蓝色生死恋
Practitioners’ Insights: How did a decade of dedicated effort build China a self-reliant ‘digital track’?_我的网站
一 | Editor's Note:
Ahead of the opening of the 2026 World AI Conference and High-Level Meeting on Global AI Governance, Chinese AI start-up Moonshot AI released its Kimi K3 large language model. As the world's largest open-source model by parameter count to date, this launch marks a significant step forward in the development of China's artificial intelligence models.
From the C919 airliner soaring into the skies to Unitree's humanoid robots stealing the show; from DeepSeek pushing the AI frontier to Moonshot AI's landmark unveiling of Kimi K3 - China's wave of homegrown innovations has dominated global headlines in recent years, delivering a steady stream of breakthroughs.
At a meeting in Beijing that brought together the national science and technology award conference, the general assemblies of the members of the Chinese Academy of Sciences (CAS) and the Chinese Academy of Engineering (CAE), and the 11th national congress of the China Association for Science and Technology in July, Chinese President Xi Jinping, also general secretary of the Communist Party of China (CPC) Central Committee and chairman of the Central Military Commission, stressed that the 15th Five-Year Plan period (2026-2030) is a critical phase for tackling tough challenges in building up the country's strength in science and technology.
"We must seize the historic opportunity, rise to the challenges of the times, accelerate efforts to achieve high-level self-reliance and strength in science and technology, and make steady progress toward the 2035 goal of becoming a leading country in science and technology," he said.
In the article "Strive for Greater Strength and Self-Reliance in Science and Technology" included in the fourth volume of Xi Jinping: The Governance of China, Xi pointed out, "Through years of endeavor, our country's overall strength in science and technology has improved substantially. We therefore have a solid foundation, and are fully confident in our ability to seize the opportunities offered by the new revolution in science, technology and industry to achieve greater results." The article also mentioned that "We should participate to the full in global science and technology governance, contribute Chinese wisdom, and shape a philosophy of technology for good purposes, so that science and technology better serve human wellbeing, and enable China's science and technology industry to contribute more to building a global community of shared future."
In the 27th installment of the special series "Decoding the Book Xi Jinping: The Governance of China," the Global Times, along with the People's Daily Overseas Edition, continues to invite Chinese and foreign scholars, translators of Xi's works, practitioners with firsthand experience, and international readers to focus on the theme of striving for greater strength and self-reliance in science and technology. Together, they share views on China's tech growth, governance principles, and global cooperation in science and technology.
This is the 25th installment of "Practitioners' Insights." We turn our focus to a "digital track" laid for China's industrial internet: the Industrial Internet Identifier Resolution System. A decade ago, when a research team at the China Academy of Information and Communications Technology (CAICT) took on the task, the prevailing view in the industry was that it was "almost impossible." At that time, China had neither a strong voice in international standard-setting nor a deep technological foundation, yet it set out to build a self-reliant and controllable digital identity infrastructure spanning the entire industrial chain.
Today, the system's core indicators rank among the world's best, and its application scale is the largest globally. From dependence to autonomy, from "follower" to "pacesetter," the leap was made possible by the unwavering footsteps of Chinese researchers on the path toward high-level scientific and technological self-reliance and self-strengthening.
At daybreak, in an electric-vehicle factory, the assembly line was operating at full capacity. Each freshly produced power battery was assigned a special "ID" code. That code would stay with the battery as it entered storage, was loaded onto vehicles and hit the road, accompanying it through its entire lifecycle.
From the laboratory to the factory, this invisible data flow connects every stage of production, circulation, use and recycling across industries. And the underlying infrastructure making all of this possible is the "Industrial Internet Identifier Resolution System," which Liu and his colleagues helped build.
In simple terms, the system is like a "digital ID card plus navigation system" for the industrial world. In the real world, everyone has an identity card and every address has a street number; on the internet, every website has a domain name. Likewise, in the era of the industrial internet, every device, component, product and piece of data needs a unique, trustworthy and resolvable digital identity.
The Industrial Internet Identifier Resolution System is China's national-level digital infrastructure that supports the generation, query, circulation and management of those digital identities. Built and operated over nearly a decade under the leadership of CAICT, the system has become a fully self-reliant and controllable "digital track" serving China's industrial internet.
In the article "Build China into a Sci-Tech Powerhouse" included in Volume IV of Xi Jinping: The Governance of China, Chinese President Xi Jinping emphasized "increasing our country's strength and self-reliance in science and technology, and keeping us well-positioned in vital areas of science, technology and development."
Looking back on the decade-long campaign, Gao Qi, deputy chief engineer at CAICT's Industrial Internet and Internet of Things Research Institute, feels this especially deeply.
"As a critical piece of digital infrastructure, the Industrial Internet Identifier Resolution System is a vivid embodiment of this important principle in the field of industrial digitalization," Gao told the Global Times. What this system carries is precisely the historic mission of safeguarding the lifeline of China's digital industry, and achieving industrial self-reliance and controllability, he said.
Pioneering in uncharted territory
Fast-forward to 2015.
By then, Liu had only recently joined the CAICT. Before that, his work had focused mainly on developing software for internet domain-name resolution, so he already had a basic understanding of the underlying logic of "resolution." But when he was assigned to take part in planning and building China's Industrial Internet Identifier Resolution System, he knew immediately that this was something entirely different.
At the outset of the project, the most pressing challenge was a classic "chokepoint" problem. China gained full access to the global internet in 1994, which essentially meant the comprehensive adoption of technical protocols, industrial ecosystems and governance rules developed elsewhere. Building a Chinese-owned, independent and controllable identifier resolution system was not a matter of filling in a single missing technology. It meant creating an entire system from scratch. At the time, many in the industry believed the challenge was enormous, and some even said it was "practically impossible."
Why? Gao, one of the key strategic drivers behind the system, recalled that the challenge came on three levels.
First, China then lacked a voice in international standards and had little deep technical accumulation of its own, yet it was expected to build a system capable of covering the entire industrial chain and interoperating with dozens of heterogeneous systems. The technical difficulty was immense.
Second, in a landscape long reliant on foreign technologies, persuading tens of thousands of enterprises to adopt domestic identifiers was an enormous challenge in terms of industrial deployment and promotion. Many people dismissed the idea as sheer fantasy.
Third, with no clear expectation of return on investment, who would take responsibility for pushing the project forward and explore a sustainable operating model?
Gao said that from the project feasibility study stage, the team was thinking simultaneously about the technical roadmap, the system architecture and the industrial ecosystem as a closed loop. "We had to break through core technologies, but we also had to design a mechanism that would ensure companies would be willing to connect to the system and able to benefit from it," he recalled.
On the front line, Liu felt the pressure in a direct and concrete way. Time was tight, the mission was heavy and there was no ready-made experience to copy. Everything had to be researched and designed at the same time, developed and verified in parallel.
"There was also the pressure on the team," Liu told the Global Times. With the team under intense pressure, uncertainty in direction or repeated setbacks could easily cause anxiety among engineers. As both the leader of a development group and the person responsible for core module development, Liu had to solve technical problems while also keeping the team's pace steady.
The work was tense, but Liu felt calm and confident. "The project had already been thoroughly validated by experts and academicians in China, so the direction was clear. What we, as R&D personnel, needed to do was find a way to bring the design goals to life in the system," he said.
It was precisely this belief - the direction is clear, just keep moving forward - that sustained the team as it navigated uncharted territory step by step.
The road from zero to one is never smooth. Liu recalled a particularly tough test the team faced on the eve of deployment, just before the system was to go live. That day, testers suddenly reported a performance bottleneck. It was not a single-point failure; it was the kind of problem that rippled through the entire system, bringing many functions to a standstill. From morning until late at night, then from the early hours until daybreak, Liu and his teammates combed through the code line by line and tested again and again.
"As the launch drew near, it felt like the pressure had automatically spread to everyone," said Liu. "Everyone was staring intently at the system, and no one felt sleepy."
Finally, the problem was precisely identified and fixed. When the testers said the system had "passed," there were no cheers or high-fives. They simply exchanged glances and smiled. "That smile wasn't one of relief. It was the kind of smile that comes after winning a hard-fought battle," Liu told the Global Times. "I was really happy at that moment. Not just because the problem was solved and the system could go live, but because I confirmed something even more important: our team had cohesion. When it mattered most, we didn't fall apart. We could take on the toughest fights."
After countless days and nights, this group of Chinese engineers laid down a solid foundation for China's high-quality industrial development with technology and conviction. The "digital track" was finally in place.
Serve the industrial chain
In the "Build China into a Sci-Tech Powerhouse" included in Volume IV of Xi Jinping: The Governance of China, Xi noted that, "we should make solid efforts to promote the deep integration of sci-tech innovation and industrial innovation, and boost new quality productive forces."
On July 8, at a meeting that brought together the national science and technology award conference, the general assemblies of the members of the Chinese Academy of Sciences and the Chinese Academy of Engineering, and the 11th national congress of the China Association for Science and Technology, Xi further emphasized the need to promote the deep integration of sci-tech innovation with industrial innovation, opening up channels for accelerating the transformation of sci-tech achievements into real productive forces.
The value of technological innovation must ultimately be reflected in industry. From the very beginning, the Industrial Internet Identification and Resolution System was never designed to "publish papers." It was designed to address longstanding pain points in the development of the real economy.
According to Gao, enterprises have long faced three common challenges in the process of digital and intelligent transformation: data could not be found because different systems lacked a unified identity; data could not be connected because cross-company mutual recognition was absent; and data could not be put to good use because intelligent decision-making could not be formed. The Industrial Internet Identifier Resolution System was built precisely to tackle these three issues and provide foundational capabilities.
But for a new technology to truly integrate into industry, writing code in front of a computer is not enough. Over the past decade, Gao has led his team on repeated visits to key industries, pushing the system from "technically feasible" to "commercially usable." Liu and his colleagues have also visited many factory workshops and spoken face-to-face with business owners.
"Back in the office, what we saw were system metrics, interfaces, logs and data," Liu said. "But once we were on site, we saw individual parts, production lines and real business processes. It felt completely different."
During field research at enterprises, Liu and his coworkers witnessed firsthand how the system they had helped develop was enabling data connectivity across systems on production lines. They also learned that, in areas such as quality traceability, product management and supply-chain coordination, the system was helping companies reduce management costs and defect rates.
"At that moment, I felt that the identifiers were no longer just something on paper or on a platform - they were really being put to use. And once they start being used, they can go deeper and spread further," Liu recalled. As an R&D engineer, Liu said he felt proud to see companies using the system he had helped develop on the ground. "The late nights and the pressure we went through were all worth it."
Stories like this are countless in the system's rollout. In the new-energy vehicle sector, for example, unified identifiers give each battery pack a unique identity, allowing it to be "recognizable at all times and traceable throughout the chain." In high-end equipment manufacturing, data generated during the operation of large machinery is uploaded in real time through the system and analyzed to predict failure risks, enabling maintenance to be arranged in advance. In the pharmaceutical industry, identifiers make it possible to trace medicines through the entire process: from factory to pharmacy, every box can be verified.
Gao said that he is pleased to see more and more companies have shifted from "I have to use it" to "I want to use it." He told the Global Times the system not only helps enterprises cut costs and improve efficiency in practice, but more importantly, "some scenarios that used to rely on foreign identifier systems are now being gradually replaced by our independently controllable system."
From 'follower' to 'pacesetter'
A decade of dedicated work has yielded a remarkable harvest for China's Industrial Internet Identifier Resolution System.
The system now comprises 407 second-level nodes and more than 600,000 connected enterprises; the number of registered identities has surpassed 700 billion, with core indicators ranking among the world's best. The system has been deeply deployed across more than 40 industries, including automotive, electronics, pharmaceuticals and equipment manufacturing, and now covers every province on the Chinese mainland. Viewed globally, it has become one of the largest industrial internet identifier practices in terms of scale, the broadest in application and the most mature in industrial ecosystem.
Looking back on the past decade, Gao summed up the system's journey from "almost impossible" to global leadership as four key leaps.
The first was a strategic breakthrough, when the state explicitly incorporated the identity resolution system into its national strategy and completed the top-level design. The second was a technological breakthrough, marked by the independently designed national top-node architecture and advances in core protocols. The third was large-scale rollout, achieved through the innovative three-tier structure of national top nodes, second-level nodes and enterprise nodes. The fourth was value realization, as identities evolved from simply "enabling connectivity" to "creating value," with deep integration into blockchain, AI and other emerging technologies.
Each leap was driven not by chance, but by the coordinated advancement of technology, industry and ecosystem development. "We did not just build a system. We explored a China-specific path characterized by national strategic guidance, breakthroughs led by research institutions, industry leadership from enterprises and ecosystem co-construction across the industrial chain," Gao told the Global Times.
In Gao's view, the most essential "code" behind the building of this system is a steadfast commitment to greater self-reliance and strength in science and technology. "We did not simply copy foreign technologies. Instead, we focused on national strategic needs and pursued independent innovation, advancing basic research, key technological breakthroughs and industrial applications in parallel."
On July 8, at the meeting bringing together the national science and technology award conference, the general assemblies of the members of the Chinese Academy of Sciences and the Chinese Academy of Engineering, and the 11th national congress of the China Association for Science and Technology, Xi underscored that, "We must seize the historic opportunity, rise to the challenges of the times, accelerate efforts to achieve high-level self-reliance and strength in science and technology, and make steady progress toward the 2035 goal of becoming a leading country in science and technology."
In the field of the Industrial Internet Identifier Resolution System, China has, over the course of 10 years, transformed itself from a passive "follower" in global internet rules into a "pacesetter," completing the full journey from following, to running alongside, to leading in some areas.
Gao said the greatest significance of building this system lies not only in creating an independent infrastructure, but in forging a development path with Chinese characteristics. "Having our own identifier system means China can participate more proactively in global digital governance and contribute Chinese wisdom in new areas such as trusted identity, trusted data and intelligent interconnectivity, while providing a more solid technological foundation for building a community with a shared future in cyberspace."
On another ordinary evening at work, Liu's desk lamp was still on. He was fine-tuning a new data exchange protocol, while test results flickered continuously across his screen. Nights like this have long become routine for him.
Outside the window, the lights of the city were coming on one by one. Behind him, countless streams of data were flowing, being parsed and transmitted through the system - traces left by products across the global industrial chain, and the digital imprint of China's manufacturing sector as it moves toward high-quality development.
"I work on identity (system) for a day; I'll work on it for a lifetime." That is a phrase Liu often repeats. He says the road from "zero" to "one" has already been carved out; what matters now is to set out once again from "one." Beyond that road lies an even faster digital frontier.
。
二 | 想想美国的顶尖人工智能开发者吧。
三 | 他们将心血、汗水和巨额资金倾注到每一个前沿模型的新版本中。然而，他们几乎来不及喘口气，竞争对手就已经追了上来。如果只是收费的竞争对手紧追不舍，情况已经够糟了。但像Anthropic和OpenAI这样的公司还必须全力冲刺，才能勉强领先那些免费提供模型的实验室几步。 7月16日，中国初创公司月之暗面（Moonshot AI）发布了Kimi K3——一个仅比Anthropic和OpenAI的最新作品Fable和Sol分别略逊一筹的模型。其参数（或称“权重”）将在本月晚些时候免费公开，这意味着它可以被下载并在任何能够运行的服务器上使用。7月19日，中国互联网巨头阿里巴巴（Alibaba）也发布了其通义千问（Qwen）模型系列新成员的预览版，该公司称其性能仅次于Fable。
四 | 根据研究公司Epoch AI的分析，最强大的开源权重模型（通常来自中国）落后于技术前沿仅几个月（见图表）。
五 | 这些系统的采用正在迅速扩大。5月，在热门模型市场OpenRouter上，客户使用美国顶尖实验室提供的系统量与使用中国顶尖实验室的系统量大致相当——以tokens（行业首选输出单位）消耗量计算。到了6月，美国模型的使用量增加了三分之二——但中国模型增加了两倍。 这不仅引起美国领先模型制造商的警觉，也引起美国政府的担忧，他们担心国家在AI领域失去领先地位并为黑客攻击创造新机会。
六 | 据说特朗普（Trump）政府内部一些人正在推动各种措施来减缓中国模型的传播，例如将其制造商加入贸易黑名单或警告美国公司不要使用它们。然而，美国政府在授予Fable和Sol等先进模型访问权方面的不可预测做法，也在刺激人们采用中国的开源权重替代方案。 甚至在K3发布之前，中国的开源权重模型就已引起全球关注。GLM-5.2是另一家中国实验室Z.ai于6月推出的模型，在执行所谓的智能体任务（agentic tasks，如创建软件）方面，比美国另一家顶级模型制造商Google目前提供的任何产品都要出色。
七 | 来自另两家中国实验室DeepSeek和MiniMax的模型也很受欢迎。根据Firefox浏览器制造商Mozilla的报告，开源权重模型在推理和复杂智能体工作方面仍落后于前沿，但现在被普遍认为对许多任务“足够好”（good enough）。 它们的成本也低得多——在公司应对AI不断上升的成本时，这是一个重要优势。
八 | 另一家研究公司Artificial Analysis追踪选定模型在其一系列测试中执行每项任务的成本。DeepSeek提供的最强系统平均每个任务花费0.04美元。尚未开源的K3平均每个任务花费0.95美元。Fable的用户预计需要支付2.75美元。 部分成本差异来自几乎任何云提供商都可以出售对开源权重模型的访问权，从而压低利润率。例如，DeepSeek以每百万tokens 0.28美元的价格提供其v4 flash模型的访问权，而最便宜的第三方提供商则提供每百万0.18美元的价格。
九 | 更重要的是，许多开源权重模型的训练特别注重运行效率，部分原因是中国实验室因美国对先进芯片的限制而在获取计算能力方面面临挑战。
十 | 像“专家混合”（mixture of experts）模型这样的创新——仅激活AI系统中与查询相关的部分——已被开源权重开发者广泛采用，DeepSeek和阿里巴巴等实验室在这一技术上取得了重大进展。 然而，开源权重模型的吸引力更深层次。
十一 | 由于美国政府仓促试图将Fable模型仅限美国公民使用，导致该模型在全球范围内三周无法使用，这促使许多公司审视自己对单一实验室的依赖。这鼓励他们尝试可以通过多个提供商访问、甚至可能在内部服务器上运行的开源权重替代方案。
十二 | Mozilla首席技术官Raffi Krikorian指出，这一事件起到了“警示射击”（shot across the bow）的作用。 （注：本文编译自《经济学人》7月22日文章，原链接：https://www.economist.com/business/2026/07/21/americas-ai-labs-are-under-threat-from-cheap-chinese-rivals）。
Current article:http://5gvmk.shangyaohujunwei.bond/news/20260826_25318.html
Published on:12:35:01
奇葩说
刺客信条