Huang calls for faster progress
Nvidia CEO Jensen Huang urged AI cloud company Nscale to resolve problems encountered during data center construction and accelerate the start of facility operations.
As demand for AI computing rises rapidly, completing physical facilities and securing electricity and equipment are emerging as major challenges for the industry.
Bloomberg reported that Huang had called for a swift response to delays in Nscale's data center construction.
Nscale is a UK-based AI cloud developer backed by Nvidia. It provides companies and developers with the computing infrastructure needed to train and run AI models.
However, it reportedly faces difficulties bringing planned facilities into operation on time as construction schedules slip and equipment and power supply problems arise at some data center projects.
During an Nvidia event in Berlin, Huang said Nscale needed to resolve its problems more quickly. Citing strong AI computing demand and a solid customer base, Huang emphasized the growth potential of neocloud companies.
Construction and power constraints
The situation illustrates a gap between the pace of growth in the AI industry and the pace of physical infrastructure construction. Securing high-performance semiconductors alone is not enough to expand AI services. Space for servers, cooling facilities, a stable electricity supply, network connections and financing must also be secured.
A scheduling disruption affecting any one of these elements can delay the start of operations for an entire data center.
Electricity is a particularly important factor. In a separate report, Reuters said grid connection had been delayed for an Nscale data center project planned in Loughton, Essex, in the UK, and that Microsoft had withdrawn from an agreement to secure computing capacity at the facility.
Local grid officials reportedly expected electricity to become available to the facility in the 2030s. Nscale said other contracts were unaffected and reviewed alternative power supply options.
This case illustrates how AI data center projects depend heavily on grid infrastructure and construction schedules as well as technological competition. Even if semiconductor performance improves, it is difficult to increase computing capacity available to customers as planned unless facilities are ready to put those chips to use in actual services. Data center developers must manage power supply timelines, construction schedules and facility operating plans comprehensively alongside equipment procurement.
Financing and modular infrastructure
Financing is also attracting industry attention. Bloomberg reported that Nscale had targeted a valuation of about $30 billion, approximately 42 trillion won, as it pursued an initial public offering, but that some investors disagreed with that valuation.
Neocloud companies must make substantial upfront investments to expand large-scale AI computing facilities. However, delayed facility completion can also postpone revenue generation while increasing borrowing costs and operating burdens.
Customers changing their plans or withdrawing under contractual terms can also affect the business outlook. Investors need to assess actual facility delivery capabilities and cash flow alongside the scale of long-term contracts.
Nvidia also presented equipment and design technologies intended to shorten data center construction times. According to Bloomberg, the DSX Pod system introduced by the company at the Berlin event is a modular system integrating servers, cables and piping.
Nvidia explained that deploying equipment that has been preinstalled and tested can simplify infrastructure construction. However, actual installation times may vary depending on site conditions, including site readiness and power and cooling facilities.
Huang's remarks reflect both expectations for the AI industry's growth potential and the practical difficulties encountered in infrastructure construction. Even as computing demand increases, business expansion is unlikely to proceed as planned without facility operations and financing to support it.
How effectively Nscale and other neocloud companies resolve construction delays and power procurement problems is expected to become a key criterion for assessing investor confidence.
Original Korean article: Jemin Ilbo
English brief by Korea Wind Intelligence, koreawindintel.com
