The Winners and Losers From the Agentic Boom

The artificial intelligence story is entering a new phase. The first wave was dominated by chatbots that could answer questions, summarize documents and generate code.

The emerging phase is increasingly about AI agents: systems capable of planning tasks, using software, calling tools, retrieving information and acting with less human intervention.

This shift could redistribute value across the technology economy, creating clear winners while putting pressure on companies whose business models depend on human attention or repetitive digital work.

The biggest potential winners are the companies providing the infrastructure on which agents operate. Semiconductor manufacturers, cloud providers and data-center operators stand to benefit as agents require substantial computing power.

Unlike a chatbot that may respond to a handful of prompts, an autonomous agent can perform dozens or hundreds of model calls while completing a complex assignment.

That creates a potentially larger and more persistent demand for GPUs, networking equipment, storage and electricity. Cloud companies also occupy an important position because enterprises increasingly need secure environments in which agents can access corporate databases, applications and internal tools.

The companies capable of combining computing infrastructure with identity management, cybersecurity and enterprise software could capture significant value as businesses move from AI experimentation toward deployment.

Another group of winners could be software companies that successfully transform their products into agent-driven platforms. Enterprise applications that once required employees to navigate menus.

Spreadsheets and dashboards could increasingly become destinations where agents execute workflows directly. Customer support, accounting, procurement, software development, research and sales are particularly exposed because many processes already follow structured digital rules.

Financial infrastructure could become another important beneficiary. If agents eventually transact independently, they will need machine-readable identities, permissions and payment systems.

Stablecoins, programmable accounts and blockchain-based settlement networks could become useful infrastructure for machine-to-machine commerce, particularly where agents need to make small or frequent payments across borders.

But the agentic boom also creates losers. Companies selling repetitive digital labor face some of the clearest disruption. Outsourcing businesses, basic customer-service operations, data-entry providers and certain administrative services could experience pressure as enterprises automate portions of their workflows.

The impact will not necessarily mean immediate mass unemployment. More likely, individual jobs will be redesigned, with employees supervising automated systems rather than performing every task themselves.

Some traditional software businesses may also struggle. Applications built around human interaction can lose value if customers increasingly access their functionality through agents.

If an AI assistant can search multiple services, compare options and execute a transaction, the application that previously controlled the customer relationship may receive less direct traffic.

Advertising-driven platforms face a similar structural question. The traditional internet monetizes human attention: people browse pages, watch videos and click advertisements. Agents do not necessarily behave like humans.

They can retrieve information without viewing advertisements, compare products without visiting dozens of websites and complete transactions without spending time inside a social feed. That could challenge business models built around impressions and engagement.

Yet the agentic economy is unlikely to produce a simple winner-takes-all outcome. Its economic consequences will depend on reliability, regulation, security, computing costs and consumer adoption. Agents must be trusted with increasingly consequential tasks, and failures could impose financial or legal costs.

The deeper transformation is therefore not simply that AI is becoming smarter. It is that software is beginning to act rather than merely respond. Companies selling the infrastructure, permissions, computing and financial rails for that activity may capture new markets.

While businesses dependent on repetitive human labor or passive digital attention could face structural pressure. The agentic boom is ultimately a reallocation of economic activity-from humans operating software toward software operating software.