Artificial intelligence has spent the past several years proving what the technology can do. The next test may be considerably harder: proving that the economics can keep pace with the infrastructure being built around it. The global AI market would need to approach $6 trillion in annual revenue by 2031 to support the scale of investment flowing into AI infrastructure, according to Bain & Company’s latest Global Technology Report. The projection brings a new dimension to the AI boom. The challenge is no longer simply developing more capable models or building more computing capacity; it is creating enough economic value to support what is rapidly becoming one of the world’s largest technology buildouts. Bain Bain estimates that existing consumer and enterprise applications of AI could generate between $1.2 trillion and $1.8 trillion annually by 2031. Even at the upper end, that would leave approximately $4.2 trillion to come from new products, services and markets that are still developing today. Bain That gap could become one of the defining business questions of the next phase of AI. Infrastructure Is Growing Faster Than the Revenue Model Much of the AI conversation has focused on capability: larger models, more powerful chips and increasingly sophisticated applications. Behind those advances sits an enormous physical infrastructure requirement. Bain projects annual AI infrastructure spending could reach approximately $1.5 trillion by 2031, encompassing data centers, computing capacity and upgrades to accelerators, memory and networking equipment. The firm also estimates cumulative data-center investment of roughly $5 trillion to $6.5 trillion by 2030, with about 150 gigawatts of additional capacity potentially being added. The National That creates an unusual economic situation. Infrastructure is being developed in anticipation of future AI demand, meaning investment is running ahead of the revenue that will eventually be needed to support it. For technology companies, the question is therefore shifting from whether demand for AI exists to whether that demand can be converted into sufficiently large and durable businesses. Productivity improvements alone may not be enough. The $4.2 Trillion Gap AI is already finding commercial applications across software development, marketing, sales, customer service, IT operations and consumer products. Bain estimates enterprise applications could eventually contribute between $1 trillion and $1.4 trillion annually, while consumer subscriptions and advertising could add another $200 billion to $400 billion. BeInCrypto Those are substantial markets. But compared with the capital required to sustain the infrastructure boom, they leave a significant gap. Bain’s analysis suggests the industry will need entirely new sources of value rather than depending primarily on AI helping existing employees perform their jobs more efficiently. Some of those opportunities are beginning to take shape. Autonomous vehicles, trucks, drones and industrial automation could become important revenue generators. Physical AI, including robotics and digital twins, represents another potentially large category. AI-powered search and advertising could create additional commercial models, while less mature applications in areas such as drug discovery, materials science, mental health and energy could open markets that are difficult to measure today. BeInCrypto The scale required, however, means that incremental products may not be sufficient. The industry needs AI to create businesses that either did not previously exist or dramatically expand the economic value of existing ones. AI’s Next Competition May Be About Monetization For much of the current AI cycle, competition has revolved around model performance, computing resources and access to advanced semiconductors. The next stage could increasingly revolve around monetization. A technically impressive AI product does not automatically become a sustainable business. Companies still need customers willing to pay, use cases capable of producing measurable returns and business models that can operate at sufficient scale. This distinction is particularly important because the infrastructure supporting AI is expensive before many of its future applications have reached maturity. Major technology companies are committing extraordinary sums to computing capacity. At the same time, the eventual distribution of AI revenue remains uncertain. Some value will flow to model developers, some to cloud and infrastructure providers, some to enterprise software companies and potentially much more to businesses that use AI to create entirely new products. The companies capturing the most economic value from AI may therefore not necessarily be those producing the largest models. They could be the businesses that discover where AI solves problems valuable enough for customers to consistently pay for. Data Centers Are Becoming an Energy Story Too The challenge is not purely financial. AI infrastructure has a physical footprint, and its rapid expansion is placing increasing demands on electricity grids, water supplies and equipment supply chains. Bain’s report says data-center size and cost have been roughly doubling every 12 to 16 months. As facilities become larger, the industry is encountering constraints involving power availability, transformers, cooling requirements and other infrastructure needed to operate them. The National This means the economics of AI increasingly intersect with energy and industrial policy. A new data center cannot simply be placed wherever computing demand appears. Developers need access to substantial and reliable power, appropriate land, water or alternative cooling infrastructure, grid connections and local approvals. As AI capacity expands, decisions once viewed primarily as technology investments are becoming questions about physical infrastructure and resource allocation. That could affect where future AI clusters are built and which regions are able to support them. From AI Adoption to AI Absorption Another important theme emerging from Bain’s report is the difference between acquiring AI technology and successfully putting it to work. Organizations can purchase tools relatively quickly. Integrating them into workflows, changing processes and producing measurable economic value is considerably harder. Bain describes the speed at which organizations can effectively put AI to use as an increasingly important competitive variable. The National That distinction matters because widespread adoption alone will not necessarily solve the industry’s revenue challenge. Businesses need to move beyond experimentation and demonstrate that AI investments can either create revenue, lower meaningful costs, improve products or enable services that were previously impractical. The conversation may therefore become less focused on whether a company “uses AI” and more focused on what measurable

Artificial intelligence is beginning to move beyond chatbots, digital assistants and content generation. The next major frontier could be machines that better understand the physical world, and AMD is making a multibillion-dollar bet on that possibility. AMD has agreed to acquire World Labs, the AI startup led by renowned computer scientist Fei-Fei Li, in an all-stock transaction valued at approximately $8.2 billion. The acquisition is expected to close by the end of 2026, subject to customary closing conditions. Following completion, Li is set to join AMD as executive vice president and chief scientist, reporting to CEO Lisa Su. The scale of the transaction is significant, but the technology behind it may be the more important story. World Labs is working on spatial intelligence, an emerging area of artificial intelligence focused on enabling AI systems to understand, generate and reason about three-dimensional environments. For AMD, the acquisition represents an opportunity to bring advanced AI research closer to the computing infrastructure that will power the next generation of intelligent systems. What Is World Labs Building? Much of today’s generative AI is designed to understand and create digital information. Large language models work with text, while image and video models generate visual content from instructions. World Labs is pursuing a different challenge: helping AI develop a richer understanding of space and the physical world. Its work centres on world models and spatial intelligence. The broader goal is to develop systems capable of understanding objects, environments and physical relationships in three dimensions. Such capabilities could eventually have applications across areas including robotics, simulation and design. This is where the idea of physical AI becomes increasingly relevant. Instead of AI operating primarily within digital interfaces, physical AI involves intelligent systems that can perceive, reason about or interact with real-world environments. The commercial potential remains an evolving story, but the field could become increasingly important as AI expands into machines and applications that need more sophisticated awareness of their surroundings. Why AMD Is Making the Move For AMD, World Labs offers more than another entry into the fast-growing AI market. Advanced AI models require enormous computing resources. As those models evolve, they also create new requirements for processors, accelerators, software and computing systems. Bringing World Labs’ researchers inside AMD could therefore give the chipmaker greater insight into the kinds of workloads future AI systems will generate. AMD has said the research could help guide its technology roadmaps across hardware, software and systems. That creates an interesting relationship between AI research and infrastructure. Rather than simply building hardware for technologies developed elsewhere, AMD can potentially gain earlier visibility into emerging model architectures and the computing demands they create. At the same time, World Labs could gain closer access to the hardware and software capabilities required to develop and scale its research. The two companies already had an established relationship before the acquisition announcement. Li said World Labs began a deeper technical partnership with AMD in 2025 involving model training and inference optimisation on AMD GPUs. AMD also participated in a $1 billion World Labs funding round earlier in 2026, according to Reuters. Fei-Fei Li Takes a Senior Role at AMD The deal also brings one of AI’s prominent researchers directly into AMD’s leadership structure. Fei-Fei Li is widely known for her contributions to computer vision and artificial intelligence research. Once the transaction closes, she is expected to become AMD’s executive vice president and chief scientist, reporting directly to Lisa Su. That makes the acquisition as much about research capability and talent as technology. In rapidly developing fields such as AI, acquiring a company can provide access not only to products or intellectual property but also to teams capable of shaping what comes next. For AMD, integrating World Labs could strengthen the connection between fundamental AI research and the infrastructure on which future models will run. Why Physical AI Matters for Business Leaders The acquisition is also worth watching beyond the semiconductor industry. For many businesses, the current AI conversation revolves around productivity: automating administrative work, analysing data, assisting customers, generating content or accelerating software development. Physical AI could broaden that conversation. AI systems with stronger spatial understanding could eventually influence industries where interaction with physical environments is central, from robotics and manufacturing to engineering, product design and simulation. That does not mean businesses should immediately rebuild their strategies around physical AI. The technology remains an emerging field, and its eventual commercial impact will depend on continued research, practical applications and the economics of deployment. But AMD’s willingness to pursue an $8.2 billion transaction demonstrates the strategic importance it places on this direction. For leaders, there is a broader lesson here as well. Technology strategy is not only about adopting what is already mainstream. It also involves identifying capabilities that could reshape an industry before their full commercial impact becomes obvious. AI’s Next Chapter May Extend Beyond the Screen The AMD–World Labs deal arrives as competition in artificial intelligence increasingly stretches across chips, software, models and applications. AMD is effectively bringing advanced spatial-intelligence research closer to the infrastructure that could ultimately support it. Whether the acquisition delivers its anticipated strategic value will depend on execution, integration and how quickly physical AI develops into commercially significant applications. But the direction itself is notable. The AI revolution began by teaching machines to work with information at extraordinary scale. Its next chapter may increasingly involve teaching them to understand the environments in which people actually live and work. AMD’s $8.2 billion World Labs bet suggests that the race toward that future is already underway.