Company Profile
Shenzhen Longsight Technologies Co., Ltd. (hereinafter referred to as "Longsight") specializes in the R&D, manufacturing, sales, and service of high-end electronic measurement instruments. Serving sectors such as ICT, the Internet, smart terminals, intelligent vehicles, advanced manufacturing, metrology and testing, and educational research, the company provides test products, solutions, and services. Its main product lines cover ultra-high-speed real-time/sampling oscilloscopes, vector signal generators, signal and spectrum analyzers, vector network analyzers, optical modulation analyzers, and high-speed digital network test platforms, providing testing support for scenarios including high-speed optical interconnects, AI computing power networks, advanced semiconductors, and commercial spaceflight.
Editor's Note
As AI computing power continues to rise and the iteration of optical modules and optical engines accelerates, a frequently overlooked issue has become prominent: who verifies these devices? High-end oscilloscopes and sampling oscilloscopes have long relied on imports. What domestic instruments lack is not just a single piece of equipment, but an entire chain composed of materials, processes, algorithms, standards, and customer trust.
This episode of "Chip Unveiled: Experts Discuss Chips" features an interview with Ms. Cui Tianyu, General Manager of Product Lines at Longsight, to dissect the breakthrough logic of domestic high-end measurement instruments. In its three years since establishment, Longsight's product line has gradually covered test scenarios ranging from chips and modules to networks. Its self-developed 90GHz real-time oscilloscope is hailed as China's first breakthrough; its 65GHz sampling oscilloscope can even process 4 channels in parallel and deliver results in 4 seconds. Against the backdrop of AI implementation, advancement of cutting-edge processes, and acceleration of high-speed optical interconnects, high-end test and measurement instruments are becoming the critical "rulers" that determine whether the industrial chain can "run faster, measure more accurately, and see more clearly." So, what makes building this ruler so difficult? How can it be truly adopted by customers? This episode reveals the answers.
*The following is the transcribed text of the podcast, confirmed and authorized by the guest.
What Makes High-End Measurement Instruments Difficult?
Huanshi: I believe many people have heard of Longsight, especially given how grand your name sounds.
Cui Tianyu: Yes, it is actually a very simple name.
Huanshi: The name isn't simple; I think your leadership had quite some vision when choosing it.
Cui Tianyu: We originally had a platform called "Qianliyan" (Thousand-Mile Eye). However, for test and measurement instruments, the precision typically needs to be one order of magnitude higher than the object being tested. Therefore, we upgraded from "Thousand" to "Ten Thousand," naming it "Wanliyan" (Ten-Thousand-Mile Eye). Du Fu once wrote a poem stating—"Eyes spanning ten thousand miles across heaven and earth, a heart focused on the timeline of a hundred years." These few words condensed the iteration of time, the precision and timing required for test and measurement instruments, and a sense of national responsibility all into one. Thus, we finalized the name Longsight.
Huanshi: Today, during my conversation with General Manager Cui, I learned that there is a poem by Du Fu behind the name. Everyone assumed upgrading from "Thousand-Mile Eye" to "Ten-Thousand-Mile Eye" was just a matter of increasing the order of magnitude, unaware that the story behind it was so poetic.
You mentioned test and measurement instruments. Why did Longsight target this specific track? There are many startups in China. Did you choose this path because you saw a special market demand, or did you have a unique perspective?
Cui Tianyu: Many decisions are driven by the call of the era. As a major manufacturing country, China has a mature supply chain for basic instruments, and numerous test and measurement companies have emerged as excellent players. However, as demands upgrade, we find that while everyone can easily buy small oscilloscopes or multimeters—even pocket-sized ones—imports of high-end instruments face restrictions, and domestic substitution in the high-end sector remains a blank space.
In other words, as we advance cutting-edge processes, AI, and high-speed optical interconnect industries, we look back and realize that the high-end instruments needed for chip and device R&D and verification still rely on imports. Once supply is restricted, the entire industrial chain gets stuck. Based on this context, Longsight chose to enter the high-end test and measurement instrument market.
Huanshi: Everyone knows there are many practitioners in the traditional instrument industry. How do you define the high-end instruments you build? What constitutes "high-end"? Is it based on application scenarios or performance metrics?
Cui Tianyu: "High-end" is defined by comparison and is a result of generational shifts in demand. Five or eight years ago, an oscilloscope with 1GHz bandwidth was considered high-end; today, only those with 33GHz, 59GHz, or even 90GHz bandwidth are viewed as high-end.
Take another example: early switch backplanes had speeds of around 10Gbit/s, which we considered "high-speed" at the time. But as technology evolved, the threshold for high-speed has moved to 112Gbit/s or even 224Gbit/s. So-called "high-end" means whether the product can match the testing requirements of high-speed interconnect hardware corresponding to the current era.
Huanshi: Indeed, different eras and scenarios bring different demands. Domestic test and measurement suppliers may still focus on relatively basic products, meaning entering the high-end instrument track inevitably requires greater investment in manpower and financial resources. I want to delve deeper: what exactly makes high-end instruments difficult? Is it merely a difference in numbers? From your introduction, people have gained an intuitive understanding. But looking at development investment, how significant is the change in barriers? Can traditional companies achieve this simply through multiple rounds of financing and sustained investment?
Cui Tianyu: That is a good question. Discussing "what makes it difficult" in this era, it is not solved by a single chip, nor by AI writing some code. It is difficult due to the entire industrial chain. Some Western companies have been operating for seventy, eighty, or even over a hundred years, accumulating deep expertise in processes, materials, manufacturing nodes, protocol standards, and vast amounts of customer data. They are already one step ahead, or even several steps ahead.
Huanshi: So, Western companies have established a comprehensive ecosystem.
Cui Tianyu: Yes. Therefore, as we push forward, we need to break through fundamental material and process support at the technical level, involving massive software algorithms and various challenges. Additionally, there is the ecosystem issue you mentioned. Customers have been accustomed to using a certain brand for years or even decades. All their company data is measured using that brand, and their production line accumulation relies on it. When a new company appears, why should customers trust you? This is a problem of ecosystem and standards, representing a harder-to-break path dependency.
Where Does the Confidence to Build High-End Products Come From?
Huanshi: I would like to ask General Manager Cui, why does Longsight dare to build high-end instruments? What is the underlying logic supporting this?
Cui Tianyu: Conventional product R&D often starts with addition, iterating from 1 to 2, then to 3. But if we set our initial goal to 100, the product design logic and technical difficulties are entirely different levels.
For example, with oscilloscopes, going from 1GHz to 10GHz bandwidth likely allows us to first solve low-end link chip and component issues, create a prototype, and then iterate upwards. However, when targeting high-end products, we find too many missing links. At this point, we often need to deconstruct from a system level. Moreover, this deconstruction must land precisely on positions within the industrial chain that are "just right" or "reachable with a jump," otherwise, related technologies might remain unachievable for three to five years.
Therefore, solution design itself is a process requiring wisdom in deconstruction, akin to converting calculus problems into addition and subtraction problems, making each individual technical point easier to solve. Taking ultra-high-bandwidth real-time oscilloscopes as an example, the system is deconstructed into multiple chips. While a single chip might be a difficulty, a system-level splicing solution ultimately forms a higher-end device. This is the first key technical point.
Second, we must drive the entire industrial chain to form a cyclical upward progression. When we invite a manufacturer to co-develop a component for a world-leading product, they often say: "I don't have the corresponding technology now. Even if I make a sample, I have no equipment to measure it." This creates a contradiction: building higher-end products requires high-end equipment to prove them; building equipment requires high-end components. Therefore, we must form a cyclical, spiral ascent with the entire industrial chain.
Third, is using algorithms to supplement chips and software to supplement hardware. After all, we still lag behind overseas counterparts in chips and hardware. Here, our strong algorithm and software capabilities shine, enabling system-level optimization. We often say that it is hard for a solution to be perfect at every single point, but the overall game state can be optimal. This is precisely Longsight's logic for deconstructing solution designs.
Huanshi: To build a high-quality instrument, one must possess unique definition capabilities on the chip side or be able to develop chips independently to achieve this result. For general companies, this involves huge investment: these chips are primarily for internal use, so volumes aren't large, but the R&D cycle is very long. This may well be the barrier formed by years of accumulation by overseas giants, serving as their moat. For latecomers, the time, energy, and determination required to break this moat are undoubtedly immense.
Cui Tianyu: Yes, I am truly grateful to all colleagues who fought selflessly and dedicatedly throughout this process. Everyone remained calm and steady, holding their breath, determined to produce the most advanced domestic products.
Huanshi: In terms of technical indicators, where do your currently launched products stand compared to international peers? Have you conducted relevant comparisons? When presenting to customers, does this also boost your confidence?
Cui Tianyu: One of the first questions many customers ask is, "Have you compared yourself with Brand X?" We have indeed conducted comparisons at some customer sites. This process enhanced both our confidence and the customers' trust in us.
A real-world case occurred during testing: a customer used our instruments to test a newly returned chip IP. Suddenly, they called me saying, "Ms. Cui, the results from your instruments seem unsatisfactory." I said, "Okay, let's take a look." Later, we coordinated several high-end industry-standard instruments for comparative testing. Ultimately, it was precisely our instruments that most accurately identified a minor bug, and we assisted the customer in resolving it quickly. Following this incident, the customer immediately ordered two of our devices. So, the promotion process inevitably goes from skepticism to trust: initially doubting your instruments, then discovering they are indeed rulers capable of pinpointing issues. This is normal.
Huanshi: This is an excellent case study, which I believe offers inspiration and confidence to peers. After all, purchasing such equipment puts them under pressure too.
Cui Tianyu: Yes, the best requirements come from customers. Over the past two to three years, we have found that upgrades in the domestic industrial chain have fostered confidence in domestic products. At this year's WESEMiBAY exhibition, vendors launched semiconductor industrial chain-related products, and attendance was very high. I think this is the best support for the achievements of strengthening and complementing the supply chain.
Pressure from Both Ends: Pursuing Advancement in Research, Competing on Throughput in Production
Huanshi: Longsight has been established for three years. What achievements has the company made in product progress in its third year?
Cui Tianyu: As instrument industry practitioners, last year we launched the 90GHz real-time oscilloscope at WESEMiBAY; this year, we concurrently launched three products—a 110GHz signal source, spectrum analyzer, and vector network analyzer—along with an optical vector network analyzer.
Simultaneously, in AI network testing, we launched an intelligent computing lossless network tester, including a 1.6T tester. Besides real-time oscilloscopes, we also introduced sampling oscilloscopes. We have indeed released many new products, and we feel proud of this.
Huanshi: This is certainly worth pride. You must have invested enormous effort, after all, the accuracy of this "ruler" largely determines the success or failure of subsequent customer outcomes. Please share your feelings regarding the specific promotion process. If you are only targeting research institutions or non-R&D site quality inspection, requirements might differ; but at mass production sites, where large quantities of online detection equipment are used, instrument requirements vary.
Which end have you currently pushed towards? Are there differences or specific designs in your approach? Simply put, how do you make production-line customers feel confident using your products? Do you have any experience or confidence to share?
Cui Tianyu: There is an essential difference between scientific research high-end instruments and mass production line-test instruments. Research instruments focus more on advancement: as long as high-level papers can be published, indexed, or beautiful eye diagrams can be generated using the instrument, it is helpful to users.
But production testing is different. Production testing implies high throughput and high concurrent testing, meaning many products pass through this "ruler" before delivery to customers. Therefore, in this process, customers care about four points:
First, is the instrument accurate and of good quality? Can it serve as my "ruler"?
Second, is the efficiency high enough? Everyone wants to expand capacity and leap forward leveraging the AI wave.
Third, is automation and adaptation smooth? Managers want fewer operators.
Fourth, returning to cost, how much actual fixed asset investment can your instruments reduce for me?
Focusing on these four aspects, we have put great effort into products like sampling oscilloscopes and vector network analyzers. First, we must convince customers that we are comparable to benchmark products; then, through long-term parallel testing comparisons, provide credible data to build customer trust; next, improve efficiency and reduce costs, ultimately achieving a truly valuable replacement for customers.
Huanshi: Letting data speak is indeed crucial. I also understand that overseas competitors' equipment is expensive with long delivery cycles. Many companies buy them gritting their teeth; without buying, they cannot proceed, but upon buying, they face various constraints. Thus, having companies like Longsight in the industrial chain gives everyone another option.
The Hotter AI Optical Interconnect Gets, The Busier High-End Measurement Instruments Become
Huanshi: Looking at current emerging tracks, everyone is talking about silicon photonics, AI, and optical chip testing. What new opportunities do these emerging tracks bring to your field? What preparations have you made in this area, or what results do you hope to achieve?
Cui Tianyu: Currently, the entire optical track revolves around the main AI track. Around AI, extensive optical interconnection is required. Previously, the industry mainly relied on electrical interconnection; now it has expanded to optical interconnection. Optical interconnection has spawned many solutions and products, such as the well-known optical modules, which are extremely popular. Besides pluggable optical modules, the industry now combines silicon photonic chips, optical DSPs, etc., to create many different products. This includes LPO (Linear Pluggable Optics) solutions removing DSPs, NPO/CPO (Near Package Co-Packaged Optics) solutions, and USFP recently announced by NVIDIA.
Looking from within the optical domain, this track features various products and companies, attracting much attention. This track involves a complete industrial chain. For example, the optical industrial chain contains many optoelectronic and electro-optic chips. What tests do these chips require? They require optical VNA, sampling oscilloscopes, real-time oscilloscopes, and chip testing.
Huanshi: Does every chip need to be tested? Or is sampling inspection used?
Cui Tianyu: Chips undergo relatively comprehensive testing during the early design phase and the first and second batches. If the batch is confirmed free of issues, mass shipment can proceed. Who procures the shipped chips? Mainly manufacturers producing modules and optical interconnections.
Huanshi: Like top-tier companies such as Zhongji InnoLight.
Cui Tianyu: Yes, Chinese optical modules account for 70% or more of the global market. These leading companies procure chips and begin making their own products, which involve complex production lines. The production line stage is intricate, including optical coupling, DC characteristic testing, and eye diagram consistency testing. Eye diagram consistency testing requires sampling oscilloscopes, so sampling oscilloscopes are also a very hot track domestically, and Longsight is a participant.
Optical modules undergo full inspection; every module must be checked before leaving the factory. Leading companies ship tens of millions of units, meaning tens of millions must pass inspection via multiple sampling oscilloscopes.
Huanshi: Are your sampling oscilloscopes implemented in an online manner? What is their level of automation?
Cui Tianyu: Their efficiency is very high. We are very confident in our products: four channels process in parallel at high speed, delivering results directly in 4 seconds.
Huanshi: If optical module manufacturers wish to further improve production line efficiency, they can add Longsight as an option.
Cui Tianyu: Yes. Moreover, manufactured modules must eventually be applied. Previously, modules were mainly used in national backbone networks and core networks for connectivity; now, AI data centers and intelligent computing centers also require large quantities of modules.
Issues arise during the operation of these centers: after modules are racked, who is responsible for detecting and maintaining the stable operation of the entire center? This is essentially an O&M requirement, necessitating our newly launched network testers for testing and verification during deployment and operations. Thus, from the front-end chips, to modules, to the final mature network, testing is indispensable at every stage.
Huanshi: Seen this way, you indeed cover the entire product lifecycle, making this track very attractive. Besides the optical track, does Longsight have layouts in other markets? What target markets do you hope to enter in the future?
Cui Tianyu: We currently focus mainly on the electronic information industry, a track that already encompasses the optical interconnect, electrical interconnect, and electronic manufacturing directions mentioned earlier. Simultaneously, we pay close attention to emerging industries such as 6G, satellite communications, and commercial spaceflight, making many preliminary deployments. For instance, our RF instruments are already applied in frontier research in these tracks.
Huanshi: The directions you listed are also key priorities in the "15th Five-Year Plan," echoing the sense of national responsibility you mentioned at the beginning. May I ask General Manager Cui, does Longsight position its products as selling single units or providing comprehensive testing services? What role do you hope to play in the industry and customer service in the future? Do you have an advanced concept?
Cui Tianyu: Accurately speaking, previously we had no choice; if you only have single instruments, you can only sell single instruments. But now, the demand from the industry for instrument manufacturers like us is for testing to be interconnected, covering everything from individual links to the final product.
If data flows seamlessly throughout the process and efficiency reaches its optimum, that is ideal, saving customers significant trouble. Otherwise, customers need to separately purchase brands A, B, and C, then develop a complete system themselves. Moreover, purchasing instruments is not a one-time deal; more importantly, how to use the instruments effectively and serve their own goals. Therefore, subsequent service support places higher demands on manufacturers.
Instrument manufacturers often think managing the backend is sufficient: deploying more field engineers to assist customers on-site. Actually, it is not. If we look further ahead and intervene at the initial design stage, co-defining and co-designing with customers, predicting potential future issues, and preparing products in advance, this effectively forms a complete test solution toolchain. I believe this is what customers expect.
Huanshi: If so, customers can indeed save much trouble. At least you can provide more comprehensive turnkey solutions. For many customers crossing over from other fields, this seems like a good development strategy.
Beyond Building Instruments: How to Plan Long-Term for Domestic Measurement
Huanshi: As Longsight reaches today's milestone, what predictions do you have for the future regarding breakthrough paths and strategies? For example, in three years, what changes will occur in the currently booming optoelectronic measurement industry? Where will this field head with the joint efforts of numerous Chinese manufacturers? From your professional perspective as frontline practitioners, what are your feelings?
Cui Tianyu: Frankly, as a populous and major manufacturing country, China is firmly investing in autonomous controllability and independent R&D. Combined with the current international competition in AI, the overall expectation for the next three to five years is very positive. Meanwhile, optical and electrical interconnections will continue to evolve. For example, at recent international conferences, we have seen many top domestic and foreign enterprises release excellent results, so the overall outlook is very optimistic.
Looking further ahead, 6G R&D continues to advance, U6G has begun trial operations, and ground-air-space integration is progressing: there are AI computing centers on the ground, satellites in the sky, communication between satellites, and computing power sent to space. These are excellent concepts and plans that will bring continuous testing and verification demands, and we hope to exert effort in this direction.
However, returning to the instrument track, many directions discussed are promising, but instruments are somewhat back-end. Their role is not to stand on the front stage, but to support these tracks to run faster, measure more accurately, see more clearly, and perform better. Precisely because of this, instruments are a long-term track, not subject to drastic fluctuations, but rather flowing steadily like water, requiring long-term technical accumulation and gradual strength. So we remain very bullish, believing that looking back in three to five years, the instrument track will still be an excellent choice.
Everyone can feel that mobile phones have iterated from Nokia in the 2G era, through 3G, 4G, 5G, to smartphones, with industry companies undergoing many rounds of turnover. But one thing hasn't changed—testing hasn't changed. Modules are the same; previously rates might have been below 100Gbit/s, now reaching 800Gbit/s, 1.6Tbit/s, or even NPU interconnect rates hitting 6.4Tbit/s and 7.2Tbit/s, but the corresponding test solutions continue.
Huanshi: Every product needs testing, and the more advanced the product, the more testing required. My understanding is that if there are no new breakthroughs in future scenarios, growth rates may slow down, but testing demand will always exist. Therefore, it is particularly important for China to have autonomous and controllable tool vendors.
Finally, please use our platform "Chip Unveiled" to formally introduce the performance, features, advantages, and specific suitable customers for the new products launched at WESEMiBAY. We are happy to provide this opportunity for potential customers to find you through us.
Cui Tianyu: Thank you very much. Longsight launched multiple new products this year, including the 65G sampling oscilloscope. All manufacturers involved in optical modules, optical interconnects, and next-generation optical interconnects can try this highly efficient sampling oscilloscope, providing great opportunities for capacity expansion and cost reduction.
Huanshi: Let's talk about cost-performance ratio.
Cui Tianyu: Extremely high cost-performance; buying it is earning it.
Huanshi: Can shipment be guaranteed?
Cui Tianyu: Yes, ready stock is king. Additionally, for the hot track of high-speed connectors and cables, I strongly recommend our vector network analyzer. It supports simultaneous testing of hundreds of channels with test intervals less than 15 microseconds, marking a major revolution in efficiency. For cable and connector testing manufacturers, improving efficiency is a mandatory requirement. We can provide a complete cable tree test solution, balancing efficiency and quality, meeting high-frequency testing needs while covering a very large dynamic range.
For computing, AI, and lossless testing of network switches, we recommend our network testers. We have over 100,000 ports in our network, with very strong stability. You could say, as the Earth turns, our instruments turn too. Overall reliability and stability are sufficient to support network assessment and product evaluation.
Meanwhile, we must also "look up to the sky," with starlight filling our eyes. In the satellite and commercial spaceflight fields, our RF signal sources and spectrum analyzers are equally indispensable. We have achieved single-port direct output for 110G networking, featuring simple structure, good performance, and extremely low noise floor, capturing any signal.
Finally, returning to last year's starting point—the instantly famous real-time oscilloscope. We hope everyone continues to anticipate bigger breakthroughs in real-time oscilloscopes, aiming for larger scale in industry rankings and actual customer usage, occupying a greater share nationwide.
Huanshi: Thank you, General Manager Cui, for your passionate summary. After listening, I really want to buy one to try immediately, though I don't have a module factory at home. However, I will definitely recommend you to friends working in optical module companies. I wish Longsight continued bravery in the wave of domestic breakthroughs, winning a larger share and helping strengthen domestic high-end instruments.
From a concept, to Longsight showing its sharp edges, to ultimately claiming half the market, I truly believe it is worth expecting.
Cui Tianyu: Thank you. I believe that only by aiming to be number one can you truly keep running on the road. Instrument-making sometimes has a slower pace, but we have sufficient patience, meticulousness, and rigor in our products; in continuous running, we always maintain perseverance and determination. Thanks to everyone; we hope you have more expectations and interest in Longsight.
Interview | Huanshi Edit | Xiao Qie Review | Huanshi