# ANALYSIS: The MIIT Plan 2026-2028, Beijing's Industrial Weapon Against the West

> On June 18, 2026, seven Chinese government ministries — led by the Ministry of Industry and Information Technology (MIIT), with the National Development and Reform Commission (NDRC) and the Cyberspace Administration of China (CAC) — jointly published an Action Plan for Promoting

- Author: Maxime Marquette
- Publisher: MadMax (https://mad-max.co)
- Section: Analysis
- Published: 2026-06-24
- Canonical URL: https://mad-max.co/en/article/analyse-le-plan-miit-2026-2028-l-arme-industrielle-de-pekin-contre-l-occident

## Key takeaways

- The plan, published June 18, 2026, targets 450+ industrial platforms, 120 million industrial IoT devices, 55%+ platform penetration in key sectors, and 100+ pilot projects by 2028.
- Its three axes are innovation coordination (forcing giants such as Alibaba, Huawei, Baidu and Tencent to share resources), ecosystem coordination (yuan tokens, AI agents, algorithm standards), and coordinated overseas expansion.
- The author argues the AI deployment uses Doubao, Ernie Bot and Tongyi Qianwen, and runs on domestic chips such as Huawei Ascend, linking the plan to circumventing US export controls and the FDPR.
- The author contends that civil-military fusion doctrine makes the platforms' data potentially accessible to the state, creating an "industrial intelligence architecture."
- The author compares the EU's slower regulatory approach (DMA, Chips Act) unfavorably with the plan's pace and calls for a faster coordinated Western response.

## Introduction: When seven ministries rewrite the rules of the game

### A coordinated move that flew under the radar

On **June 18, 2026**, seven Chinese government ministries — led by the **Ministry of Industry and Information Technology (MIIT)**, with the **National Development and Reform Commission (NDRC)** and the **Cyberspace Administration of China (CAC)** — jointly published an **Action Plan for Promoting the Coordinated Development of Large, Medium and Small Enterprises in the Platform Economy (2026-2028)**. The document passed almost unnoticed in the noise of the week's geopolitical crises. And yet its ambitions are staggering: **raise the number of industrial platforms from 340+ to 450+ by 2028**, connect **120 million industrial IoT devices** (up from **100 million**), achieve **55%+ platform penetration in key sectors**, establish **3 resource-sharing lists** and launch at least **100 pilot projects**. This is not an industrial roadmap. It is a geopolitical declaration wrapped in the language of bureaucratic coordination.

The plan was first flagged by **Newsquawk** on **June 18, 2026**, and later analyzed in depth by the newsletter **Sinocism**. It fits directly into a broader strategic framework that Beijing's **State Council unveiled on January 13, 2026** for integrating the real and digital economies. The timing is deliberate: as **US** export controls tighten, as the European Union deploys its **Digital Markets Act**, as Western governments attempt to decouple critical supply chains from China, Beijing responds not with retreat but with industrial mobilization at a scale that dwarfs any comparable Western effort. **The plan is not a response to Western pressure. It is a countermove designed to make that pressure irrelevant.**

### Three strategic axes — and what lies beneath them

The plan is organized around three strategic axes. The first is **innovation coordination**: forcing tech giants — **Alibaba, Huawei, Baidu, Tencent** — to share computing resources, data infrastructure, and technical capabilities with smaller enterprises. The second is **ecosystem coordination**: deploying yuan-denominated token services, **AI** agents, and algorithm transparency standards across all **450** platforms. The third — the one that should concern Western policymakers the most — is labeled **coordinated overseas expansion**. That phrase, in the context of a plan linking **450** industrial platforms to state-directed **AI** rollout and civil-military fusion doctrine, is not a commercial ambition. It is a projection strategy.

The **AI** dimension of the plan is explicit and central. General-purpose large language models — **Doubao** (ByteDance), **Ernie Bot** (Baidu), **Tongyi Qianwen** (Alibaba) — plus sectoral **AI** models and autonomous **AI** agents are to be deployed across all **450** industrial platforms. This is not about chatbots for consumer apps. This is about embedding **AI** decision-making into **the physical-digital infrastructure of Chinese industrial production** — logistics, manufacturing, supply chains, finance. And the **Pentagon added 188 Chinese companies** to its military-linked list in June **2026**, including Alibaba, Baidu, and **BYD** — precisely the companies whose platforms the **MIIT** plan designates as the backbone of this rollout.

*Seven ministries. One plan. 450 platforms. 120 million connected devices. Beijing does not move this way by accident — this level of cross-ministerial coordination is rare in any government on earth, authoritarian or democratic. When it happens in China, it means Xi Jinping has personally decided this is a strategic priority. And when Xi decides something is a strategic priority, the question is not whether it will happen. The question is how fast.*

## The three strategic axes: coordination, ecosystem, expansion

### Forcing the giants to share — and why that matters

The first axis of the **MIIT** plan — innovation coordination — targets a structural problem that China shares with the West: the concentration of digital infrastructure and **AI** capabilities in a handful of dominant platforms. Alibaba's cloud computing infrastructure, Huawei's industrial IoT stack, Baidu's **AI** platform, Tencent's payment and data ecosystem — these giants have, over the past decade, built walled gardens of capability that smaller Chinese manufacturers struggle to access. The **MIIT** plan mandates a structured sharing of these resources. **It is, in effect, a forced opening of the platform economy's crown jewels — not by market competition, but by state directive.**

This approach has no real equivalent in the Western tech world. Imagine **Washington** ordering Google, Amazon, and Microsoft to share their cloud **AI** capabilities with every **US** manufacturer below a certain revenue threshold, under standardized terms, within a two-year deadline. The antitrust debates alone would take a decade. In China, seven ministries issue the order and the giants comply — or face consequences that corporate lobbying cannot deflect. Whether this top-down approach produces the innovation China needs, or whether it creates bureaucratic coordination nightmares that slow the very SMEs it claims to empower, is a genuinely open question. But the speed and scale of the attempt are real.

### Ecosystem coordination — the yuan token and the algorithm layer

The second axis — ecosystem coordination — is technically dense but strategically clear. The plan calls for deploying yuan-denominated token services across industrial platforms, integrating **AI** agent standards, and imposing algorithm transparency requirements on the **450** designated platforms. The yuan token dimension is particularly significant: it is a direct effort to reduce the dollar-denominated friction in Chinese industrial supply chains, building a payment and settlement infrastructure that operates entirely within the renminbi ecosystem. In the context of **US** secondary **sanctions** and **SWIFT** exclusion risks, this is not just financial engineering. It is **sanctions**-proofing at the industrial level.

The algorithm transparency requirements are also worth noting — not because Beijing has suddenly become a champion of digital rights, but because standardizing algorithm interfaces across **450** platforms creates the technical conditions for state oversight of every **AI** decision made across Chinese industrial infrastructure. **Transparency to the state, not to the users. Algorithmic standardization that enables central monitoring. This is the architecture of the digital command economy — and it is being built platform by platform, standard by standard.**

*Yuan tokens across 450 industrial platforms. Algorithm standards that open every AI decision to state review. Anyone who thinks this is just about economic efficiency hasn't read enough Chinese policy documents. It's also about control — the ability of the Party to see, monitor, and ultimately redirect what 450 platforms do with 120 million connected devices. That's not conspiracy theory. That's the explicit logic of civil-military fusion doctrine, written into law since 2017.*

## Artificial intelligence at the core: the battle of foundation models

### General LLMs, sectoral models, and the 450-platform deployment

The **MIIT** plan's **AI** dimension is its most consequential axis. The plan explicitly designates **Doubao** (ByteDance), **Ernie Bot** (Baidu), and **Tongyi Qianwen** (Alibaba) as the general-purpose foundation models to be deployed across the **450** industrial platforms. These are not experimental systems. **Doubao** reported over **100 million** monthly active users in early **2026**. **Ernie Bot** has been deeply integrated into Baidu's industrial and cloud services. **Tongyi Qianwen** underpins Alibaba's enterprise offerings across logistics, manufacturing, and financial services.

Beyond general-purpose LLMs, the plan mandates the development and deployment of **sectoral AI models** — models specialized for manufacturing, energy, transportation, agriculture — and **AI agents**: autonomous software systems that go beyond answering questions to actively managing inventory, scheduling maintenance, optimizing logistics flows, and executing financial transactions. This distinction matters enormously. An **AI** agent embedded in a manufacturing platform is not a productivity tool. It is an autonomous decision-maker integrated into the operational fabric of the industrial economy. **DeepSeek's 2025 breakthrough** — demonstrating that frontier **AI** performance could be achieved at a fraction of the compute cost — showed that China's **AI** capabilities are not merely derivative of Western models. They are developing their own trajectory. The **MIIT** plan is the deployment vehicle for that trajectory.

### Huawei Ascend, Biren, Cambricon — the domestic chip foundation

The **AI** deployment the **MIIT** plan envisions requires massive computational infrastructure. And here, the plan intersects directly with the **US** export control regime and its gaps. **Huawei's Ascend 910B and 910C chips** — developed to replace banned Nvidia H100 GPUs — are the primary domestic alternative for training and deploying large **AI** models in China. **Biren Technology** and **Cambricon** produce additional **AI** accelerator chips for inference workloads. None of these chips match the performance of Nvidia's H100 or H200 at equivalent power envelopes. But for the industrial AI applications the MIIT plan targets — sectoral models running inference on manufacturing platforms, AI agents managing logistics — the gap is narrower and the availability is what matters.

The Pentagon added **Biren Technology** to its military-linked entity list in June 2026. **Huawei** remains under FDPR restrictions. And yet the Ascend chip supply continues to scale, backed by China's **Big Fund** (~$47 billion in its third phase) and by state procurement mandates that guarantee a domestic market even if export markets remain inaccessible. **The MIIT plan does not depend on Western chips. It is designed to work with what China can build — and to build more of it faster than the export control regime can anticipate.**

*I've read dozens of Western analyses of China's AI chip gap. They're mostly right on the technical specs. Where they sometimes go wrong is in assuming that the gap translates mechanically into strategic limitation. The MIIT plan is not trying to build GPT-5. It's trying to automate 450 industrial platforms with AI agents that run on Ascend chips and domestic LLMs. That's a different — and in some ways more immediately achievable — ambition. The performance bar for industrial AI is lower than for frontier reasoning. And lower bars are easier to clear.*

## Chinese SMEs as vectors of platform dominance

### The hidden leverage in the supply chain

One of the MIIT plan's most underappreciated dimensions is its explicit targeting of small and medium enterprises as vectors of platform adoption. The plan's title — promoting **coordinated development of large, medium and small enterprises** — is not rhetorical filler. It describes a deliberate strategy: use state-mandated resource sharing from the giants to onboard hundreds of thousands of SMEs onto the 450 designated platforms, then leverage those SMEs' international supply chain connections to extend the platforms' data reach globally. A Chinese SME supplier to a German automotive manufacturer, running its operations on a Huawei industrial platform with embedded AI agents, is a data node in a network that extends beyond China's borders — regardless of where the final car is assembled.

This is the mechanism behind the MIIT plan's third axis — **coordinated overseas expansion**. Not direct market entry by Chinese platforms (which faces increasing regulatory resistance in Western markets) but indirect expansion through the supply chain tentacles of Chinese SMEs that are already deeply embedded in global manufacturing. **The 120 million IoT devices the plan targets by 2028 are not all in China. Many are already in the factories, logistics hubs, and warehouses of companies that source components from China.** The MIIT plan is, in part, a plan to turn those devices into data-gathering and AI-processing nodes in a network that Beijing can monitor, direct, and, under civil-military fusion doctrine, potentially mobilize for intelligence purposes.

### Resource-sharing lists and the discipline of coordination

The three resource-sharing lists the MIIT plan establishes — covering computing resources, data assets, and technical capabilities — are the operational mechanism for forcing the giants to share with SMEs. These lists are not voluntary. They are state directives backed by the regulatory power of seven ministries. Companies that fail to comply face the same consequences that have historically met Chinese tech giants that resisted Party directives: regulatory investigations, antitrust actions, public humiliations of CEOs, market access restrictions. **Alibaba's experience in 2020-2022 — $2.75 billion antitrust fine, forced restructuring, Jack Ma's disappearance from public life — established the template for what non-compliance looks like.** The MIIT plan's resource-sharing mandate is not an invitation. It is a command with a historical track record of enforcement.

For the 100+ pilot projects the plan mandates, the resource lists create a testing ground for the platform coordination model before national rollout. By 2028, if the plan proceeds on schedule, the 450 platforms will have been tested against real industrial use cases, with AI agents deployed at scale, with yuan token payment systems operating, with algorithm transparency standards enforced. That is not a prototype. It is the architecture of a fully operational industrial digital economy — one that has been stress-tested, optimized, and embedded in the operational muscle memory of Chinese manufacturing. **The West has no equivalent program.**

*The resource-sharing lists remind me of something a supply chain analyst told me once: the most dangerous competitor is not the one who beats you at your own game. It's the one who changes the rules of the game so that your strengths become irrelevant. Western tech platforms compete on market efficiency, network effects, investor returns. Chinese platforms will compete on state-mandated coordination, resource sharing, and embedded AI agents — all backed by the regulatory muscle of seven ministries. It's not a fair fight. It was never designed to be.*

## The MIIT plan in the context of US export controls

### China sanctions the US back — the escalation ladder

The MIIT plan was published against a backdrop of accelerating mutual economic sanctions. In June 2026, **China sanctioned 10 US firms** and restricted **46 US companies from public procurement** in response to the Pentagon's military-linked entity list expansion. **Alibaba filed suit against the Pentagon** over its inclusion on that list. The US added **188 Chinese companies** — including Alibaba, Baidu, and BYD — to the military-linked list. And Congress continued debating an extension of the **CHIPS Act** ($52 billion) and new restrictions on AI chip exports.

This mutual sanctions escalation is the geopolitical context in which the MIIT plan was designed. It is explicitly a response to — and an attempt to make irrelevant — the US export control regime. **If you cannot import the most advanced Nvidia chips, you build domestic alternatives. If you cannot use SWIFT-denominated payment systems, you build yuan token infrastructure. If Western cloud platforms face regulatory pressure in China, you mandate domestic platforms and force the giants to share their capabilities.** Every axis of the MIIT plan has a corresponding US or EU policy that it is designed to neutralize. This is not paranoia. It is the explicit logic of technological self-sufficiency that Beijing has articulated since 2015 under the Made in China 2025 program.

### The FDPR — Beijing's most feared instrument

The **Foreign Direct Product Rule (FDPR)** — which extends US export control jurisdiction to any product made anywhere in the world using US technology — is the instrument Beijing fears and respects most. It was the FDPR that allowed Washington to cut off TSMC's supply of advanced chips to Huawei and SMIC. It is the FDPR that gives the US extraterritorial reach into the supply chains of allies and partners. And it is the FDPR that the MIIT plan is most directly designed to circumvent — by building an AI chip and industrial platform infrastructure that uses zero US-origin technology wherever possible.

The **Huawei Ascend** chips are designed to FDPR-proof China's AI infrastructure. The domestic LLMs are trained on Chinese data with Chinese architecture. The yuan token system removes dollar-denominated friction from the settlement layer. The algorithm standards are set by Chinese regulatory bodies, not by W3C or IEEE. **The MIIT plan is the most comprehensive FDPR-mitigation strategy ever assembled by a sovereign government — and it is being executed not by a single ministry, but by a coordinated coalition of seven.**

*The FDPR is Washington's most powerful extraterritorial tool. And Beijing's most systematic response to it is not a legal challenge at the WTO — it's the MIIT plan. Not fighting the rule. Building around it. There's something both admirable and deeply unsettling about that level of strategic patience. The US imposed the FDPR. China spent three years building the infrastructure to make it irrelevant. The race between restriction and circumvention is the defining economic contest of this decade.*

## Implications for global supply chains

### The fragmentation accelerates — and the MIIT plan is a catalyst

The MIIT plan does not exist in isolation. It is the Chinese domestic side of a global supply chain fragmentation that is already underway — driven by US export controls, European industrial policy, India's "Make in India" push, and the general trend toward friend-shoring and near-shoring that has accelerated since the COVID supply chain disruptions of 2020-2022. But where Western policies are largely defensive — securing supply chains against dependence on China — the MIIT plan is offensive. It is designed to make Chinese industrial platforms so deeply embedded in global supply chains, through SME integration and IoT device deployment, that the cost of decoupling from them becomes prohibitive for Western manufacturers.

The **120 million IoT devices** the plan targets by 2028 are the physical expression of this strategy. Each device is a data point in a network. Each AI agent running on those devices is a decision-maker embedded in an industrial process. If those processes run on Huawei industrial platforms with Alibaba cloud infrastructure, replacing them is not a regulatory decision. It is a multi-year, multi-billion-dollar operational undertaking that most manufacturers — under competitive pressure from cheaper Chinese alternatives — will rationally choose to avoid. **The MIIT plan is an attempt to make decoupling economically irrational before the West can make it politically mandatory.**

### Critical minerals, rare earths, and the platform convergence

There is a dimension of the MIIT plan that connects to the broader resource competition between China and the West: the plan's industrial platform ecosystem covers sectors that include **mining, processing, and logistics for rare earths and critical minerals** — sectors where China already dominates global supply chains. If AI agents deployed on MIIT-designated platforms are managing the logistics and processing of rare earth elements — materials essential for EV batteries, semiconductor manufacturing, and military hardware — then the platform economy becomes a control layer over the physical resource economy.

This convergence is not speculative. China processes approximately **85% of the world's rare earth elements**. Its dominance in battery materials is even more complete. The MIIT plan's industrial platform ecosystem, if it achieves its 2028 targets, will embed AI agent decision-making into the supply chains that the West depends on for the physical inputs of its own tech and defense industries. **Digital control of physical resources is the strategic endpoint of the platform economy — and the MIIT plan is a significant step toward that endpoint.**

*I keep coming back to the phrase "coordinated overseas expansion." It appears once in the plan, almost as an afterthought. But read it in the context of civil-military fusion doctrine, the Pentagon's military entity list, and the 120 million IoT devices — and it stops being an afterthought. It's the whole point. The overseas expansion is not commercial. It's structural. It's about embedding Chinese platform architecture into global supply chains before Western decoupling policies can reach those chains. The phrase is quiet. The strategy is not.*

## US export controls: pressure that keeps mounting

### October 2022, October 2023, 2024 — and beyond

The US export control architecture — built around the **Export Administration Regulations (EAR)** administered by the **Bureau of Industry and Security (BIS)** — has been progressively tightened since **October 2022**. Successive rounds of restrictions have targeted advanced semiconductor equipment, AI chips, and high-bandwidth memory. The **CHIPS Act** ($52 billion) funded domestic US semiconductor manufacturing as the supply-side complement to the demand-side restrictions. In June 2026, senators **Jim Banks and Andy Kim** were pushing for stricter rules on foundries like TSMC to prevent advanced chips from reaching Chinese-controlled foreign subsidiaries.

These controls have had real effects. Direct US chip equipment exports to China have fallen to their lowest level in eight years. China's access to the most advanced GPU nodes — Nvidia's H100, H200, B200 — remains constrained. **SMIC's achievement of 7nm production using SADP (Self-Aligned Double Patterning) with DUV equipment** — circumventing the EUV ban — demonstrates that the controls delay rather than prevent Chinese semiconductor advancement. But each year of delay is a year of maintained Western advantage in the AI race. And the MIIT plan's domestic chip strategy — Ascend, Biren, Cambricon — is the long-term answer to that delay.

### What the BIS can and cannot do

The BIS Entity List — which subjects designated companies to additional export licensing requirements — now includes over **600 Chinese entities**. But the list's effectiveness is constrained by the same structural problem that limits all unilateral export controls: **enforcement depends on the compliance of intermediaries, and intermediaries are located in jurisdictions where US regulatory reach is incomplete.** The Southeast Asia bypass documented by **Nikkei Asia in April 2026** — Chinese chip equipment imports via Malaysia and Singapore surpassing direct US imports — illustrates the limits of even the most sophisticated unilateral export control architecture.

The BIS faces a structural dilemma: tighten controls enough to be effective, and you impose costs on US companies that lose Chinese market revenues; maintain commercial relationships, and you risk enabling the very capabilities you are trying to restrict. This dilemma has no clean resolution. It can only be managed — through multilateral coordination with allies, through positive economic incentives for compliant intermediaries, and through investment in enforcement capacity. None of these solutions are fast. All of them are politically costly. **The MIIT plan is Beijing's bet that Washington will not sustain the political will to maintain and expand the control regime long enough to matter.**

*The BIS Entity List. I look at the 600+ Chinese entities on it and I think: this is a policy instrument fighting a network effect. You can add 600 companies to a list. You cannot add 120 million IoT devices to a list. You cannot add an AI agent embedded in a manufacturing platform to a list. The export control regime is fighting the MIIT plan's world with the tools of the pre-MIIT world. That asymmetry is the central strategic problem.*

## The European response: regulation versus competition

### The DMA's 22 gatekeepers versus Beijing's 450 platforms

The **European Union's Digital Markets Act (DMA)**, which designated **22 gatekeepers** — Alphabet, Amazon, Apple, Meta, Microsoft, and others — in 2023-2024, represents Europe's primary instrument for managing platform power. Its logic is regulatory: impose interoperability requirements, data-sharing obligations, and anti-self-preferencing rules on dominant platforms to enable competition. It is, in many ways, a mirror of the MIIT plan's first axis — forcing large platforms to share resources with smaller actors.

But the comparison stops there. The DMA is a competition regulation tool, designed to prevent market abuse. The MIIT plan is a **strategic industrial policy**, designed to create a nationally coordinated platform economy capable of global expansion. The DMA has 22 designated gatekeepers. The MIIT plan has 450 designated platforms. The DMA operates through legal proceedings that take years. The MIIT plan operates through ministerial directives that take months. **Europe is trying to regulate its way to a competitive digital economy. China is planning its way there. These are not equivalent strategies.**

### The European Chips Act — and why it is not enough

The **European Chips Act** — which aims to double Europe's share of global semiconductor production to 20% by 2030 — is the EU's supply-side response to the chip competition. It has committed approximately €43 billion in public and private investment. Intel's planned fab in Magdeburg (Germany), TSMC's Dresden fab, and STMicroelectronics' Crolles expansion are its flagship projects. These are real investments that will produce real chips. But they will not be operational at scale before 2027-2028, and their initial production will focus on mature node technologies (28nm-12nm) rather than the frontier 3nm-2nm nodes that the most advanced AI applications require.

Meanwhile, the MIIT plan's AI deployment does not wait for frontier chips. It deploys on the chips China already has — Huawei Ascend, domestically produced DRAM, mature-node processors — for industrial applications that do not require frontier performance. **Europe is investing to compete in a chip race while China is winning an industrial platform deployment race that uses whatever chips are available.** The misalignment between Western competitive strategy and Chinese competitive strategy is the core of the current technological confrontation — and the MIIT plan is the clearest articulation of that misalignment.

*Europe is debating the DMA and building chip fabs while China deploys AI agents in 450 industrial platforms. I don't say this to be dismissive of European policy — the DMA is serious, the Chips Act is necessary. But the pace differential is real. A DMA enforcement action takes three to five years. A MIIT pilot project takes 12 months. That's not a policy gap. It's a strategic clock gap. And clocks that run at different speeds eventually produce very different outcomes.*

## What the MIIT plan means for national security

### Civil-military fusion and the 450 platforms

China's **civil-military fusion doctrine (军民融合)**, codified in law since 2017 and elevated to a core state strategy under Xi Jinping, requires that all Chinese civilian enterprises make their technologies, data, and capabilities available to the military upon state request. This is not a theoretical provision. It has been applied to Huawei's telecom infrastructure, to ByteDance's data systems, to DJI's drone technology. The MIIT plan's 450 industrial platforms — and the 120 million IoT devices they connect — are, by definition, subject to civil-military fusion obligations.

What does this mean in practice? It means that the operational data flowing through those 450 platforms — logistics flows, manufacturing throughput, supply chain inventories, financial transaction patterns — is potentially accessible to Chinese state security services upon request. It means that AI agents embedded in global supply chains and running on MIIT-designated platforms are operating in an architecture that is structurally linked to Chinese intelligence collection capabilities. **This is not speculation. It is the explicit legal framework within which the MIIT plan operates.** Western companies doing business with Chinese suppliers running on these platforms are operating within that framework, whether or not they acknowledge it.

### The intelligence architecture of the platform economy

The convergence of the MIIT plan's physical-digital infrastructure with civil-military fusion doctrine creates what analysts are beginning to call an **industrial intelligence architecture**: a network of connected platforms, devices, and AI agents that generates real-time data on global supply chains while being legally accessible to Chinese state security. This architecture does not require cyberattacks. It does not require espionage in the traditional sense. It requires only that Western companies continue doing business with Chinese suppliers running on MIIT-designated platforms — which most of them will, because the cost of not doing so is too high.

The Pentagon's June 2026 decision to add Alibaba, Baidu, and BYD to the military-linked list is an acknowledgment of this architecture. But the list is a labeling exercise. It does not sever the supply chain connections that make the architecture operational. **Labeling a risk is not the same as mitigating it. And the gap between the Pentagon's list and the operational reality of global supply chains running on MIIT-designated platforms is the space in which China's industrial intelligence architecture will continue to expand.**

*Civil-military fusion. It's a phrase that appears in every serious analysis of Chinese tech policy — and it remains underweighted in corporate risk assessments. Because if you take it seriously, the implications for supply chain compliance are enormous and expensive. Most companies prefer to note the risk and move on. I understand why. I also understand that "noting the risk and moving on" is exactly the behavior the MIIT plan was designed to exploit.*

## The 450 platforms and 120 million devices: architecture of digital dominance

### From industrial internet to physical-digital integration

The MIIT plan's target of **450 industrial platforms** connected to **120 million IoT devices** is not an abstraction. It describes a specific physical-digital integration architecture: manufacturing equipment, logistics systems, energy grids, agricultural sensors, transportation networks — all connected through standardized platform interfaces, all generating data that flows into AI systems capable of optimizing, predicting, and autonomously managing industrial processes. This is the **Industrial Internet of Things** at national scale, with AI agents as the operating layer.

The **55%+ penetration target** across key industrial sectors means that by 2028, more than half of China's industrial activity — by volume or value, depending on the sector — will run on MIIT-designated platforms. At that penetration level, the platforms are not optional infrastructure. They are the operating environment of Chinese industry. Companies that want to do business with Chinese manufacturers will increasingly need to interface with these platforms. And interfacing with these platforms means operating within the civil-military fusion architecture — with all the data exposure and intelligence implications that entails.

### Why 120 million devices is not a number — it's a strategy

The progression from **100 million to 120 million connected industrial IoT devices** between 2024 and 2028 represents a **20% increase in the data-generating and AI-processing infrastructure** of the Chinese industrial economy. But the strategic significance is not the 20% incremental increase. It is the **network effect** that comes from cross-platform data integration at scale. AI systems trained on data from 120 million connected devices across 450 platforms develop pattern recognition capabilities that no single platform or company can replicate. The MIIT plan is, among other things, a data aggregation strategy: create the conditions for the world's largest industrial AI training dataset, then deploy models trained on that dataset across global supply chains via SME integration.

**No democratic government has a comparable program.** The US CHIPS Act funds chip manufacturing. The EU AI Act regulates AI deployment. Neither has a plan for coordinating 450 industrial platforms across 120 million connected devices with AI agents embedded in every node. This asymmetry — between China's offensive industrial AI strategy and the West's defensive regulatory response — is the core strategic challenge of the coming decade. And the MIIT plan makes it more visible than any previous Chinese policy document.

*120 million IoT devices. I try to visualize what that number means physically. It's more connected industrial sensors than exist in the entire European Union. It's more data-generating nodes than any Western industrial policy has ever contemplated coordinating. And it's all going to run on platforms that are legally required to share their data with the Chinese state upon request. That's not a tech policy story. That's a geopolitical architecture story. And the MIIT plan is the blueprint.*

## AI agents: the qualitative leap of the MIIT plan 2026-2028

### Beyond chatbots — autonomous industrial decision-making

The AI agent dimension of the MIIT plan deserves particular attention because it represents a qualitative leap beyond the conversational AI applications that most Western commentary focuses on. An AI agent is not a chatbot. It is an **autonomous software system** that perceives its environment through sensor data, makes decisions according to trained objectives, executes actions in real-world systems, and learns from the outcomes of those actions. Deployed in an industrial context — managing a warehouse, optimizing a logistics network, scheduling production runs, executing financial settlements — an AI agent operates continuously, at machine speed, without human intervention for routine decisions.

The MIIT plan's mandate to deploy AI agents across 450 industrial platforms means embedding autonomous decision-making into the operational fabric of Chinese industry at a scale that has no precedent. **Doubao, Ernie Bot, and Tongyi Qianwen are not just deployed as conversational tools. Their agent frameworks — ByteDance's Coze, Baidu's ERNIE Agent, Alibaba's Tongyi Agent — are the operating systems for this autonomous industrial layer.** The difference between an AI agent and a chatbot is the difference between a system that answers your questions and a system that runs your factory. The MIIT plan is deploying the latter at national scale.

### The agent governance gap — a Western blind spot

Western AI governance frameworks — the EU AI Act, the US Executive Order on AI, the UK AI Safety Institute's work — are primarily focused on high-risk AI applications in consumer-facing and public sector contexts. They are not designed to govern the deployment of autonomous AI agents across industrial supply chains at the scale the MIIT plan envisions. **The EU AI Act categorizes most industrial AI applications as "limited risk" or "minimal risk" — categories that attract disclosure requirements and transparency obligations, but not the stringent oversight applied to high-risk systems.**

This categorization made sense when industrial AI was primarily about predictive maintenance and quality control. It does not make sense when industrial AI agents are making autonomous procurement decisions, managing cross-border logistics, and executing financial settlements in real time across supply chains that span multiple jurisdictions. The governance gap is real, and it will become more visible as the MIIT plan's AI agent deployment matures. **The West has regulations for the AI of 2022. China is deploying the AI of 2028. That gap is not primarily a technology gap. It is a governance imagination gap.**

*AI agents running factories. AI agents managing supply chains. AI agents executing financial settlements. I'm not describing science fiction — I'm describing the explicit targets of the MIIT 2026-2028 plan. And I'm also describing something that our regulatory frameworks were not designed for. The EU AI Act was a serious effort to govern AI responsibly. It will need a serious update — because the deployment environment it was designed for no longer exists.*

## The relationship with Huawei: a revealing parallel

### How Huawei prefigured the MIIT strategy

**Huawei's trajectory since 2019** — when the Trump administration placed it on the Entity List and cut off its access to advanced chips and Google's Android ecosystem — is the most instructive preview of how China responds to Western technology restriction. Within three years, Huawei had developed its own operating system (HarmonyOS), its own application ecosystem, its own 5G chip (Kirin 9000S, manufactured by SMIC at 7nm), and its own cloud services infrastructure. The restriction that was supposed to cripple Huawei instead accelerated its self-sufficiency program. **By 2025, Huawei was the dominant industrial IoT platform provider in China — the largest single contributor to the network of connected devices that the MIIT plan now targets.**

This Huawei parallel is the template for the MIIT plan. The logic is identical: Western restrictions on the most advanced components → accelerated domestic substitution → deployment of domestically developed alternatives at scale → achievement of functional self-sufficiency in target applications. The MIIT plan applies this logic at the level of the entire platform economy, not just one company. **Huawei was the proof of concept. The MIIT plan is the national rollout.** And the proof of concept worked well enough that seven Chinese ministries are now coordinating to replicate it across 450 platforms.

### Civil-military fusion and the Huawei model

Huawei's civil-military fusion obligations are the most documented case of this doctrine in action. The company's founder, **Ren Zhengfei**, stated publicly in 2019 that Huawei would comply with Chinese law requiring companies to cooperate with national intelligence activities. This statement, made in the context of Western governments banning Huawei from 5G networks, was honest but also clarifying: a Chinese company has no legal option to refuse state security demands under civil-military fusion doctrine. The same obligation applies to every company operating on the 450 platforms the MIIT plan designates. **Huawei demonstrated what civil-military fusion looks like for a single company. The MIIT plan demonstrates what it looks like for an entire industrial economy.**

The Western response to Huawei — exclusion from 5G networks, Entity List designation, chip supply cutoffs — was a reactive measure taken after Huawei's infrastructure was already deeply embedded in the networks of dozens of countries. The MIIT plan's industrial platform deployment is proceeding faster, with more state coordination, and with deeper integration into global supply chains than Huawei's 5G rollout ever achieved. The window for a comparable reactive Western response is narrowing with each pilot project completed.

*Huawei in 2019. The MIIT plan in 2026. Same logic, different scale. We had seven years to learn from the Huawei experience — to build industrial platform governance frameworks, to develop supply chain security standards, to create positive alternatives to Chinese platform dependency. I'm not sure we used those years well. The MIIT plan suggests that Beijing, at least, has been paying attention to what worked and scaling it up.*

## Chinese SMEs as vectors of international expansion

### The supply chain Trojan horse — and why it's not actually hidden

The MIIT plan's strategy of using Chinese SMEs as vectors of international platform expansion is not hidden. It is explicitly stated in the third axis: **coordinated overseas expansion**. The mechanism is straightforward: Chinese SMEs that adopt MIIT-designated platforms gain access to AI agent tools, computing resources, and financing instruments that make them more competitive in international supply chains. As they win more contracts with Western manufacturers, their platform infrastructure — Huawei industrial IoT, Alibaba cloud services, Baidu AI agents — becomes embedded in the operational processes of those manufacturers' supply chains. The data generated flows back into Chinese AI training datasets. The platform architecture extends beyond China's borders.

This is not a conspiracy. It is a rational industrial policy. Every country with a platform economy advantage — the US, the EU — does exactly this: deploys its domestic platform infrastructure through international commerce and ensures that the data generated in that commerce flows back to domestic AI training. The difference is that in the US and EU, this process is driven by market competition among private companies. In China, it is driven by state directive among companies that are legally required to share their data with the state. **Same mechanism, fundamentally different governance implications.** And the MIIT plan accelerates this process by reducing the barriers for Chinese SMEs to adopt and integrate these platforms — making them more competitive, and their platform infrastructure more pervasive, faster.

### The 100 pilot projects as international templates

The **100+ pilot projects** the MIIT plan mandates are not just domestic experiments. They are templates for international deployment. Each pilot project that successfully demonstrates AI agent management of a supply chain, IoT-connected manufacturing process, or cross-platform logistics optimization becomes a documented use case that Chinese SMEs can present to international partners. **The pilot project is the sales pitch.** A German automotive manufacturer looking to source components from China, comparing suppliers — one running on a traditional ERP system, one running on a MIIT-designated platform with AI agent optimization, real-time IoT connectivity, and lower operational costs — faces a commercial decision that increasingly favors the platform-enabled supplier, regardless of the geopolitical context.

This is the competitive pressure the MIIT plan is designed to generate. Not a frontal assault on Western platform dominance — which would trigger defensive responses — but a gradual, supply-chain-by-supply-chain infiltration that makes platform adoption the rational commercial choice for anyone sourcing from China. **By the time Western policymakers recognize the security implications, the commercial integration will be deep enough that untangling it is more costly than accepting it.** That is the strategic timeline the MIIT plan is operating on.

*The 100 pilot projects. I keep thinking about the German automotive engineer sitting across a table from a Chinese supplier, looking at the cost savings that come from AI agent optimization on a Huawei industrial platform. That engineer is not making a geopolitical decision. He's making a procurement decision. And procurement decisions, made ten thousand times across European industry over the next three years, become a geopolitical fact. The MIIT plan doesn't need to defeat Western policy. It just needs to make Western supply chains dependent enough on Chinese platforms that policy cannot reverse the dependence.*

## Western digital sovereignty: an institutional response under construction

### The EU's response — serious but slow

The European Union's response to the challenge the MIIT plan represents is the most comprehensive among Western governments — and also the most constrained by its own institutional architecture. The **DMA**, the **EU AI Act**, the **European Chips Act**, the **Cyber Resilience Act**, and the proposed **Cloud Certification Scheme** together form a regulatory framework for the digital economy that is technically coherent and strategically relevant. But each of these instruments operates through the EU's legislative and regulatory process — a process that produces solid, enforceable rules over three-to-five-year timelines. The MIIT plan operates on 24-month deployment cycles.

The pace differential is the structural problem. **The EU has the regulatory ambition and the market size to shape global digital standards — it has done so with GDPR, with the DMA, with the CSRD.** But it does not have a **strategic industrial policy** comparable to the MIIT plan — a coordinated, resource-backed, cross-ministerial plan to build and deploy a European industrial platform ecosystem. The Gaia-X cloud initiative, the IPCEI-CIS microelectronics program, and the European Industrial Data Space are partial attempts in this direction. They remain fragmented, underfunded relative to their stated ambitions, and slower than the challenge they are meant to address.

### What Western digital sovereignty actually requires

Digital sovereignty is a phrase that has proliferated in Western policy discourse over the past five years without producing the institutional substance its urgency requires. What the MIIT plan makes clear is that **digital sovereignty is not achievable through regulation alone**. It requires: a coordinated industrial platform strategy that matches the MIIT plan's 450-platform ambition; positive economic incentives that make Western platform infrastructure as commercially attractive as Chinese alternatives; supply chain security standards with real enforcement mechanisms; and a governance framework for AI agents in industrial supply chains that the West does not yet have.

None of these requirements are beyond the institutional capacity of Western democracies. The CHIPS Act demonstrated that the US can move quickly on strategic industrial policy when the political will exists. The DMA demonstrated that the EU can create globally significant regulatory frameworks. AUKUS demonstrated that allied coordination on technology security is possible. **What is missing is not the capacity to respond. What is missing is the recognition that the MIIT plan represents a challenge that requires a response at its own scale — and the political will to act at that scale before the window closes.**

*Digital sovereignty. I've heard this phrase in thirty policy conferences over the past four years. Each time, I ask the same question: what specific platform, deployed by what deadline, with what resources, replaces the Chinese alternative? I've rarely gotten a specific answer. The MIIT plan has specific answers. 450 platforms. 120 million devices. 2028 deadline. Three resource-sharing lists. 100 pilot projects. Specificity is not everything. But the absence of specificity is not nothing. It means the urgency hasn't landed yet.*

## Conclusion: The plan that names the game

### Why the MIIT plan matters more than it seems

The MIIT Action Plan for 2026-2028 will not make headlines the way a military exercise around Taiwan does. It will not generate a Security Council debate or a G7 communiqué. It is a bureaucratic document, published by seven Chinese ministries, about industrial platform coordination. And precisely because it is a bureaucratic document — not a provocation, not a threat, not a weapon — it is the most important Chinese strategic document of 2026. It describes, with administrative precision, how Beijing intends to build the technological and industrial infrastructure that will determine the balance of economic and strategic power for the next decade.

The 450 platforms and 120 million IoT devices are not just economic targets. They are the physical substrate of a civil-military fusion architecture that extends China's intelligence reach into global supply chains. The AI agents are not just productivity tools. They are autonomous decision-makers embedded in the industrial processes of companies that will have limited practical ability to remove them once they are integrated. The coordinated overseas expansion is not just commercial ambition. It is a structured strategy for making Western decoupling economically irrational before it can be made politically mandatory.

### The window that is closing

The window for effective Western response to the MIIT plan is not infinite. Every pilot project that succeeds narrows it. Every Chinese SME that integrates onto a MIIT-designated platform and wins an international contract closes it slightly further. The plan is designed to operate faster than Western regulatory and industrial policy cycles — to create facts on the ground that policy cannot reverse. **Understanding this is the prerequisite to responding. And responding requires specificity, coordination, and urgency that Western governments have not yet consistently demonstrated.** The MIIT plan is not a mystery. It is a published document. The question is not whether the West can read it. The question is whether it will act at the speed that reading it requires.

*June 18, 2026. Seven ministries. One plan. The document is public, translated, cited. I read it and I think: this is the clearest statement China has ever made about its technological strategy. It is not hidden. It is not obscured. It is written in the administrative language of industrial policy — which makes it invisible to most political and media attention. That invisibility is a feature, not a bug. The most consequential strategic moves are often the ones that look like paperwork until they don't.*

**By Maxime Marquette, columnist**

## Columnist's transparency note

### Editorial stance

I am not a journalist but a **columnist and analyst**. My expertise lies in observing and analysing the geopolitical, economic and strategic dynamics that shape our world. My work is to take political strategies apart, to understand global economic movements, to put the decisions of international actors in context and to offer analytical perspectives on the changes that are redefining our societies.

I do not claim the cold objectivity of traditional journalism, which stops at reporting facts. **I claim analytical clarity**, rigorous interpretation and a deep understanding of the complex issues that concern us all. My role is to give meaning to the facts, to place them in their historical and strategic context, and to offer a critical reading of events.

### Method and sources

This text keeps the basic distinction between **verified facts and interpretive analysis**. The rule never changes: a factual claim is published only when a verifiable source supports it, and the sources this article actually used are listed under **Sources**, never here.

**Primary source families** used by the publication, when the story has them: official statements from governments and international institutions, public statements by political leaders, reports from intergovernmental organisations, dispatches from recognised international news agencies.

**Secondary source families**: specialised publications, internationally recognised news media, analyses from established research institutions, reports from industry organisations.

When an article cites statistical, economic or geopolitical data, it comes from the institutions that produce it (intergovernmental organisations, central banks, national statistics offices), and the exact institution is named under **Sources**.

### Nature of the analysis

The analyses, interpretations and perspectives in the analytical sections of this article are a **critical, contextual synthesis** based on the information available, the trends observed and the expert comments cited in the sources consulted.

My role is to interpret these facts, to set them in the context of today's geopolitical and economic dynamics, and to give them a coherent meaning within the larger story of the changes shaping our time. These analyses reflect expertise built through continuous observation of international affairs and an understanding of the strategic mechanisms that drive global actors.

This article describes a documented state of affairs on its publication date, not a prediction: later developments may shift its outlook. No update is promised in advance; when an article is corrected or completed, the change is dated in the text.

## Sources

### Primary sources

- [Newsquawk — China cross-department action plan for platform economy 2026-2028 — June 18, 2026](https://www.newsquawk.com/headlines/china-has-issued-a-cross-department-action-plan-for-promoting-the-collaborative-development-of-large-medium-and-small-enterprises-in-the-platform-economy-2026-2028-18-06-2026)

- [Chinese Government (english.www.gov.cn) — State Council framework for real-digital economy integration — January 13, 2026](https://english.www.gov.cn/news/202601/13/content_WS6966464cc6d00ca5f9a088c4.html)

- [Sinocism — Seven Departments Issue Action Plan to Promote Coordinated Development (analysis) — June 18, 2026](https://sinocism.notion.site/Seven-Departments-Issue-Action-Plan-to-Promote-Coordinated-Development-of-Large-Medium-and-Small-E-38384ece41d7810e931bd2d4bea0721c)

### Secondary sources

- [ORCA — China's industrial cyber ecosystem and civil-military fusion — March 31, 2026](https://orcasia.org/article/cicm-31st-march-2026)

- [South China Morning Post — Pentagon: 188 Chinese companies added to military-linked list including Alibaba, Baidu, BYD — June 23, 2026](https://www.scmp.com/news/china/military/article/3357065/quiet-escalation-unfolding-around-taiwans-remote-outposts-beijing-sends-ships)

- [CGTN — China unveils plan to boost integration of real and digital economies — June 19, 2026](https://news.cgtn.com/news/2026-06-19/China-unveils-plan-to-boost-integration-of-real-and-digital-economies-1O641kYUOqI/p.html)

- [The Asia Cable — Asia Daily June 22, 2026 (MIIT platform economy plan) — June 22, 2026](https://www.theasiacable.com/p/asia-daily-june-22-2026)

---

Source: Maxime Marquette, MadMax. When citing, name the author and link https://mad-max.co/en/article/analyse-le-plan-miit-2026-2028-l-arme-industrielle-de-pekin-contre-l-occident