continue supporting future improvements. This makes technology replacement part of an ongoing investment process rather than a separate disruption.

The challenge is maintaining that process alongside immediate operating demands. Current-week execution and next-quarter financial pressures still matter, but investments intended to shape a company’s position over a decade require a longer view. Decisions about architecture, talent, customer relationships, and proprietary information can determine which opportunities remain available years later.

For CEOs, the most useful strategic test concerns direct competitors. Where can AI help the company deliver greater customer value, operate at lower cost, or build an advantage rivals will struggle to reproduce? The answer determines where capital should go, which capabilities deserve protection, and how the company can position itself to capture profits as industry economics change.

In conclusion

AI’s projected $4.7 trillion impact on corporate profits represents more than an opportunity to improve efficiency. It signals a shift in how companies create value, compete for customers, and maintain their market positions. With approximately 75% of the projected changes coming from innovation and competitive redistribution, productivity improvements alone will not determine which businesses benefit most.

For CEOs and business leaders, the priority is understanding where AI will change their industry’s economics and how those changes could affect their competitive position. Companies with proprietary data, established customer relationships, and strong operational capabilities may have an advantage, but those assets will only create lasting value if they support better products, services, and business models.

The ability to learn and adapt will become increasingly important as AI capabilities improve and become more accessible. Organizations that turn deployment experience into proprietary knowledge, stronger workflows, and better customer outcomes can build advantages that extend beyond any individual technology investment.

The companies that capture the most value from AI will not necessarily be those that adopt it fastest or spend the most. They will be those that understand where profits are moving, invest in capabilities that strengthen their competitive position, and continuously improve how they deliver value to customers.