The artificial intelligence boom may be entering a new phase, with the biggest gains expected to come not from the construction of data centres and the sale of advanced chips, but from how businesses redesign everyday work.
While companies such as Nvidia, Dell Technologies and major cloud providers have benefited enormously from the massive investment in AI infrastructure, the productivity gains expected from those investments have so far remained largely incremental.
Workers are using AI tools to summarise meetings, prepare drafts and assist with coding, but businesses are increasingly moving towards a more ambitious model in which AI systems take on entire workflows.
This emerging shift is transforming AI from a simple digital assistant into what is increasingly being described as a digital coworker.
Companies at the forefront of the technology are already experimenting with AI agents capable of handling multiple stages of business processes, with human employees supervising and making final decisions.
OpenAI, for example, says nearly all of its employees use Codex regularly. The tool, initially developed for software development, is now being used for tasks ranging from investigating customer billing issues and building dashboards to supporting legal reviews and analysing internal information.
Google has similarly deployed AI agents within its finance operations to compare invoices with contract terms before human employees examine exceptions.
The company says the system has enabled finance teams to process significantly more invoices without increasing staff numbers and could help prevent millions of dollars in overpayments.
At Anthropic, AI is being used in marketing operations to automate event creation, campaign management and data imports. Some systems are even designed to review the work of other AI agents before a human gives final approval.
The examples highlight a growing trend in which AI is not necessarily replacing employees but taking over repetitive and time-consuming tasks, allowing professionals to focus on work requiring experience, judgment and critical thinking.
The transformation is also spreading beyond technology companies.
Financial institutions are using AI for investment research and fraud detection, retailers are applying it to inventory and supply chains, pharmaceutical companies are using it to accelerate drug research, while manufacturers are deploying the technology for predictive maintenance and factory automation.
The infrastructure supporting this expansion remains equally important. Dell’s growth in AI-optimised servers, storage, networking and systems integration illustrates how companies that provide the tools needed to deploy AI may benefit significantly, even if they do not develop foundation models themselves.
Every large-scale AI deployment requires computing power, storage, networking, cybersecurity, cooling and integration.
However, the biggest beneficiaries of the next phase of AI could be small and medium-sized businesses.
Unlike large corporations, many smaller companies operate with fewer legacy systems and less technical debt. This could allow them to introduce AI into areas such as scheduling, invoicing, customer communication, proposal writing and compliance more quickly.
For some businesses, the technology could effectively increase their capacity without requiring a proportional increase in staff numbers.
The transition will not be without challenges. As AI increases the volume and speed of work, companies could face new bottlenecks if human teams are unable to review the output generated by automated systems.
That makes governance, security, accountability and human oversight increasingly important.
The central question surrounding AI is therefore shifting from whether the technology will replace entire professions to how it will reshape the work professionals perform.
Just as spreadsheets did not eliminate accountants and computer-aided design did not make architects obsolete, AI is likely to automate routine tasks while increasing the importance of strategic thinking and human judgment.
The companies that gain the most from the AI revolution may ultimately be those that successfully combine human expertise with intelligent automation.
The next stage of the AI boom will therefore be measured less by the number of chips, data centres or models deployed and more by whether businesses can reduce costs, improve productivity, accelerate innovation and deliver better services.
The infrastructure boom may have created the foundation for the AI revolution. The real productivity payoff, however, could depend on how effectively organisations now redesign the way work gets done.
