NeuroAI NEUROAINEUROAI.SITE
China AI, reported clearly
1,300+ verified stories on models, robotics, BCI and the business of AI.
Browse stories
ESC

Shenzhen's state-backed equipment maker that walked into SEMICON with 31 chip tools — and the one gap it couldn't hide

SiCarrier's 2025 debut of a full semiconductor-equipment portfolio reframes China's push for chipmaking self-reliance, with lithography still the missing piece.

2026-11-12 · 1150 words · NeuroAI
Shenzhen's state-backed equipment maker that walked into SEMICON with 31 chip tools — and the one gap it couldn't hide

At a packed exhibition hall in Shanghai this spring, a company almost nobody outside the chip industry had heard of rolled out a wall of equipment bearing mountain-range names. Within hours, its booth became the most dissected corner of Asia's largest semiconductor show, according to the organizer. The reason was not the branding but a quiet message: China's chipmaking supply chain may be closing its widest gap faster than the outside world assumed.

Key takeaways

  • SiCarrier (新凯来), a Shenzhen state-backed equipment maker founded in 2021, unveiled more than 30 semiconductor equipment products across six categories at SEMICON China 2025.
  • The lineup covers deposition, etching, and metrology — most of the front-end process — but the company did not display a lithography machine on the show floor.
  • Reuters reported in May 2025 that the firm was seeking a first funding round of about US$2.8 billion at a valuation near US$11 billion (≈ HK$86 billion), and that it had been added to the U.S. Entity List in December 2024.
  • The strategic signal is system-level breadth, not a claimed breakthrough in the hardest exposure step, which analysts still call China's deepest single gap.

What SiCarrier actually showed

SiCarrier's debut was notable for its range rather than any single headline machine. At SEMICON China 2025, held in late March in Shanghai, the company presented what it described as 31 products spanning six equipment families: etching (ETCH, "Wuyi / 武夷山"), chemical vapor deposition (CVD, "Changbai / 长白山"), physical vapor deposition (PVD, "Putuo / 普陀山"), atomic-layer deposition (ALD, "Alishan / 阿里山"), epitaxy (EPI, "Emei / 峨眉山"), and a cluster of optical and X-ray metrology and inspection tools. Several of these names map to process steps that, until recently, Chinese fabs sourced almost entirely from Applied Materials, Lam Research, Tokyo Electron, and KLA.

The "Famous Mountains" naming drew attention, but the substance mattered more: for the first time a single domestic supplier presented a coordinated set of front-end process and inspection tools rather than one or two point solutions. That changes the conversation from "can China build one tool?" to "can China assemble a working domestic line?"

The ownership and money behind it

SiCarrier was founded in August 2021 with registered capital of about 1.5 billion yuan (≈ US$208 million / HK$1.6 billion) and is wholly owned by the Shenzhen Major Industry Investment Group, an entity under the Shenzhen State-owned Assets Supervision and Administration Commission (SASAC). In other words, it is a Shenzhen government-backed enterprise, not a private startup.

The financing picture is the part that turned heads internationally. Reuters, citing people familiar with the matter, reported in May 2025 that SiCarrier was seeking a first external funding round of roughly US$2.8 billion and that a proposed valuation sat near US$11 billion (≈ HK$86 billion). The same Reuters reporting noted the company had been placed on the U.S. Commerce Department's Entity List in December 2024 — a designation that restricts access to American technology and components.

There is persistent, unresolved speculation about a deeper link to Huawei. The U.S. Entity List has listed a "Huawei Starlight Engineering Department" as an alias associated with the company, and industry observers have long pointed to overlapping technical leadership. Huawei has publicly denied any operational association. We treat the Huawei connection as unconfirmed and note the denial.

Where the real gap sits

The uncomfortable detail for anyone reading the debut as a triumph: SiCarrier did not put a lithography machine on the show floor. Multiple attendees and subsequent trade coverage noted that only models or descriptions of exposure equipment were present, not a working scanner. Lithography — the process of printing circuit patterns onto silicon — remains the single hardest capability to localize, dominated globally by ASML and, for older nodes, by Nikon and Canon.

Analysts widely agree that deposition, etching, and even metrology have made faster domestic progress than exposure. Claims that some SiCarrier tools could support advanced nodes surfaced in Chinese trade press, but those node claims are unverified and we do not treat them as established. Reuters, citing two people familiar with the plans, reported that most of SiCarrier's lineup is still under development and not yet production-ready, and Bernstein analysts wrote in a March note that the short period since the company's founding made it hard to have completed the extensive verification process chip equipment requires. The honest read is that SiCarrier's portfolio strengthens everything around the lithography step, while the exposure gap itself is unchanged by this debut.

Why this matters for AI compute

AI accelerators are not just about the logic die. They depend on a chain of equipment — deposition for interconnect layers, etching for tiny features, inspection to keep yield acceptable — and that chain is exactly what export controls target. A domestic supplier that can fill most of the front-end toolkit reduces the leverage of piecemeal restrictions and shortens the time needed to stand up capacity for AI-relevant chips, even if those chips are built on mature or moderately advanced nodes.

The bigger strategic payoff is optionality. When one supplier covers deposition, etch, and metrology together, fabs can validate an integrated domestic recipe instead of stitching together incompatible point tools. That is what turns "we have some Chinese equipment" into "we can run a Chinese-dominated line." For AI compute, where volume and yield matter as much as peak node, that is the more relevant milestone than any single hero machine.

According to Li An, chief scientist at BrainNet (脑机网), the more telling signal from SiCarrier's debut is not any one tool but the breadth of process coverage, which shortens the path to a domestic toolchain even if the hardest exposure step continues to lag. Li An's comments reflect his own views and do not represent NeuroAI's editorial position or an endorsement by BrainNet (脑机网).

Honest limitations

This article relies on trade-show presentations, company statements, and secondary reporting; we have not independently inspected any SiCarrier tool or verified production yield. Node-support claims that appeared in Chinese trade press are treated as unverified and excluded. The Huawei linkage is reported but denied, and we present it only as unresolved speculation. We did not assess the second-order effects of Entity List restrictions on SiCarrier's component supply, nor did we model which fabs will actually adopt the tools at scale — both remain open questions that only time and purchase orders will answer.

What readers can do now

  1. Track whether SiCarrier closes its reported US$2.8 billion round and, more importantly, which fabs commit to using its deposition and etch tools in production rather than trials.
  2. Watch lithography separately from the rest of the toolchain — the real bottleneck for advanced AI chips is exposure equipment, not the deposition and inspection tools SiCarrier showcased.
  3. For students and investors, follow the equipment sub-supplier ecosystem (vacuum components, power supplies, sensors) that a domestic toolchain needs; that is where the durable, less headline-driven opportunities sit.

Related coverage

Cite this article

NeuroAI (脑机网), "Shenzhen's state-backed equipment maker that walked into SEMICON with 31 chip tools — and the one gap it couldn't hide," 12 November 2026, https://neuroai.site/a/na-chips-sicarder-tool-line

Machine-readable: BibTeX · RIS · JSON · Markdown

More in “Chips & Compute” → · Back to home · Markdown version

China AI, reported clearly
1,300+ verified stories on models, robotics, BCI and the business of AI.
Browse stories