CXMT Was the Headline. China Is Financing an Entire Semiconductor Supply Chain

That is the less important question.

The more consequential question is which missing part of the semiconductor supply chain China’s capital markets will finance next.

Yangtze Memory Technologies, or YMTC, has begun formal preparations for a possible domestic listing that could bring China’s leading NAND manufacturer to public investors. SJ Semiconductor completed an April 2026 IPO to finance advanced multi-chip integration and packaging projects. VeriSilicon, already listed in Shanghai, submitted a Hong Kong listing application as it seeks a broader capital base for semiconductor intellectual property and custom-chip design.

Those events are not at the same stage, and none should be treated as a repeat of CXMT’s completed offering.

Together, however, they reveal a wider strategy.

By April 2026, Shanghai’s STAR Market hosted 128 semiconductor companies spanning design, manufacturing, packaging and testing, equipment, and materials. They had raised more than RMB320 billion through IPOs and reported more than RMB360 billion in combined 2025 revenue.

CXMT was the headline. The larger development is the construction of a public-market financing system around most major layers of the semiconductor industry.

That does not mean China has achieved semiconductor independence.

It means that companies addressing most of the industry’s major layers now have access to domestic public equity.

Capital can finance fabrication plants, packaging lines, design teams, equipment suppliers, patent portfolios, and research programs.

It cannot automatically produce competitive yields, reliable equipment, globally qualified products, leading-edge manufacturing, or a complete AI hardware system.

China has built financial coverage across the semiconductor chain.

Its next test is whether that financial breadth can become technical depth.

YMTC could complete China’s publicly financed memory pair

CXMT manufactures dynamic random-access memory, or DRAM.

DRAM acts as working memory. It holds data temporarily while processors and accelerators perform calculations.

YMTC manufactures NAND flash.

NAND provides persistent storage. It retains data when power is removed and is used in solid-state drives, smartphones, consumer devices, enterprise storage, and data-center systems.

A modern AI infrastructure stack needs both.

Servers require DRAM to operate workloads. They require NAND-based storage to retain model files, datasets, checkpoints, applications, and system data.

That makes YMTC the closest strategic companion to CXMT.

In May 2026, YMTC entered formal IPO tutoring with CITIC Securities and CSC Financial. Tutoring is a meaningful preparatory step, but it is not the same as filing a full prospectus or receiving approval to list. No final offering size, share price, valuation, or trading date had been established in the public materials reviewed for this article.

The accurate description is therefore straightforward:

YMTC has begun formal preparations for a possible domestic IPO.

It has not completed one.

That distinction matters because the stages between preparation and trading can include regulatory review, prospectus submission, exchange questions, registration, pricing, allocation, delay, or withdrawal.

The story is not that YMTC has become the next CXMT.

It is that China’s second major memory manufacturer is preparing to use the same public-market system.

The Wuhan model resembles the Hefei model

CXMT’s development was closely associated with Hefei and Anhui-backed investment platforms.

YMTC’s development is similarly tied to Wuhan, Hubei, and national semiconductor investment funds.

Reporting surrounding the IPO preparation identified Hubei Changsheng Development as YMTC’s largest shareholder, with a 26.54% interest. The entity is associated with Wuhan’s Optics Valley administrative structure. Other investors reportedly include national integrated-circuit funds and additional state-backed capital.

The ownership structure points to a repeatable industrial-policy pattern:

1. Local and national public capital assumes early development risk.
2. The company builds manufacturing capacity over a long private period.
3. Domestic customers and suppliers create a commercial ecosystem.
4. The company approaches public markets for the next stage of expansion.
5. Public investors inherit both the opportunity and the remaining technical risk.

Hefei helped incubate a DRAM manufacturer.

Wuhan and Hubei-backed investors helped incubate a NAND manufacturer.

The public market becomes a handoff point from state-supported industrial development to wider ownership.

It is not a complete transfer away from state capital. State-backed funds can retain shares, participate in strategic placements, support suppliers, and continue financing infrastructure after listing.

Public listings broaden the financing base rather than replacing public capital entirely.

YMTC is not an early laboratory project

YMTC has already developed commercial 3D NAND products.

Its official product materials identify the X4-9070, a fifth-generation TLC 3D NAND product based on the company’s Xtacking 4.0 architecture. YMTC markets the product for enterprise, embedded, consumer, cloud-computing, and big-data applications.

YMTC says Xtacking manufactures the NAND memory array and peripheral logic separately before connecting them. That approach allows each portion to follow a different development and manufacturing path.

The architecture is strategically important because it gives YMTC a distinct route for improving interface performance and scaling products under difficult manufacturing constraints.

YMTC has also moved beyond raw NAND dies into selected finished storage products.

Its PCIe 5.0 PC550 solid-state drive has been advertised in capacities up to 2 TB and with sequential read speeds reaching 10,500 MB per second.

That means YMTC now participates in more than one storage layer:

• NAND fabrication
• Memory packaging
• Selected controller and firmware integration
• Branded client storage products

It would still be inaccurate to describe YMTC as a complete enterprise-storage provider.

Enterprise systems require additional capabilities, including controllers, firmware, endurance qualification, system software, error management, service, and long-term customer support.

YMTC produces a foundational storage technology.

It does not represent the entire storage stack.

YMTC’s market position remains difficult to reconcile

Published estimates of YMTC’s global NAND position vary materially.

Reporting surrounding its 2026 IPO preparation placed first-quarter revenue above RMB20 billion and its global NAND share above 10%. Other industry analysis from portions of 2025 placed the share below 5%, attributing slower expansion to equipment restrictions and production constraints.

Those figures may not measure the same thing.

They may differ by:

• Revenue
• Bit shipments
• Wafer output
• Quarterly timing
• Product mix
• Branded sales
• Production capacity

Unlike CXMT, YMTC had not published a public IPO prospectus that allowed investors to reconcile audited financials, capacity, customer concentration, and market-share methodology.

The article should therefore not choose the highest estimate and present it as settled fact.

A future prospectus would matter for more than the capital raise. It could provide the first comprehensive public test of YMTC’s actual scale, profitability, cash flow, customer base, equipment exposure, and planned use of proceeds.

YMTC would enter the market under direct U.S. restrictions

YMTC’s geopolitical position differs from CXMT’s.

The U.S. Commerce Department added Yangtze Memory Technologies and identified affiliates to the Entity List in 2022, creating licensing requirements for covered exports, reexports, and transfers.

The practical effect depends on the listed legal entity, the controlled item, whether the item is subject to the Export Administration Regulations, foreign direct-product rules, end use, and licensing policy.

YMTC can nevertheless face significant restrictions involving:

• Semiconductor-manufacturing equipment
• Replacement parts
• Process software
• Technical support
• Tool installation
• Calibration
• Maintenance
• Controlled components

This creates one of the companion article’s central questions:

Can domestic public capital compensate for restricted access to foreign production technology?

Money can help YMTC purchase domestic equipment, redesign processes, stock spare parts, recruit engineers, and support local suppliers.

Money cannot instantly give a domestic toolmaker the throughput, defect control, reliability, service network, and accumulated manufacturing knowledge of a mature international supplier.

An eventual YMTC IPO would therefore finance more than NAND expansion.

It would finance a real-time experiment in semiconductor substitution under export controls.

Advanced packaging is the next bottleneck

Memory production alone does not create a high-performance AI system.

Modern AI hardware combines multiple components:

• Compute dies
• Memory
• Interposers
• I/O dies
• Substrates
• Chiplets
• Power-delivery components
• Thermal systems

Those components must be connected at high bandwidth while controlling power consumption, heat, signal loss, and manufacturing defects.

That is the role of advanced packaging.

SJ Semiconductor represents this layer of the financing system.

The company listed on Shanghai’s STAR Market on April 21, 2026, under code 688820. The exchange describes it as a provider of 12-inch mid-end wafer processing, wafer-level packaging, and chiplet multi-chip integration for high-performance products, including GPUs, CPUs, and AI chips.

SJ Semiconductor issued approximately 255.47 million shares at RMB19.68 each, producing gross proceeds of about RMB5.03 billion.

The online portion of the offering was initially subscribed more than 4,100 times, while strategic investors received a substantial allocation.

The transaction was far smaller than CXMT’s listing.

Its market mechanics were similar:

• Heavy demand
• Strategic allocations
• Restricted shares
• A narrow immediately tradable portion
• Capital directed toward a national technology bottleneck

SJ Semiconductor is financing physical integration capacity

The company’s intended use of proceeds was more narrowly defined than CXMT’s broad DRAM expansion plan.

SJ Semiconductor identified projects involving:

• Three-dimensional multi-chip integration
• Ultra-high-density interconnection
• Chiplet packaging
• Bumping capacity
• 3D integrated-circuit research and industrialization

The distinction is important.

CXMT and YMTC manufacture memory.

SJ Semiconductor addresses the technical layer that allows memory, processors, and other dies to function together in one high-performance package.

Its pre-IPO financial growth suggests that this was not a speculative research concept.

The Shanghai Stock Exchange reported that revenue increased from RMB1.633 billion in 2022 to RMB4.705 billion in 2024. The company reported RMB435 million in parent-attributable profit during the first half of 2025, while cumulative R&D spending from 2022 through the first half of 2025 exceeded RMB1.5 billion.

Within months of listing, SJ Semiconductor announced steps toward beginning construction of the first phase of a 3D integrated-chip project.

That sequence provides one of the clearest examples of the financing system working as intended:

1. A company develops an existing commercial platform.
2. It raises public equity.
3. The proceeds are assigned to defined capacity and research projects.
4. A construction program begins moving forward.

The new project was not yet complete or producing commercial output.

But public capital had begun moving from a listing into physical semiconductor infrastructure.

Packaging can become a bottleneck even when chips exist

A country may have access to processor designs, memory dies, substrates, and packaging companies while still lacking sufficient ability to assemble those elements at leading performance and volume.

Advanced packaging requires precision across:

• Alignment
• Bonding
• Thermal design
• Interconnection density
• Testing
• Materials
• Yield management
• Reliability

A defect in one component can reduce the value of an entire multi-die package.

That makes packaging economics especially difficult.

More capital can build additional lines.

It cannot guarantee acceptable yields or customer qualification.

SJ Semiconductor is therefore a useful test of the larger thesis.

China’s public markets can finance the packaging bottleneck.

The company must still prove that the added capacity can meet frontier requirements at scale.

VeriSilicon represents the design layer

VeriSilicon occupies a different part of the semiconductor system.

It does not operate like CXMT or YMTC.

It provides semiconductor intellectual property and custom-chip design services.

Its business includes processor technology, interface IP, platform-based silicon services, and support for customers developing application-specific chips.

In April 2026, VeriSilicon submitted an H-share listing application in Hong Kong while retaining its existing Shanghai STAR Market listing.

Major offering details remained redacted, including the number of shares, pricing, proceeds, and timetable. The Hong Kong exchange also warned that publication of the application materials did not guarantee that an offering would occur.

The correct description is therefore:

VeriSilicon applied for a Hong Kong listing. It had not completed the offering.

That procedural distinction reveals another part of the financing strategy.

Shanghai and Hong Kong serve different capital roles

The STAR Market gives Chinese semiconductor companies access to:

• Domestic institutional investors
• Retail demand
• State-backed funds
• Strategic placements
• Renminbi financing
• Technology-sector indices
• Exchange-traded funds

A Hong Kong listing can potentially add:

• Offshore institutional capital
• Greater international visibility
• Foreign-currency financing
• A second valuation venue
• Broader acquisition currency

VeriSilicon is part of a broader movement by mainland-listed semiconductor companies to seek additional access to Hong Kong capital.

China’s semiconductor companies are no longer being financed through one market alone.

They are building a layered capital structure around the industry.

VeriSilicon matters because fabrication is only one part of the problem

A semiconductor system needs more than factories.

It needs the intellectual property and engineering required to create chips worth manufacturing.

Custom-chip development can involve:

• Processor architectures
• Neural-processing units
• Graphics technology
• Video and vision systems
• Memory interfaces
• Physical implementation
• Verification
• Software support

As AI workloads become more specialized, companies increasingly develop application-specific integrated circuits rather than relying entirely on general-purpose processors.

VeriSilicon says AI-related custom-chip demand has become an increasingly important part of its order book.

That makes it relevant to the AI infrastructure cycle even though it does not operate a memory fab or packaging plant.

CXMT raises capital to manufacture memory.

SJ Semiconductor raises capital to integrate dies.

VeriSilicon seeks a broader capital base for the design and intellectual-property layer.

A complete semiconductor strategy requires all three.

The STAR Market is functioning as industrial infrastructure

The STAR Market is often described as a stock exchange for technology companies.

For China’s semiconductor sector, that description understates its practical role.

The exchange hosts companies across design, manufacturing, packaging and testing, equipment, and materials. Those companies have raised hundreds of billions of renminbi through IPOs and use public shares for expansion, research, employee ownership, and acquisitions.

In practice, the STAR Market is functioning as financial infrastructure for China’s semiconductor strategy.

It supports:

• Fabrication plants
• Packaging facilities
• Equipment production
• Materials qualification
• Chip-design teams
• Research programs
• Employee ownership
• Acquisitions
• Supplier development

It also creates valuations and securities that can be used for future fundraising.

A semiconductor company can list, enter an index, attract ETF demand, raise follow-on capital, and use shares in acquisitions.

The capital market becomes part of the production system.

The system now covers most boxes in the diagram

A simplified map looks like this:

These are representative examples only. A listed company’s presence in a category does not establish full domestic substitution or frontier-level competitiveness.

The map is financially significant.

It is not proof that every layer can yet perform at the level required by the most advanced AI systems.

Strategic placements coordinate the supply chain

CXMT’s IPO demonstrated how strategic investors can align customers, suppliers, insurers, public funds, and equipment companies around one issuer.

SJ Semiconductor’s offering used a similar structure on a smaller scale.

Strategic placements can help create long-term relationships among:

• Equipment suppliers
• Materials companies
• Packaging providers
• Cloud companies
• Device manufacturers
• Vehicle companies
• Institutional investors

Strategic investment can signal industrial alignment, but it should not be treated as proof of a commercial contract.

A supplier may invest because it expects future business.

A customer may invest because it wants supply security.

A state fund may invest because the company fills a strategic gap.

An insurer may invest because a restricted float and national priority create valuation potential.

Those motives are not identical to a conventional investor assessing discounted cash flow.

That does not make them illegitimate.

It means valuations should be interpreted within the structure that created them.

Index and ETF demand can create a second financing loop

Once semiconductor companies enter eligible indices, passive and index-linked funds can become another source of demand.

The potential loop works like this:

1. A semiconductor company lists.
2. Its market capitalization increases.
3. It becomes eligible for one or more technology indices.
4. ETFs and passive funds purchase shares.
5. Higher valuations improve future financing capacity.
6. The company raises more capital or uses shares for acquisitions.

This can support long-term industrial expansion.

It can also magnify valuation cycles.

The same market structure that channels capital toward strategic technology can push prices far beyond near-term earnings.

CXMT’s first-day debut made that risk visible.

It is unlikely to be unique to CXMT.

Financial coverage is not technical independence

China now has publicly traded companies in most semiconductor categories.

That does not mean every category is globally competitive or independent of foreign technology.

The unresolved bottlenecks include:

• Advanced lithography
• High-end metrology
• Electronic-design automation
• Process-control software
• Certain deposition and etching applications
• High-bandwidth memory commercialization
• Advanced substrates
• Leading-edge foundry access
• Packaging yield
• Equipment maintenance and spare parts
• International customer qualification

This leads to the article’s central distinction:

China has built financial coverage across the semiconductor supply chain. It has not yet demonstrated complete technical independence across that chain.

A stock exchange can finance an equipment company.

It cannot guarantee that the tool matches the throughput, precision, defect control, and reliability of an international competitor.

A public offering can finance a packaging line.

It cannot guarantee acceptable yields for a complicated multi-die package.

An IPO can fund a memory fab.

It cannot guarantee that the company will remain profitable through the next oversupply cycle.

Capital creates opportunity.

Execution determines independence.

State-backed capital creates resilience and risk

The state-to-market model has clear advantages.

It can support projects that require:

• Decade-long development periods
• Heavy upfront capital
• Years of losses
• Large research teams
• Domestic equipment qualification
• Low initial yields
• Strategic patience

Many semiconductor projects would not survive if judged only by short-term private returns.

State-backed capital gives them time.

Public markets then provide a larger pool of follow-on financing.

The same model can create allocation problems.

Local governments may compete to build similar projects.

Funds may support companies because they fill a policy category rather than because they have a sustainable commercial advantage.

High valuations may weaken investment discipline.

Public shareholders may inherit projects developed under strategic mandates they did not design.

The system can finance:

• Too much capacity
• Duplicated capacity
• Technically immature capacity
• Capacity without enough customers
• Capacity dependent on continuing support

The relevant question is not whether the capital is public or private.

It is whether the investment produces durable technical and commercial capability.

Memory companies still face the memory cycle

CXMT and YMTC are approaching public markets during a strong period for AI-related memory demand.

That is favorable for fundraising.

It is also dangerous.

The memory cycle often follows a familiar sequence:

1. Prices rise.
2. Profits improve.
3. Manufacturers expand.
4. New capital enters.
5. Supply increases.
6. Customer inventories normalize.
7. Prices fall.
8. Margins contract.

A public listing does not remove this cycle.

It may accelerate it by giving manufacturers more capital during the strongest part of the market.

Public-market financing could therefore accelerate capacity additions and increase the risk of a future oversupply cycle.

That risk applies to established international producers as well.

China’s expansion makes it more consequential.

Restricted floats can magnify valuations

CXMT’s first-day market capitalization was established through a relatively narrow immediately tradable portion of the company.

SJ Semiconductor also began trading with large portions of the broader ownership base restricted.

This pattern matters.

When a strategically important company has:

• Heavy oversubscription
• Strategic-investor lockups
• Pre-IPO shareholder lockups
• Limited public float
• Index expectations
• National-policy significance

a relatively small amount of tradable stock can establish the headline value of the entire company.

The market capitalization is calculated according to standard conventions.

It does not mean every shareholder could sell at that price.

A high valuation can help a company raise money.

It can also create unrealistic expectations and future volatility when more shares become tradable.

What capital can build

China’s semiconductor financing system can produce real industrial outcomes.

It can build:

• Memory fabs
• Packaging plants
• Research laboratories
• Equipment factories
• Materials capacity
• Design teams
• Patent portfolios
• Domestic supplier networks

It can help companies survive development cycles that would defeat shorter-term investors.

It can create customers and suppliers willing to coordinate through strategic investment.

It can spread risk across local governments, national funds, institutions, and public shareholders.

That is why the system matters.

What capital cannot guarantee

Capital cannot guarantee:

• Competitive yield
• Low defect rates
• Frontier process technology
• HBM leadership
• Reliable domestic equipment
• A complete software ecosystem
• International customer trust
• Sustainable pricing
• Rational capacity decisions
• Long-term shareholder returns

Money can buy time, tools, talent, and facilities.

It cannot buy accumulated manufacturing experience instantly.

That gap will determine whether the public-market strategy succeeds.

CXMT was the headline, not the finish line

CXMT’s debut made China’s semiconductor financing system visible because of its scale.

YMTC shows that the same system may soon be used for NAND flash.

SJ Semiconductor shows that public equity is already being converted into advanced-packaging projects.

VeriSilicon shows that Shanghai-listed chip companies are seeking additional access to Hong Kong capital for the design and intellectual-property layer.

The STAR Market connects these cases.

It has become more than a venue where successful companies list after they mature.

It is becoming part of the machinery through which China develops semiconductor capacity.

That is a real strategic advantage.

It is not proof of technical self-sufficiency.

China has built a public market capable of financing almost every box on a semiconductor supply-chain diagram.

The next test is whether those boxes can perform together as one competitive system.

Sources