Author: Xia Mo, Editor: Chao Yi
Recently, the OCP Global Summit (Open Compute Project Global Summit) is being held in San Jose, Silicon Valley. It is one of the most important annual technical conferences in the global data center infrastructure field.
If Hot Chips focuses more on chips and computing architectures, and NVIDIA GTC focuses more on AI computing platforms, then the uniqueness of the OCP Global Summit lies in discussing how to integrate chips, servers, racks, networking, power supply, cooling, and software into infrastructure that can be deployed at scale.
Under the global consensus that "the end of AI is computing power, and the end of computing power is electricity," over the past three years, the focus of OCP has shifted from open server hardware to open AI infrastructure, and from rack-level standardization to system-level collaborative design for AI factories. The basic philosophy of OCP is that data center hardware should not rely entirely on the closed designs of a single vendor, but should improve the efficiency of large-scale deployment through open specifications, modular interfaces, and cross-vendor collaboration.
In other words, OCP not only formulates specifications but also promotes reference designs, engineering validation, and the actual implementation of products; participants include Meta, Google, Microsoft, AWS, NVIDIA, AMD, Intel, Samsung, SK Hynix, Micron, as well as infrastructure suppliers such as Taida, Schneider, and Vertiv, along with a large number of startups.
Therefore, OCP is not just a conference, but also an important window to observe how hyperscale cloud providers are reallocating the design power in the data center industry chain.
We have observed that the NVIDIA MGX AI Factory Ecosystem exhibition area can be described as the most signal-significant section at this OCP conference. It sends a clear signal: the 800V DC power supply architecture has entered a critical stage from "proof of concept" to "engineering readiness."
Among the seven companies shortlisted for the NVIDIA SST ecosystem exhibition area, multiple Chinese enterprises have made their presence felt, and all have released heavyweight products. For example, Yangguang Yuanzhi showcased the core components of the EnerNeo Solid State Transformer (SST) in a concentrated manner. It is introduced that relying on silicon carbide high-frequency power electronics technology, the EnerNeo SST can directly convert 10kV-13.8kV medium-voltage AC into 800V DC, with a system efficiency of 98.5% and a 50% reduction in footprint.
In addition, as a co-exhibitor with Lixun Precision, Zhizhan Technology released the NovaLink™ 35kV / 10MW solid-state transformer. According to the introduction, the product adopts 24 standardized liquid-cooled power cells, which further improves the system efficiency to 98.7%, while achieving multiple important technical breakthroughs such as a whole machine power density of > 1 MW/㎡. It realizes an efficient and compact medium-voltage DC power supply solution for AI data centers, and can be called the strongest dark horse in the SST field at this summit.
In fact, accompanied by the global explosion of AI, a new CNY 100 billion-level market—SST has entered the eve of an explosion. Almost all new energy companies, and even tech giants, are trying to get a foothold in this strategic highland.
A CNY 100 Billion-Level Feast Under the AI Wave
Globally, it is an indisputable fact that an increasing number of tech giants and tech newcomers are flocking into the AI wave.
A set of data and industrial realities are sufficient to demonstrate the growth potential of SST, and even the entire AIDC (Artificial Intelligence Data Center) industry ecosystem.
For example, according to forecasts by the CAICT (China Academy of Information and Communications Technology), the total global computing power scale is expected to exceed 16 ZFlops by 2030, with a compound annual growth rate of over40% from 2022 to 2030. The proportion of intelligent computing power, mainly for AI training and inference workloads, will reach over 30%.
The sustained high growth of global computing power marks that computing infrastructure is becoming a strategic real estate allocation as important as office space and production factories. JLL (Jones Lang LaSalle) expects that by 2030, the total global data center capacity will increase from 103 GW to 200 GW, an increase of nearly double.
Market research firm Dell'Oro Group predicts that global data center capital expenditure will approach the milestone of 1 trillion USD in 2026 and reach 1.7 trillion USD in 2030. Goldman Sachs has a more long-term estimate: between 2026 and 2031, the cumulative capital expenditure around computing power, data centers, and electricity may reach approximately 7.6 trillion USD.
The industrial vision is already very clear, but the industry still faces severe challenges. As is well known, the power supply chain in traditional data centers is a lengthy conversion chain. Medium-voltage AC is first stepped down to low-voltage AC via power frequency transformers, and then undergoes multiple "AC-DC-AC" conversions through UPS, and is finally converted to DC by server power modules to supply the chips. Each additional conversion adds another layer of loss.
According to a previously released SST research report by Goldman Sachs, when the power consumption of NVIDIA DGX NVL8 surges to 3.7 times that of A100, and when the power density of a single rack rushes from 30 kW to 1000 kW by 2030, the traditional data center power chain of "grid-transformer-UPS-PDU-server" is turning into an energy disaster—for every 10 kWh of electricity entering the data center, 1 to 1.5 kWh is wasted in the conversion process.
What is changing is not just the technology itself, but the entire computing power industry's understanding and cognition of electricity. Starting from first principles, SST is highly likely to be the most feasible solution for the sustainable development of the AIDC (Artificial Intelligence Data Center) industry.
A set of data is sufficient to summarize the full significance of this transformation. First is the leap in efficiency. Limited by multi-stage power conversion losses, the end-to-end efficiency of the traditional five-stage power supply architecture is only 75%~80%; in contrast, the direct supply architecture based on SST+800V HVDC can improve the end-to-end efficiency to 85%~90%. Data from Zhizhan Technology shows that its NovaLink™ 35kV / 10MW liquid-cooled solid-state transformer product natively supports 800V & 1500V DC output, while achieving a system efficiency of 98.7% and a power density of greater than 1 MW/㎡.
Taking a 2.5 MW system as an example, switching from UPS to the SST solution can save nearly 600,000 kWh of electricity per year. Estimated at an industrial electricity price of CNY 0.5/kWh, only the electricity bill can save about CNY 300,000. For a large data center of tens of megawatts, this number will be amplified to the level of millions or even tens of millions of CNY.
Second is the leap in power density. Mass-production-level SST based on SiC can reduce volume and weight to 1/4~1/6 of traditional solutions, with power density increasing by more than 4 times. A traditional 10kV/1MW transformer requires 50㎡ of space, while SST only requires 1-5㎡, a 90% reduction. The traditional solution for a 1 MW cabinet requires 200 kg of copper busbars, while the 800V DC architecture can reduce copper usage by 45%, decrease power distribution space occupation by 80%, and increase available cabinet space by 30%.
Third is the leap in intelligence and response speed. The response speed of traditional transformers is at the second level, while SST can achieve the millisecond level, capable of precisely matching the severe fluctuations of AI computing power loads. At the same time, SST supports bidirectional power transmission, harmonic suppression, and active voltage regulation, and can seamlessly adapt to renewable energy and energy storage systems, providing the possibility for data centers to realize a "flexible grid."
Overall, for a 2.5 MW data center switching from traditional UPS to the SST solution, electricity costs can be saved by over CNY 500,000 annually, and the released power distribution space can additionally deploy more than a dozen cabinets, with copper usage reduced by tons. For any AIDC project, this is not just an incremental improvement, but a systematic reconstruction starting from the underlying architecture.
As early as October 2025, NVIDIA listed medium-voltage rectifiers and solid-state transformers as the ultimate solutions for future outside-cabinet power in the white paper "800V DC Architecture for Next-Generation AI Infrastructure." According to its roadmap, the single-rack power of Rubin Ultra will reach 600 kW in 2027, and the Feynman platform will even break through 1 MW in 2028 (Note: Reports in July 2026 indicated that the Kyber rack has been delayed to 2028, and there is uncertainty in the actual promotion pace). When the cabinet power density crosses 600 kW and evolves towards the MW level, the traditional 54V/415V power supply architecture is approaching its limits in the three dimensions of expansion elasticity, copper loss control, and cable bearing capacity. Relying on its ability to directly step down medium voltage to DC and a power density several times that of traditional solutions, SST has become the only feasible technical choice in this range.
The deeper change lies in the fact that SST naturally possesses multi-port access capabilities. DC power sources such as photovoltaics and energy storage can be directly connected to its DC bus without the need for independent AC/DC conversion stages. This means that AIDC is no longer a mere power-consuming terminal, but can become an integrated node of source, grid, load, and storage, which is precisely the most scarce interface capability to resolve the contradiction between power supply and demand in the AI era.
Along with the acceleration of the commercialization process, SST is highly likely to grow into a new CNY 100 billion-level track. According to estimates by Huaan Securities, the global SST market size will rapidly climb from CNY 2.82 billion in 2026 to CNY 101.04 billion in 2030. Xingye Securities is more optimistic in its forecast, analyzing that the global outside-cabinet power market space will be about CNY 271 billion in 2030; among which, the penetration rate of SST in terms of installed capacity is about 40%, corresponding to an installed demand of about 51.9 GW. Calculated at an average price of CNY 3.4/W, its market size is about CNY 176 billion.
In summary, in the future, whoever can achieve technological breakthroughs and commercialization layout earlier will have more voice and growth potential globally.
The Figure of a Dark Horse Among Top SST Players
Over two months ago, Mike Tu, Chief Electrical Engineer at NVIDIA, released the white paper "800VDC Architecture for Next-Generation AI Infrastructure" (800VDC V2.0). Unlike the V1.0 stage which merely demonstrated the advantages of high-voltage DC technology, NVIDIA's V2.0 version has completed the critical leap from technical demonstration to standardized implementation, officially establishing 800V high-voltage DC (800 VDC) as the unified power supply foundation for gigawatt-level AI computing power factories, clearly dividing the phased construction path, and simultaneously defining solid-state circuit breakers (SSCB) as a near-term rigid demand and solid-state transformers (SST) as the long-term ultimate solution.
At this OCP conference, NVIDIA has sent a clear signal that the 800V DC power supply architecture has entered a critical stage from "proof of concept" to "engineering readiness."
NVIDIA also joined hands with 7 tech companies to make a collective appearance in the SST ecosystem exhibition area, namely Taida, Yangguang Yuanzhi (a subsidiary of Sungrow Power), Heron, DG Matrix, Lixun Precision, Maigemite, and SolarEdge. These are almost the top players with the most sufficient technical reserves and the fastest commercialization process in the current SST field.
For example, Taida. In November 2025, Taida joined hands with Meituan and Qinhuadata to globally launch the SST-based intelligent DC power supply system solution, becoming the first publicly disclosed commercial implementation case. Its single power cabinet output power reaches 1 MW, occupying only 1 m², with an efficiency of 98.5%. In March 2026, Taida further launched a containerized SST DC mobile intelligent computing center, integrating CPU, GPU, and SST into an integrated delivery unit. Currently, Taida SST has completed a long cycle of grid-connected testing in data centers of top cloud providers.
Another heavyweight player, Sungrow Power, also officially released its fully self-developed solid-state transformer EnerNeo for AIDC scenarios on July 9. It is introduced that relying on silicon carbide high-frequency power electronics technology, the EnerNeo SST can directly convert 10kV-13.8kV medium-voltage AC into 800V DC, with a system efficiency of 98.5% and a 50% reduction in footprint; its pioneering "medium and low voltage zoning + quick plug-and-pull" design brings system availability to 99.999%, which can adapt to next-generation GPUs such as NVIDIA Vera Rubin and the native 800V DC AI Factory power supply architecture, and already possesses mature commercial implementation capabilities.
And the most attention-grabbing in the SST field this time is still Lixun Precision Machinery and its co-exhibitor—Zhizhan Technology. Its released NovaLink™ 35 kV / 10 MW liquid-cooled solid-state transformer has 5 major highlights in its core power cell: 1. Cell power output 420 kW, volume 216 L; 2. Power density 1.94 kW/L, including 35 kV insulation; 3. Cell end-to-end efficiency reaches 99.0%; 4. Adopting Zhizhan Technology's self-developed ZPAK SiC power module, collaboratively integrating high-frequency power conversion, medium-voltage insulation, and liquid cooling; 5. Built-in auxiliary power supply, communication, and control redundancy, supporting continuous system operation after a single power cell failure.
At the whole machine system level, it has further improved the efficiency to 98.7%, higher than competitors such as Taida and Sungrow Power. In addition, it is also in the top tier of the industry in terms of power density and other aspects, with highlights including: whole machine power density > 1 MW/㎡; whole machine power output 10 MW, whole machine footprint 9.6㎡; natively supports 800 V & 1500 V DC output; complies with OCP SST V 0.3.0 specifications; adopts an integrated modular liquid-cooled power cell design, facilitating operation and maintenance as well as large-scale production, etc.
According to the analysis and judgment of Guosheng Securities: the stable operation of SST requires the accumulation of practical experience. The leap from 10 kV to 35 kV is difficult. Enterprises with rich practical experience in SST system project delivery have more opportunities to seize the initiative in AI data center demand.
As is well known, in the fields of UHV DC transmission and flexible DC distribution networks, China's engineering practices and grid-connected operation records are the most in the world. Core technologies required for SST, such as grid-forming control, medium-voltage rectification, and DC protection, have been preliminarily verified on the grid side by domestic enterprises, which is the natural advantage of Chinese enterprises.
In a research report, Guojin Securities breaks down the competitive barriers of SST into three levels: the ability to participate in defining product specifications, medium-voltage system engineering capabilities, and high-frequency power electronics productization capabilities. The time scales corresponding to the three are different: in the short term, it depends on the speed of prototype verification; in the medium term, it depends on the accumulation of system-level reliability; and in the long term, it depends on modular mass production capabilities.
The 24chao Industry Research Institute (TTIR) analyzes that the decisive point of the current competition is not at the hardware level. What truly determines the market position of enterprises is a more forward-looking issue: who enters the architecture design system of AI giants first.
Just as Pulitzer Prize-winning journalist Gary Rivlin revealed a cruel reality in the book "Silicon Valley AI Dream": the training and inference costs of foundation models are astonishingly high; this is a game that only giants can afford to play.
The facts are indeed so. At the earnings call in August last year, Jensen Huang stated that the cost of building an AI data center with a capacity of 1 GW is about 50 billion USD. According to data from authoritative institutions, in 2025, global data center investment reached 770 billion USD, a figure that has already surpassed the investment in the entire photovoltaic industry, as well as the investment in upstream oil and gas exploration and development.
According to other public reports, in the first three quarters of 2025, the capital expenditure of the four major US cloud providers has reached 284.1 billion USD; domestic enterprises Alibaba and Tencent Holdings spent over CNY 150 billion, among which Alibaba announced in 2025 an investment of CNY 380 billion over the next three years for AI and cloud infrastructure construction, with a focus on computing power centers...
Looking ahead, we believe that in the AIDC ecosystem, whoever can build an "industrial community" with leading AI enterprises will be able to seize the initiative in the new industrial competition, and even possess the ability to change the industrial landscape and direction.