Thursday, 17 September 2026
Nvidia expects another major increase in demand for artificial intelligence computing, but the company’s growth is becoming increasingly dependent on something outside the semiconductor industry: the availability of electricity, development land and data-centre capacity.
Speaking at Goldman Sachs’ Communacopia + Technology Conference, Nvidia chief executive Jensen Huang said the company could achieve revenue growth of around 70% year-on-year. The figure represents Huang’s outlook rather than formal full-year revenue guidance, but follows another period of exceptional expansion for the company as technology groups, AI laboratories and infrastructure operators increase spending on computing capacity.
Nvidia reported revenue of $96.2 billion for its second quarter of fiscal 2027, more than double the level recorded a year earlier. Data-centre activities generated $89 billion, an increase of 117% year-on-year, meaning that more than nine-tenths of quarterly revenue came from the part of the business most closely connected with large-scale computing infrastructure.
The company’s products are also changing in scale. Nvidia increasingly supplies integrated computing platforms combining processors, networking, memory, software and other equipment rather than relying simply on sales of individual graphics processors. Huang used the Goldman Sachs conference to illustrate how far the business has moved from the consumer graphics cards that established Nvidia’s early market position.
This transition has significant implications for the property sector because the infrastructure required to accommodate the latest computing systems is becoming considerably more demanding. Nvidia’s growth increasingly depends on customers having suitable buildings, sufficient electricity and locations capable of supporting high-density computing equipment.
Huang specifically identified land, electricity and available data-centre buildings among the factors Nvidia follows when assessing how quickly additional computing capacity can be deployed. The constraint on AI development is therefore shifting beyond the availability of advanced processors towards the physical infrastructure needed to operate them.
Australia provides an indication of the scale involved. Nvidia is working with data-centre and infrastructure companies on capacity of up to 2 GW by 2027. The programme involves companies including AirTrunk, NEXTDC, CDC and IREN and is intended to expand the land, electricity and buildings capable of accommodating successive generations of Nvidia computing equipment.
During the Goldman Sachs conference, Huang associated the Australian programme with approximately $80 billion of infrastructure. While individual projects within the wider programme will have different ownership and financing structures, the figure illustrates how quickly investment requirements increase when AI capacity reaches gigawatt scale.
Nvidia is simultaneously becoming more involved in the financing structures supporting this expansion. The company has announced partnerships involving Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR aimed at creating financing platforms capable of mobilising more than $500 billion of third-party capital for AI infrastructure over time. The figure represents the intended aggregate capacity of these platforms rather than committed Nvidia expenditure or company revenue.
The participation of major infrastructure and alternative-investment managers demonstrates how the AI expansion is moving closer to conventional infrastructure investment. Data centres require land and buildings, but increasingly large parts of their capital requirements are associated with substations, power distribution, cooling, backup generation and other systems necessary to support intensive computing.
Higher-density equipment is also changing building specifications. Facilities designed for traditional enterprise computing may not automatically be suitable for the latest AI installations, increasing demand for purpose-built capacity and upgrades to existing sites. Access to sufficient electricity can determine whether a development proceeds at all, particularly in markets where grid connections already involve long waiting periods.
The relationship between technology and real estate is consequently becoming more direct. Nvidia itself has recently secured land, power and building capacity through a partnership at the PORTS-Pike Technology Campus in Ohio, while its international partnerships increasingly combine computing technology with data-centre operators, energy resources and development sites.
Competition in AI processors continues to increase. Major technology companies are developing proprietary chips alongside their purchases of Nvidia equipment, while other semiconductor manufacturers are competing for AI workloads. That creates uncertainty over how much of future computing investment Nvidia will ultimately capture.
The wider infrastructure requirement is less dependent on which processor supplier prevails. Continued growth in AI computing requires facilities capable of supplying large quantities of reliable electricity and managing increasingly concentrated computing loads, whether those facilities ultimately contain Nvidia hardware or competing systems.
For commercial real estate, this makes the AI boom increasingly a question of physical capacity. The availability of powered sites, substations, grid connections, cooling infrastructure and suitable data-centre buildings is becoming closely connected with the speed at which new computing capacity can be brought online.
Nvidia’s growth outlook therefore provides more than an indication of semiconductor demand. It points to a much larger investment cycle in which the expansion of artificial intelligence is increasingly determined by how quickly the property, energy and infrastructure sectors can provide the physical capacity required to support it.
