eMemory Technology Inc. (3529) Earnings Call Transcript & Summary

Good afternoon, and welcome to eMemory's Fourth Quarter 2023 Webcast Investor Conference. Joining us today is our Chairman, Dr. Charles Hsu; President, Mr. Michael Ho; Head of IR, Ms. Li-Jeng Chen; and the Director of the Finance Department, Ms. Teresa Kuo. The format of today's event will be as follows: First, eMemory's Chairman, Dr. Charles Hsu, will give an opening remark. Afterwards, President, Mr. Michael Ho, will summarize our operations, followed by our business outlook. Next, Dr. Charles Hsu will give a talk titled, Enhancing Security from AI Servers to AI Edge Computing. Then we will give a demo of our improved IR website. We will conclude today's conference with a Q&A session, where our management team will answer your questions. [Operator Instructions] As a reminder, this conference is being recorded, and a webcast replay will be available after the conference is finished. For more information, please visit our company's website under the Investor Relations section. As usual, before we begin, we would like to remind everyone that today's presentation may contain forward-looking statements subject to risk factors associated with the semiconductor and IP business. Please refer to the cautionary statement on Page 3 of today's presentation. Now I would like to give the floor over to eMemory's Chairman, Dr. Charles Hsu.

Thank you, Millie. Good afternoon, everyone, and thank you for attending our conference call today. With a strong technology licensing momentum and new processes and applications gradually entering production, our company has begun to enter the next multiyear growth cycle. eMemory's technology licensing is a platform-based business. We develop our technologies at foundry's process platform. As customers use this process, we collect royalties by per wafer basis. Over the past 20 years, we have established 621 process platforms ranging from 0.5 micron to 3 nanometer in 25 foundries worldwide. The number of new technology license cases and the technologies under development and verification last year doubled compared to the past. This indicates our new technologies such as MTP and the PUF-based securities, are rapidly advancing and contributing to the expansion of our technology platforms. Consider this technology platform similar to the concept of production capacities that is more technology platforms, more wafer running and more royalties. So additional technology platforms and the process nodes will contribute and drive our royalty growth. The newly established technology platform, whether an advanced process incorporating PUF-based or process involving MTP-related technology will charge royalties per wafer much higher than the average, accelerating the growth of our royalties. Therefore, we are very optimistic about the future. Later, I will address a topic that everyone is very concerned about that go into whether our IP have been adopted in generative AI-related applications. Next, I would like to invite our President, Michael Ho, to share our fourth quarter performance and the future outlook.

Good morning, everyone. Now to begin with our 2023 fourth quarter financial results. The fourth quarter revenue was TWD 899 million, up 14.2% sequentially, but down 0.4% year-over-year. Operating expenses were TWD 359 million, down 2.9% sequentially, mainly attributable to the decrease in bonuses. Operating income was TWD 540 million, with an increase of 29.3% sequentially and an increase of 0.9% year-over-year. Operating margin increased by 7 percentage points sequentially and increased by 0.8 percentage points year-over-year to 60%. However, due to the strong appreciation of the Taiwan dollar at the end of December, our U.S. dollar fixed deposits incurred an exchange loss of TWD 95 million. As a result, our net income amounting for the TWD 404 million experienced a decrease of 0.5% sequentially and 6.2% year-over-year. EPS for the quarter was TWD 5.41 and ROE was 53.1%. Next, let's move on to the revenue contributions by licensing and royalty. Licensing in the fourth quarter accounted for 30.5% of the total revenue, up 5.7% sequentially and the 33.6% year-over-year. Royalties in the fourth quarter contributed 69.5% of the total revenue, increasing 18.4% sequentially but decreasing 10.4% year-over-year. Total revenue for the fourth quarter increased by 14.2% compared to the previous quarter, but decreased by 0.4% compared to the previous year. For the full year of 2023, total revenue decreased by 5.2% compared to the previous year. Total licensing and royalty increased by 24.8% and decreased by 14.1%, respectively. With that, I will comment on our revenue contribution by specific IPs. NeoBit accounted for 26.3% of total licensing revenue in the fourth quarter, increasing 1.4% sequentially and up 97.9% year-over-year. Its royalty accounted for 21.9% of total royalty, up 17.1% sequentially but down 41.1% year-over-year. NeoFuse accounted for 42.8% of the total licensing revenue in the fourth quarter, up 18% sequentially and up 18.3% year-over-year. In terms of total royalty revenue, NeoFuse royalties increased by 18.2% sequentially and 4.7% year-over-year, accounting for 75.5% of total royalties. PUF-based security IPs contributed 10.9% of licensing revenue, decreasing 47.9% sequentially and decreasing 29% year-over-year. Its royalties accounted for less than 1% of total royalties, up 87.2% compared to the previous quarter, but down 97.8% compared to the previous year. For MTP technology, accounted for 20% of the total licensing revenue, up 74.4% sequentially and up 99.1% year-over-year. Royalty from MTP increased 37.1% sequentially and up 50.5% year-over-year, accounting for 2.6% of total royalties. For the full year of 2023, the licensing and royalty revenue are as follows: NeoBit licensing revenue increased by 48.2%, while royalty revenue decreased by 39.8%, accounted for 25.5% of total revenue. NeoFuse licensing revenue increased by 4.3%, and the royalty increased 3.2% year-over-year, contributing to 63.3% of the total revenue. PUF-based security IP licensing revenue increased 13.4% year-over-year, but royalty revenue decreased 87.1%, accounting for 4.3% of the total revenue. MTP technology licensing revenue increased 88.4% year-over-year, but royalty revenue decreased 38.1%, accounted for 6.9% of total revenue. Now let's look at the royalty for 8-inch and 12-inch wafers. For 8-inch wafers accounted for 37.8% of royalties, up 23.4% sequentially, but down 28.5% year-over-year. 12-inch wafers contributed 62.2% of royalties, increasing 15.5% sequentially and up 5.9% year-over-year. In total, 149 product tape-outs were completed in the fourth quarter. We will provide more information in the management report. In the next section, I will address our future outlook. For licensing revenue, strong licensing demand will continue to drive the growth momentum of licensing fees. For royalty revenues, we expect royalties in the first quarter to be temporarily affected by a customer moving to a more advanced node, resulting in a single-digit quarter-by-quarter decline, but growth year-over-year. Overall, revenue will decline by single digits quarterly, but will increase by double digits yearly. Moving to the new IP technology and business development. First, specialty process like a high-voltage, high-K, BCD, embedded flash and emerging memory. There will be more foundry technology licensing this quarter. The per wafer royalty ASP will increase, due to the ongoing development into the more advanced nodes. The continued adoption of RRAM by more customers will drive increasing application and contribute to royalty growth in the future. Second, advanced process. Last quarter, we successfully licensed our 3-nanometer OTP and PUF technology to a U.S. foundry customer, and will be collaborating on the development of the most leading process in the future. Furthermore, our PUF-based security has been successfully adopted in a leading U.S. high-performance computing, HPC, customer's related applications. There are multiple projects in the 3-, 4- and 5-nanometer under progress. This concludes my comments. Next, I will pass the time to Charles. Thank you.

Okay. In the next several minutes, I'm going to talk about the Enhancing Data Security from AI Servers to AI Edge Computing. AI has experienced widespread adoption across various computing environment, ranging from edge devices to servers. This AI systems comprise a multitude of essential components such as the CPU, GPU, DPU, accelerators in charge of computation and processing, as well as storage components like SRAM, DRAM, solid-state disk drive and hard disk drive and peripheral devices such as sensors and actuators. Our technologies and IPs, including OTP, MTP and PUF-based root of trust have been deployed on all the mentioned components, enhancing the overall functionality and security of AI system. Our solution not only contribute to bolstering data security in memory components, but also prove instrumental in enhancing yield. There are different types of memory in the memory hierarchy of computing system, spanning from SRAM, DRAM, solid-state disk to hard disk drive. For all AI and high-performance computing, high-density SRAM and DRAM are needed. And our OTP IPs play a critical role in repairing the broken bits of memory to improve the yield. In addition, the CXL memory facilitates high-speed CPU-to-device memory access. Security is also required in the CXL memory controller. Our PUF-based security solutions are deployed to enable security functions and ensure robust secure data transmission. Another contribution to our IPs to the DRAM module is EEPROM or MTP embedded in DIMM card's, SPD hub. The EEPROM is used to store the configuration information for the DIMM and the DRAMs. And for nonvolatile memory storage in the solid-state disk and the hard disk drive, our security solutions are embedded in the disk controller to ensure the protection of the stored data. All the aforementioned memories are used in AI servers and our IP have been deployed into the memory system of these AI servers. And in addition to the memory system I just mentioned, the computing elements, such as CPU and the GPU, will transition toward confidential computing to protect data used in computation and the processing. As shown in this figure, root of trust security solutions are embedded not only in the CPU and the GPU, but also in the memory controller such as solid-state disk and hard disk drive and the CXL controllers. And for the fast response of IoT device at the network edge, AI are frequently used in edge device. Similarly, our IP used in AI servers also serve the same functionality in AI edge devices, as shown in this slide. However, there are additional component in AI edge devices compared to the AI servers that will leverage our IPs, including actuators and sensors and embedded flash. Our OTP is used for trimming functions on analog circuit design for sensors and actuators. Additionally, our NeoFlash in the near future will be utilized to store the AI model and the program code. So in summary, while eMemory does not directly developed AI accelerators, our IP plays a very critical role in AI servers and AI edge devices. Various components such as CPU, GPU, SRAM, DRAM, data storage controllers, sensors, actuators and more utilize our IP to enhance security functions and improve yield. The wide adoption of the AI application will speed up the utilization of IPs and make eMemory grow faster. For example, every AI chip needs SRAM repair, and it is a very big business. And this concludes my remarks. And next, I will hand the floor over to Millie.

Thank you, Charles. This quarter, my team and I have been dedicated to improving the IR website to ensure that all of our resources are accessible. We want to make our information transparent to all investors. So we have prepared a website demonstration to show you some of the improvements we've made, and we will show the video. [Presentation]

That concludes our prepared statement. Next, we will enter the Q&A session.

[Operator Instructions] All of our questions will follow the format of answering the Chinese version first, followed by the English version. We will now collect the questions and begin our Q&A session.

[Foreign Language]

[Foreign Language]

The question that was asked was the aggressive expansion of Chinese foundries may put severe downward pressure on wafer prices. How will this affect your company? Michael?

OTP technology is now a standard for applications like display driver, PMIC and various sensors in mature process. As foundry expand, the need to license our technology expected to boost our licensing revenue where the industry face oversupply and pricing pressures due to the foundry expansion. The effect it has on our company will be offset by the increase in royalties per wafer as existing customers move to more advanced process. In order to survive, existing foundries are focusing on developing specialty process with high value add, such as NTP, embedded flash and PUF, as is happening now. Our royalties for MTP and PUF-based are much higher than OTP. Furthermore, the increasing contribution from the higher wafer price of advanced nodes leads us to believe that the average royalties per wafer will continue to rise. Foundry expansion in Europe, the U.S. and Japan are driving the need for local production by large customers in the future. Deploying our IP simultaneously in foundries across each region will facilitate convenient adoption for customers, driving increased usage and expanding our business across diverse customers and applications. Thank you.

[Foreign Language]

[Foreign Language]

The question was, will cost constraints push your customers to use eFuse, which is free of charge to reduce expenses? Michael?

Our customers developed their first-generation products using foundry's eFuse. However, in response to competitive pressures and the need for improving cost and performance, they move to transition to our IPs. Our OTPs offers improved yield and reduced the chip area compared to eFuse and enable customers to protect their intellectual property like algorithms stored in the chip. The success of this has led to continued adoption of our IPs in subsequent generation of products. Thank you.

[Foreign Language]

[Foreign Language]

Over the past year, discussions about the slowdown of Moore's Law has become more prevalent. Will this affect your rate of improvement in foundry penetration? And considering that many IP replacements occur during process upgrades, could the deceleration of Moore's Law potentially have a negative impact on your company. Michael?

2014, we have been developing IPs for technology below 16-nanometer FinFET process. Typically, the foundry provide in-house eFuse for the initial customers for the most leading process. After the process stabilized, third-party IPs like ours are permitted for the technology verification, often occurring after the first stage of customers have entered in mass production. This means that third-party providers like us need to wait for those migrating from the previous generation to adopt our technology. As Moore's Law slows down, the launch time for the next generation extends, creating a longer interval of time like current 2 nanometer and 3 nanometer. This year, once we successfully verified the 3-nanometer OTP, we still can catch customers seeking for the most advanced process. Overall, this is a positive for us. Moreover, there are always customers migrating to more advanced process nodes, such as moving from the 7-nanometer to 5-nanometer or 16-, 12-nanometer to 7-nanometer. As long as we have completed process verification, there will still be many opportunities for customers' adoption. Even with the mature process, these opportunities remain unaffected by the slowdown or change in Moore's Law. Thank you.

[Foreign Language]

[Foreign Language]

In Charles' talk just now, he mentioned that our IPs play an important role in memory. Since memory is a huge market, how are we progressing in this area. Charles?

Our DRAM royalties will increase significantly, driven by the increase in production of existing customers and the new customers ramping up production.

[Foreign Language]

[Foreign Language]

The question was, you mentioned a customer moving to a different process in quarter 1. Could you provide details on the specific application associated with this migration? Michael?

Many major OLED DDI customers switching from the 40-nanometer to 28-nanometer, leading to a temporary decrease in the 40-nanometer shipment. However, as the 28-nanometer and other new process applications begin production, royalties will soon increase. Thank you.

[Foreign Language]

[Foreign Language]

The Chairman recently introduced the concept of production capacity, highlighting the increasing number of foundry capacity will increase our total addressable market. We have been actively establishing our analogy platform in these newly built facilities. Can they directly transform to our royalties in the future? Charles?

Yes. Our total addressable market will increase as the world expands foundry capacity and move toward more advanced technology. Our technology coverage in each foundry will increase as more technology process is developed and more fab are established. With each generation of technology, there is need for the continuity to move to the next generation. Therefore, our technology increasingly available in each foundry. Historically, our royalty income for each foundry have experienced an upward trend. Consequently, as the number of fab and technology platform increases, we anticipate a corresponding growth in license fee and [indiscernible] usage and royalties.

[Foreign Language]

[Foreign Language]

The question was how much do you spend on R&D? Michael?

About 70% of our expenses are allocated to R&D. These expenses are mainly comprised of employee salaries, software, other equipment as well as application and maintenance costs related to the patents. Our R&D expenses are mostly sustained by our partners. A key factor contributing to our high operating margin, the high cost associated with testing and qualification, including the foundry shuttle, mask, test chip and more are fully covered by our foundry partners. They not only cover these costs, but also pay us a license fee. A similar arrangement exists with the technology partner. Thank you.

[Foreign Language]

[Foreign Language]

In the past 10 years, your employees have increased by an average of 5% each year, while your revenue increased by an average of 20%. Additionally, your company also managed to develop many technologies. Why is this the case? And is the goal of increasing staff in the future still the same? Michael?

Main reason why we can keep developing new technologies by managing to keep our employee headcount relatively the same is because our IP is reusable. Once we qualify our IPs on the foundry platforms and registered on our IP library, customers can use our IPs anytime, unless they require customization. This means that our R&D team can always focus on developing the new technology on the more advanced nodes. The other main reason is that because of our unique business model, since 70% to 80% of revenue is contributed by royalty. Royalty-based business drives the revenue as more customers adopt our IPs without increasing the employee headcount. In the future, our hiring plan will mainly revolve around the advanced process, projects and special projects, such as NeoFlash, RRAM and security. We will maintain a similar hiring pace. Thank you.

[Foreign Language]

[Foreign Language]

The question was, the ForEx losses in the fourth quarter of last year seemed to have an impact on your earnings. How do you plan to address this matter? Michael?

Foundry partners have suggested paying royalties in Taiwanese dollars based on the exchange rate at the payment date and time. Our customers directly pay us in U.S. dollars. We deduct necessary fees, convert the amount to NT dollars and keep the remaining sum in U.S. dollar time deposits to gain the higher interest compared to put in NT dollar. The foreign exchange losses incurred in December will be partially offset this quarter as the NT dollar depreciation -- depreciated early in this year. We will review our current currency accordingly. Thank you.

[Foreign Language]

[Foreign Language]

You mentioned that existing customer applications are moving to more advanced process nodes. Which applications have experienced the migration to advanced nodes? Michael?

Taking OLED DDI for example, the main process now in recent years has been the 14-nanometer. However, this year marks a significant move toward 28-nanometer production, with future plans for development in the 16- and 17-nanometer. In the case of PMIC for smartphones, which were previously mainly in the edge like 0.18, 0.13 micron generation. Some will move to the like 90 and 55 production. Similarly, TV SoC used to be in the 12- and 14-nanometer. One of our customers took the lead to move to 6-nanometer to gain the market share from the 12- and 14-nanometer players. These transitions are all driven by the need for customers to incorporate additional features, achieving faster speeds and reducing power consumption to enhance the product competitiveness. Thank you.

[Foreign Language]

[Foreign Language]

Is there any progress on SRAM repair? Charles?

Yes. Multiple customers are working with us, and we are collaborating with the EDA tool company to incorporate our OTP into their SRAM compiler to facilitate customers' adoption.

[Foreign Language]

[Foreign Language]

What kind of advantages does eMemory have in developing RRAM and MRAM? Michael?

Strengths lie in our strong circuit design and device development team. With years of accumulated design experience, numerous patterns and a profound understanding of customer needs, we excel in quickly assisting foundries in customizing specific spec for the RRAM and MRAM. Our expertise extends to designing IPs toward to meet the customer requirements, and we are able to quickly complete the qualification on foundry process so that the customers can use our IPs in their designs and move forward the world of mass production. We have already completed verification at the major foundries and the customers have begun to adopt our IPs in their chip designs. Currently, we are moving toward advanced process. Thank you.

[Foreign Language]

[Foreign Language]

Is it possible for customers to simultaneously use different technologies from your company? And how do you charge royalties in that case? Michael?

Customers do have many products that simultaneously utilize different technologies from our company such as OTP plus PUF, OTP plus MTP or MTP plus MTP to meet their spec product functionality design requirements. When a product use multiple of our technologies, we will charge relevant royalties for each different technology, respectively. Thank you.

[Foreign Language]

[Foreign Language]

The question was, lately, there has been instances of M&A in Taiwan semiconductor industry. Have you considered expanding your technology portfolio through M&A to expedite your growth? And if yes, what technologies or IP products are under consideration? If not, could you provide more insight into the reason behind this decision? Charles, can you offer an explanation?

Okay. Our company has many groundbreaking nonvolatile memory invention and the technologies that surpass all those of other companies globally. As long as we continue to develop this innovation and technologies along our roadmap, we can accelerate our growth rate. Currently, acquiring other companies is not under consideration.

[Foreign Language]

[Foreign Language]

The question was, eMemory has been working on MTP for a long time. What is the reason behind MTP's slower take off? And why do you believe that MTP is now entering its growth phase? Michael?

The reason MTP faced slow adoption is because in terms of memory cell architecture and the peripheral circuits, MTP is more complex. The complexity at the fundamental physical level result in a longer lead time between the technology development and the customer production. To develop the best solution, the MTP technology specifications needs to be aligned with customer ICs, and the MTP technology itself should offer cost-effective and competitive advantages over the external alternatives. However, each authorization in IP specification require requalification and development with the foundry process technologies, typically taking a few years before achieving significant customer volume shipments. Moreover, scenarios where multiple rewrites and embedded are imperative, alternative solutions with higher specification and maturity such as eFlash are available. To choose MTP, there must be compelling reasons such as the cost considerations or challenges in integrating eFlash into the specialty process like high-voltage and BCD. However, MTP is gradually gaining momentum due to the shift in customers and market applications demand toward embedded MTP solutions. This chip aims to enhance in-house chip in terms of both functionality and the cost effectiveness. Various mainstream requiring MTP are progressively increasing. Notable example, including the DDR5 SPD, which require 100,000 rewrite times, making embedded EEPROM necessary. The enhanced complexity of the new SPD features makes external EEPROM used and suitable, and eFlash is less fitting from a cost perspective, thereby providing a favorable entry point for MTP. Other applications such as the e-paper drivers are linked toward MTP, due to the need for regular adjustments in more complex panels. Thank you.

[Foreign Language]

[Foreign Language]

Will you benefit from the 12-nanometer project between Intel and UMC? Michael?

UMC is our second largest foundry customer. We also licensed our technology to Intel's 3-nanometer in Q4 last year, and we'll keep up with their technology development in the most leading process. We are optimistic about the cooperation between 2 customers. Thank you.

[Foreign Language]

[Foreign Language]

Samsung is your largest [ end-chip ] customer. Does Samsung's S24 mobile phone use your IPs? And will their memory-related products also include your IPs? Michael?

Currently, our IPs are adopted in the ISP and OLED DDI. In the future, more applications will also adopt our IPs such as PMIC. Memory-related has already adopted our technology for the DDR5 module-related chips in the SPD hub and the PMIC. In addition, we are developing embedded memory-related technology in their own fab, which will gradually move its way into more products. Thank you.

In the interest of time, we will ask our last question.

[Foreign Language]

[Foreign Language]

What security problems does your NeoPUF solution solve? If not NeoPUF, then what is the alternative? And why will NeoPUF be increasingly adopted in the future? Charles?

NeoPUF's -- eMemory's NeoPUF solution leverages the unique physical uncomparable variation that occur in the [indiscernible] and manufacturing process to solve multiple security issues, such as device authentication. NeoPUF generates a unique identifier for each device based on its inherent physical and comparable variations. This unique identifier can be used for the device authentication, ensuring that the only legitimate devices are authorized to access a system or network. And the second is for key generation, NeoPUF can be used to generate private keys that are unique to each device, to enhance the security of cryptographic protocols such as encryption and digital signature. And the third is for the anti-counterfeiting. The uniqueness of the anti -- NeoPUF can be used to detect counterfeit or cloned devices. By verifying the password stored in the NeoPUF during the authentication process, it becomes challenging for attackers to replicate the device identity. And the fourth is for the anti-tampering. NeoPUF generates a private key and a public key pair, enabling the signature function to prevent files from being tampered with. And the fifth function is random number of generation. The inherent unpredictability of NeoPUF can be used to generate random numbers, which are crucial for various security applications, including cryptographic protocols and secure communications. If the chip does not use PUF, random number must be generated by a random number generator, we call it TNG. However, TNG relies on the competition between 2 identical circuit to generate random numbers and result in several problems. First, the randomness will be insufficient and the generated random number may not be unique. And the second, the method for generating random number is easily influenced by surrounding variables such as temperature, [indiscernible] and the noise, leading to the result is inconsistent.

Thank you. We will now begin the closing comment. Charles, please proceed.

Okay. Thank you for attending our conference. For more information about our PUF-based security IP and technology, we encourage you to visit PUF security website at www.pufsecurity.com and check out our articles and other materials. And thank you once again for your patience and the support for eMemory. We will continue to work hard to -- on the technology and IP innovation and the PUF-based hardware security solutions for our customers and bringing higher returns for our shareholders. Thank you.

Thank you, everyone, for joining us today. Please be advised that the conference recording will be accessible later. You may now disconnect. Goodbye, and have a good day.