In late September, Shenzhen Aowei Lingxin Technology Co., Ltd. (hereinafter referred to as "Aowei Lingxin") completed the full closing of its Pre-A+ round, with the amount exceeding CNY 200 million. Combined with the Pre-A round previously led by Longgang Financial Holdings together with its sub-fund Chuangdongfang Investment, the total scale of the company's Pre-A and Pre-A+ financing rounds has exceeded CNY 300 million.
Let's look at the capital structure: Su Venture Capital, Guoxin Guozheng, Puhua SME Phase II, and Chengdu Peikun have state-owned or central enterprise backgrounds; Xingchen Technology, Shenzhen Industrial Investment, and Amala Yuehai are industrial capital; well-known investor Wei Wei participated personally, and existing shareholders continued to follow on. From the angel round led by Shenzhen Capital Group to the investor composition in this round, we can see the capital market's attitude towards the company's technology route and engineering capabilities.
The entry of state-owned capital means the project is viewed within the framework of "autonomous and controllable in key information fields"; the entry of chip design companies like Xingchen Technology as industrial capital often brings downstream scenarios and supply chain resources, rather than being purely financial investments; the multiple rounds of follow-on investments by existing shareholders indicate that early investors recognize the pace of technological advancement rather than simply chasing hotspots.
Capital is moving away from early-stage low-power IoT and embedded systems, shifting towards more expensive segments such as server CPUs, high-performance cores, industrial DSPs, and heterogeneous computing: in July, Lingrui Zhixin Computing Technology completed a Pre-A round of hundreds of millions of CNY; in August, Lanxin Computing completed a Pre-B round; and in September, Zhongke Haoxin completed a B round. Aowei Lingxin is taking the "heterogeneous + industry-specific" route among them.
| Modular Heterogeneity Combining RISC-V and DSA
Aowei Lingxin's approach is based on RISC-V to build a modular and configurable heterogeneous architecture. Specifically, it uses RISC-V general-purpose cores paired with DSA (Domain-Specific Accelerators), and attaches near-memory computing units on demand to alleviate data throughput bottlenecks.
The focus of this architecture is not on stacking cores, but on configurability—the same silicon die adapts to different scenarios through different configurations, replacing the high-cost model of "one chip for all" with an engineering model of "one architecture, tailored for multiple scenarios."
Compared to companies that only sell RISC-V processor IP licenses, Aowei Lingxin has chosen a heavier path: directly delivering "processor chips plus supporting software and hardware solutions" to customers in communications, storage, industrial equipment, and edge computing. The path is heavier, but customer stickiness and barriers are also higher.
The company is also equipped with a set of self-developed AI design tools to improve efficiency and reduce costs. At a time when the marginal cost of chip design continues to rise, the self-development of the toolchain itself constitutes a barrier.
| RV9000: From Blueprint to Silicon
In late August, the company's self-developed RV9000 series heterogeneous chip completed tape-out and was successfully brought up. The chip adopts a RISC-V plus DSA many-core heterogeneous design, with heterogeneous computing power combinations up to 24 cores, built-in national cryptographic security engine, and equipped with rich high-speed I/O interfaces.
The so-called "24 cores" does not mean 24 general-purpose CPU cores, but a combination of general-purpose cores and software-defined accelerator cores. Public information shows that the RV9000 offers two general-purpose core configurations of 4 cores and 8 cores, and some models can additionally attach 16 accelerator cores. Product differentiation precisely comes from the flexible combination of "general-purpose processing plus specific business acceleration."
The physical value of the RV9000 has two points: first, it verifies that the "modular configurable heterogeneous architecture" is feasible in engineering; second, it confirms the technical path for AI inference chips. Its positioning spans from edge computing to data center infrastructure upgrades, and it uses the national cryptographic security engine to cut into key information fields with hard constraints on autonomous and controllable requirements.
However, tape-out and bring-up are only the first hurdle. After entering the physical silicon validation, it must pass complete functional debugging, performance and reliability verification, and SDK and operating system adaptation, and then deliver samples to Alpha customers for evaluation. There are still several steps before large-scale commercialization. In this regard, Aowei Lingxin is no different from typical RISC-V startups.
| A Dedicated Niche for Private AI Inference
The most noteworthy aspect of Aowei Lingxin is that it has chosen private deployment of AI inference as its landing point.
The training side is dominated by a few giants and high-computing-power GPUs. The inference side, especially the private deployment in government, enterprises, and key industries, is the real battlefield where domestic architectures can truly land: data must not leave the domain, computing power must be controllable, and costs must be affordable. The openness and configurability of RISC-V perfectly fit the needs of "industry-specific CPUs"—there is no need to compete for the peak of general-purpose computing power, but rather to optimize energy efficiency, security, and cost for specific workloads.
Centered around the RV9000, the application landing points disclosed by the company include network communications, network security, edge computing, industrial control, distributed storage, as well as industrial automation equipment, industrial PCs, intelligent robots, communication equipment, AI PCs, and power equipment. The commonality of these scenarios is that they have defined scenarios, strong demands for domestic substitution, and high sensitivity to security and compliance, which is exactly where the RISC-V heterogeneous architecture can play its role.
| Differentiation in the RISC-V Track
Placing Aowei Lingxin in the current RISC-V capital landscape, the differences are clear:
Lingrui Zhixin Computing Technology focuses on processor cores and chiplets; Lanxin Computing bets on server CPUs; Zhongke Haoxin focuses on industrial DSPs; ESWIN Computing targets smart terminals and embodied AI, and is advancing its Hong Kong stock IPO; Aowei Lingxin uses a "general-purpose core plus dedicated accelerator core" heterogeneous architecture to cut into network communications, distributed storage, and edge computing.
This track is changing its valuation yardstick. After ESWIN Computing entered the IPO review, the market's evaluation of RISC-V enterprises is shifting from "whether tape-out is done, how many cores, how flashy the architecture is" to customer validation, software ecosystem, mass production progress, revenue scale, and going-concern ability.
| Beyond the Hype: Three Things to Watch
First, orders are more important than tape-out. Bring-up is a technical milestone, but whether it can secure clear orders and achieve batch delivery is the real problem in the next stage.
Second, ecosystem is more important than architecture. The victory or defeat of RISC-V lies in the software ecosystem and developer stickiness; the progress of toolchain and operating system adaptation is more critical than the number of cores.
Third, cash flow is more important than valuation. In an environment where capital is retreating and IPO reviews are becoming stricter, the pace of cash generation under continuous losses directly determines whether the financing window can be sustained.
| Conclusion
The closing of Aowei Lingxin's Pre-A+ round of over CNY 200 million, combined with the overall Pre-A round of over CNY 300 million, is another example of RISC-V moving from an embedded narrative to a "high performance plus vertical industry" narrative. The modular heterogeneous route of RISC-V plus DSA, coupled with the physical landing of the RV9000 and the national cryptographic security engine, targets two high-value main lines: "autonomous and controllable in key information fields" and "private AI inference". Being awarded the Rising Star of the 7th China IC Unicorn is the intersection of dual recognition from both industry and capital.
However, tape-out is just the beginning; customer validation and large-scale delivery are the real answers. Whether Aowei Lingxin can turn its architectural advantages into a revenue curve is worth continuous observation over the next 12 to 18 months.
This article is an industry observation and does not constitute any investment advice.