Artificial intelligence is beginning to change what companies consider infrastructure.
The next phase of the AI investment cycle may extend well beyond chips, cloud computing and data centres. As businesses deploy AI inside factories, mines, ports, warehouses and other physical environments, they are discovering that intelligent machines require something less visible but equally fundamental: dedicated networks capable of moving huge volumes of data between machines, sensors, robots, workers and computing systems in real time.
That is opening a potentially significant new investment pool.
Global spending on private LTE and 5G networks is projected to rise from $17.3 billion in 2026 to $84.7 billion by 2031, according to new research by Juniper Research.
The headline number represents almost a fivefold increase. But the more consequential number for investors is the $67.4 billion increase in annualised spending between the two endpoints; and that changes the nature of the story.
This is no longer simply about whether private 5G becomes another enterprise connectivity product. It is about whether dedicated cellular networks become a core layer of infrastructure for the industrial AI economy — and which companies capture the investment flowing behind that transition.
From connectivity to productive infrastructure
Private cellular networks are increasingly being positioned as the connective tissue of industrial automation.
An autonomous vehicle moving through a warehouse, a machine-vision system inspecting products on a production line, sensors monitoring industrial equipment or an AI system making real-time operational decisions all depend on reliable data transmission.
In those environments, connectivity is no longer merely a communications service. It becomes part of the production system.
That distinction is important because it potentially shifts private-network spending from the traditional telecommunications budget into much larger capital-investment programmes covering manufacturing automation, logistics, energy, mining and industrial digitisation.
The investment opportunity could therefore extend well beyond radio equipment. Network infrastructure, software, edge computing, cybersecurity, systems integration and specialised industrial applications can all participate in the spending cycle.
The commercial prize is consequently much larger than selling an enterprise a faster wireless connection.
AI is changing the economics of networks
The relationship between AI and private 5G works in both directions.
AI creates demand for better connectivity, while better connectivity can make some industrial AI applications economically viable.
Consider a modern factory. Thousands of sensors can generate continuous streams of operational data. Cameras can inspect products for defects. Robots can coordinate movements. Autonomous vehicles can transport components around a facility. Predictive-maintenance systems can monitor machinery for early signs of failure.
The value of AI in each case depends partly on how reliably data can move between the physical asset and the computing system analysing it.Connectivity therefore becomes an economic variable.
If a network is unreliable, the value of an AI deployment can be undermined by downtime, delayed decisions, safety constraints or the need for redundant systems.
If connectivity is sufficiently reliable and responsive, businesses can push automation further.
That is why private-network investment increasingly needs to be considered alongside AI capital expenditure rather than as an isolated telecoms purchase.
The $67.4bn investment question
For investors, the critical issue is what sits behind Juniper’s forecast.
The projected increase from $17.3 billion to $84.7 billion represents an additional $67.4 billion of annual spending at the 2031 endpoint.
That is a substantial new pool of enterprise technology expenditure.
The question is where the money will ultimately go.
Telecom operators have an obvious route into the market, but traditional connectivity revenues alone may not be sufficient.
Network-equipment manufacturers are also positioned to benefit. Juniper’s 2026 competitor assessment identifies Nokia, Huawei, Samsung, HPE and Ericsson among the leading vendors.
But the competitive battlefield is changing.
The winning proposition may increasingly be an integrated industrial solution rather than a network box.
A mining company does not buy connectivity simply because it wants a private 5G network. It wants to automate vehicles, monitor equipment, improve worker safety, increase throughput or reduce downtime.
A manufacturer does not necessarily care about the radio technology itself. Its investment case rests on higher productivity, lower waste, fewer stoppages and greater flexibility.
The technology supplier capable of linking the network investment directly to those financial outcomes has a stronger commercial proposition.
Telecom vendors face an identity test
That creates both an opportunity and a strategic challenge for the telecom industry.
The market is expanding rapidly, but the customer is changing.
Traditional telecom sales are built around connectivity. Industrial AI projects are built around business outcomes.
Nick Maynard, vice president of research at Juniper Research, argues that network performance alone will no longer be sufficient as a differentiator.
The implication is significant. Network vendors may increasingly need to operate alongside industrial automation companies, cloud providers, AI developers, cybersecurity firms and systems integrators.
That could reshape competitive boundaries across the technology industry.
The private-network market could consequently become less about who sells the best network and more about who controls the architecture connecting the network to the industrial application.
Industrial companies face a tougher calculation
For manufacturers and other industrial users, the investment case is more demanding.
Installing a private cellular network requires capital. The resulting productivity gains therefore have to justify the expenditure.
A factory using private 5G to connect robots, cameras and sensors may seek returns through fewer production stoppages and better asset utilisation.
A logistics operator may look to autonomous vehicles, real-time inventory visibility and improved warehouse efficiency.
A mining company may use dedicated connectivity to support remote operations, autonomous equipment and predictive maintenance.
In each case, the return on investment does not come from 5G itself. It comes from what the network enables.
That distinction will become increasingly important as corporate AI budgets face greater scrutiny.
The first phase of AI spending was dominated by experimentation and software adoption. The next phase is likely to face tougher questions about measurable returns.
Businesses will increasingly have to justify not only the cost of AI models and computing capacity, but also the physical infrastructure required to deploy those systems at scale.
Private networks become part of that calculation.
Edge computing strengthens the case
There is also a geographical dimension to the investment.
Many industrial AI applications cannot rely entirely on distant cloud infrastructure.
A production line may require decisions to be made within milliseconds. A mine may operate in an environment with limited public-network coverage. A port may need secure communications across a large industrial site.
That strengthens the case for combinations of private cellular connectivity, edge computing and on-site data processing.
The architecture of industrial AI is therefore becoming increasingly distributed.
Instead of data travelling from a machine to a distant cloud and back, more processing can take place closer to where the data is generated.
Private 5G becomes one layer in that architecture.
That makes the opportunity potentially relevant to companies well beyond traditional telecoms.
Africa could become a proving ground
The opportunity is also relevant to emerging markets, including Africa, where industrial connectivity gaps remain significant while pressure to modernise manufacturing, logistics, energy and extractive industries continues to grow.
Private networks could allow businesses in mining, ports, manufacturing, energy and logistics to build dedicated communications environments around their operational requirements rather than waiting for every capability to be delivered through public networks.
For African telecom operators and technology companies, that creates a potential new enterprise market. But the economics will be critical.
A private network must demonstrate a return against alternative technologies, including Wi-Fi and public cellular services.
The business case will also depend on spectrum availability, equipment costs, technical expertise, cybersecurity requirements and the scale of the industrial operation.
Nigeria has potential applications across ports, oil and gas, manufacturing, logistics and large industrial facilities.
But the opportunity should not be mistaken for an automatic commercial win.
The question for enterprises remains straightforward: What measurable business problem does the network solve?
The investment opportunity is moving upstream
For investors, perhaps the most important implication of Juniper’s forecast is that private networks could become an infrastructure layer of the AI economy.
That creates several potential pools of value:
-
Network vendors can sell the underlying equipment.
-
Software companies can provide network management and industrial applications.
-
Cybersecurity companies can protect increasingly connected operational technology.
-
Systems integrators can design and deploy complex industrial environments.
-
Cloud and edge providers can supply the computing infrastructure required to process the resulting data.
-
Industrial companies can capture the productivity gains if the technology delivers as promised.
The resulting value chain is much broader than telecoms.
It resembles the wider AI infrastructure investment cycle, where spending on computing creates secondary demand for power, cooling, networking, storage and physical infrastructure.
Private 5G could develop along a similar path at the industrial edge.
$84.7bn is a forecast, not a guarantee
The scale of the projection should not obscure the risks.
Industrial technology projects can take years to deploy. Enterprises may postpone investment when financing conditions deteriorate. Technology standards can evolve. Public networks may become capable enough to satisfy some use cases. And companies will abandon deployments that fail to produce measurable productivity gains.
There is also a danger of confusing technological capability with economic demand.
Not every factory needs a private 5G network.
The strongest investment cases are likely to emerge where connectivity directly supports high-value operations, particularly those involving automation, safety, machine vision and continuous data flows.
That makes the quality of enterprise deployments as important as the headline market size.
The bigger bet
Juniper’s forecast points to a fundamental change in corporate technology spending. Private LTE and 5G expenditure is projected to rise from $17.3 billion in 2026 to $84.7 billion in 2031 — a $67.4 billion expansion in annualised spending between the endpoints.
The important question is therefore not whether companies will spend more on 5G. It is whether they will increasingly regard connectivity as productive capital.
If industrial AI moves from experimentation into large-scale deployment, networks connecting machines, sensors and AI systems become part of the machinery of production itself.
That would put telecom vendors in competition for a much larger technology-investment pool — while forcing them to compete on something the industry has historically struggled to quantify: the financial return generated by the business outcomes their networks enable.
The next private-5G market may therefore be won not by the company that sells the most connectivity, but by the one that can make the strongest financial case for why an industrial customer should connect its machines, AI systems and physical operations in the first place.
