As the global artificial intelligence industry enters a new cycle of capital expenditure, an AI infrastructure value chain centered on GPUs, data centers, power systems, networking, liquid cooling, and advanced manufacturing is rapidly emerging as a major area of long-term interest for global capital.
Recently, Astroflux Capital was invited to visit NVIDIA, one of the world’s leading companies in artificial intelligence and accelerated computing, for an in-depth exchange. The visit focused primarily on AI computing infrastructure, next-generation GPU platforms, AI data center development, energy demand, and the future evolution of the global AI industry. The two sides also exchanged views on the long-term investment opportunities that may emerge across the broader AI value chain.
For Astroflux Capital, this visit was not simply a corporate tour of a single technology company. Rather, it formed part of the firm’s ongoing research into the global AI industrial ecosystem.
Over the past several years, competition in artificial intelligence has gradually moved beyond large AI models themselves and expanded toward the entire infrastructure required to support model training and inference.
From GPUs and HBM high-bandwidth memory to AI servers and data centers, and further downstream to power systems, energy storage, liquid cooling, and industrial real estate, an increasingly large physical AI infrastructure ecosystem is taking shape.
This transformation is also redefining the logic of global technology investment.
From GPUs to a Complete AI Infrastructure Platform
During the visit, Astroflux Capital paid particular attention to NVIDIA’s development strategy in AI computing platforms, data center architecture, and next-generation artificial intelligence infrastructure.
For many years, NVIDIA was primarily viewed by the market as a GPU company. However, with the rapid expansion of artificial intelligence, the company’s business boundaries have increasingly extended toward a comprehensive AI infrastructure platform.
Today, NVIDIA’s technology ecosystem spans GPU computing, CPUs, AI networking, the CUDA software ecosystem, server systems, and AI data center infrastructure.
In particular, as large-scale AI models demand increasingly greater computing resources, the industry is shifting from traditional server clusters toward higher-density, higher-bandwidth, and larger-scale “AI Factory” infrastructure.
From Astroflux Capital’s perspective, this shift carries significant investment implications.
What the artificial intelligence industry will require in the future is not simply more chips, but an integrated infrastructure system capable of supporting continuous, large-scale computing operations:
AI Models → GPUs and HBM → AI Servers → High-Speed Networking → Data Centers → Power → Energy Storage → Liquid Cooling → Industrial Real Estate → Advanced Equipment and Materials
This means that the capital expenditure generated by artificial intelligence may ultimately extend far beyond traditional technology-sector investment.
Looking Beyond Blackwell to the Next Generation of Computing
During the visit, Astroflux Capital also closely examined the development of NVIDIA’s next-generation computing platforms.
As the Blackwell architecture continues to be deployed across major global cloud computing companies, AI laboratories, and data center ecosystems, the GPU industry is entering a more pronounced phase of “system-level competition.”
Compared with the past, when competition was largely focused on the performance of individual GPUs, the future of AI infrastructure will increasingly depend on the coordinated performance of chips, networking, servers, cooling systems, power infrastructure, and software.
This trend is also pushing the capital requirements of individual AI data centers steadily higher.
Across the global market, major technology companies continue to rapidly expand their artificial intelligence capital expenditures, with AI servers, GPU clusters, and data centers becoming some of the most important areas of investment within the global technology industry.
Astroflux Capital believes that one of the defining characteristics of this industrial cycle is that AI is gradually moving from software innovation into large-scale physical infrastructure construction.
This means that over the coming years, capital expenditure related to artificial intelligence may continue to spread into semiconductor manufacturing, energy, power grids, data centers, industrial real estate, and advanced equipment.
The Biggest Bottleneck for AI May Not Be Chips Alone
Energy and data center infrastructure were also major areas of focus during the visit.
As the computing density of individual AI clusters continues to increase, artificial intelligence data centers are placing significantly higher requirements on power supply, cooling systems, and network infrastructure.
Across global markets, some of the largest AI data center projects are already being planned around power requirements measured in hundreds of megawatts or even gigawatts.
NVIDIA has also continued to expand its AI infrastructure ecosystem.
For example, in September 2026, the company announced plans involving several Australian cloud computing and data center infrastructure partners, with the related AI infrastructure expected to reach a combined scale of up to approximately 2 gigawatts.
This suggests that the future constraints on AI development may not be limited to GPU supply.
Land availability, grid connections, power capacity, transformers, cooling equipment, and data center construction timelines may all become critical factors determining the pace at which AI computing capacity can expand.
From Astroflux Capital’s perspective, future AI investment therefore cannot focus solely on technology companies themselves.
Energy and infrastructure are becoming inseparable components of the artificial intelligence ecosystem.
From Technology Investing to AI Value-Chain Investing
The visit to NVIDIA further reinforced an investment framework that Astroflux Capital has been developing: rather than simply following valuation movements in individual AI companies, the firm is increasingly focused on understanding the long-term capital expenditure cycle behind the artificial intelligence industry.
Over the past period, Astroflux Capital has continued to study AI-related sectors, including semiconductors, data centers, energy infrastructure, advanced manufacturing, and industrial real estate.
In Astroflux Capital’s view, if global demand for artificial intelligence computing continues to expand, capital investment throughout the broader value chain will generate increasingly visible spillover effects.
Rising demand for GPUs will drive demand for advanced semiconductor manufacturing processes and HBM.
Growth in AI servers will increase demand for high-speed networking, liquid cooling, and power supply equipment.
As data center capacity expands, demand for electricity, power grids, energy storage, and land resources will rise accordingly.
Artificial intelligence is therefore beginning to reconnect industries that were previously relatively separate.
Technology, energy, real estate, industrial manufacturing, and infrastructure are becoming increasingly integrated around one central resource: computing power.
Rethinking NVIDIA’s Position in the Global Industrial Landscape
From the perspective of Astroflux Capital’s visit, NVIDIA’s significance can no longer be understood simply through the growth of a semiconductor company.
Instead, NVIDIA is increasingly becoming an important window through which investors can observe the broader global artificial intelligence capital expenditure cycle.
According to NVIDIA’s latest publicly disclosed figures, for the fiscal quarter ended July 26, 2026, the company generated revenue of approximately US$96.2 billion, with data center revenue reaching approximately US$89.0 billion.
The data center business has become one of NVIDIA’s most important growth engines, further reflecting the rapid expansion of global artificial intelligence infrastructure investment.
As next-generation GPUs, AI factories, and large-scale data centers continue to be deployed, the infrastructure investment cycle built around computing power may continue to spread into a growing number of industries.
Astroflux Capital Continues to Expand Its AI Industry Research
Astroflux Capital stated that this visit will form an important part of the firm’s global AI industry research framework.
Going forward, Astroflux Capital will continue to study artificial intelligence value chains across North America, Europe, and Asia, with particular attention to long-term capital opportunities linked to AI infrastructure, including:
GPU and advanced semiconductor supply chains, AI data centers, high-speed networking, power infrastructure, renewable energy and energy storage, liquid cooling technologies, industrial real estate, and advanced manufacturing.
Astroflux Capital believes that the transformation created by artificial intelligence may extend far beyond a new cycle of technology product upgrades.
AI is pushing the global economy into a new era of capital expenditure.
During the internet era, the most important infrastructure consisted largely of fiber-optic networks, servers, and cloud computing.
In the artificial intelligence era, that infrastructure is evolving into a broader system:
Chips + Computing Power + Data Centers + Electricity + Energy + Networks
Those who understand this industrial chain will be better positioned to understand how global capital flows may evolve over the next decade.
For Astroflux Capital, the invitation to visit NVIDIA also represents another step in the expansion of its global investment research from traditional technology sectors toward the core infrastructure of the artificial intelligence economy.
And the global capital expenditure cycle now forming around artificial intelligence may only be beginning.