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

Welcome to eMemory's 2020 Second Quarter Webcast Investor Conference. Today with us, we have Chairman, Dr. Charles Hsu; President, Dr. Rick Shen; Vice President of Business Development and Spokesman, Mr. Michael Ho; Director, Ms. Lijing Chen; and Director of Finance Department, Ms. Teresa Kuo. We will start off today's meeting by an opening remark given by our Chairman, Dr. Charles Hsu. Next, our President, Dr. Rick Shen, will summarize our operations in the second quarter followed by an outlook for the future. Dr. Hsu will provide a closing comment at the end. [Operator Instructions] As a reminder, this conference is being recorded, and the webcast replay will be available within 3 hours after the conference is finished. Please visit www.ememory.com.tw, 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 that are subject to risk factors associated with semiconductor and IP business. Please refer to the cautionary statement as shown on Page 3 of today's presentation. Now I would like to turn the floor over to eMemory's Chairman, Dr. Charles Hsu. Dr. Hsu, please go ahead.

Thank you. Good afternoon, everyone. I hope you all are healthy and are doing well. In Q2 2020, we delivered record high revenue. As mentioned in previous conference calls, we are poised to enter a multilayer growth cycle with a strong set of tape-outs in the pipeline. Besides the continuous strength of 12-inch royalty, 8-inch royalty has gained momentum as 2 to 3x PMIC content increase in 5G smartphone as well as medical, online working and entertainment-related demand. In addition, we are experiencing strong interest in particular for our security PUF-based technology due to security requirements for 5G, AI, automotive and IoT. So we are actively engaging with the leading companies for design in our NeoPUF into their security architectures and expect more design win for the rest of the year. Next, I would like to invite our President, Rick, to report our operating results of Q2 and the future outlook of our business. And afterwards, I will elaborate more on our NeoPUF unique quantum tunneling mechanisms and why we believe we will be the winner of the security silicon IP provider. Thank you. Rick?

Good afternoon, everyone. This is Rick. Thank you for attending eMemory's 2020 Second Quarter Investor Conference Webcast. In today's presentation, I would like to report our operating results of 2020 Q2, followed by the status update of our technologies and the future outlook. To begin with, I would like to report our 2020 Q2 results. Q2 revenue was TWD 423 million, an increase of 1.9% sequentially and 33.7% year-over-year. In terms of the U.S. dollars, Q2 revenue was USD 14 million, a growth of 2% sequentially and 38.1% year-over-year. The operating expenses of Q2 was TWD 227 million, up 2.7% sequentially and 21% year-over-year. Q2 operating income increased 1% sequentially and 52% -- 52.3% year-over-year. The operating margin was 46.3%, a decrease of 0.4 percentage points sequentially but increased 5.7 percentage points year-over-year. EPS of Q2 is TWD 2.28, and the ROE at 41.9%. For the first half of 2020, the revenue was TWD 839 million, up 17.9% year-over-year. The operating expenses increased 15.4%, and the operating margin was 46.5%, with an increase of 1.2 percentage points. EPS, up 18.3% to TWD 4.66, and ROE gained 5.9 percentage point to 42.8%. In the following section, I will break down the revenue contribution by licensing and the royalties. Licensing from Q2 is TWD 118 million, accounted for 27.9% of the revenue, up 10.9% sequentially and 12.6% year-over-year. In terms of U.S. dollars, licensing revenue is USD 4 million, up 11.6% sequentially and 17.2% year-over-year. Royalty contributed to 72.1% of the total revenue, is TWD 305 million, decreased 1.2% sequentially but increased 44.1% year-over-year. In terms of U.S. dollars, the royalty is USD 10 million, down 1.4% sequentially, but up 48.4% year-over-year. In the first half of 2020, the total revenue grew by 17.9% as compared to the previous year. Licensing and the royalty each has a growth of 6.6% and 22.6%, respectively. In terms of U.S. dollars, licensing and royalty grew by 10% and 26.2%, respectively. If we break down the revenue by technologies, the results are as follows: NeoBit accounted for 17.2% of the total licensing revenue of the second quarter, decreased 15.6% compared to previous quarter and 26.7% year-over-year. Its royalty accounted for 65.6% of the total royalty, increased 0.5% sequentially and 31% year-over-year. This was due to the contribution from PMIC, DDI and TDDI. NeoFuse accounted for 71% of the total licensing revenue of the second quarter, up 12.3% sequentially and 27.6% year-over-year. Its royalty decreased 2.3% sequentially but increased 100.4% year-over-year. Because of the increasing contribution from 40- and 28-nanometer OLED, DDI, DTV and their networking applications, the royalty of NeoFuse contributed 31.5% of total royalty. Our PUF-based security IP contributed to 2.4% of licensing revenue in the second quarter. Although this technology has not contributed royalty yet, engagement with industrial leaders are still actively ongoing. As for MTP technology, licensing revenue increased 56.3% sequentially and 3.3% year-over-year. Royalty from MTP decreased 22.2% sequentially and 18.3% year-over-year. Our MTP team is still working with partners on developing MRAM, ReRAM and AI memory. In the first half of 2020, for NeoBit, the licensing revenue decreased 6.9% year-over-year, but royalty increased 1.2%, accounting for 52.9% of total revenue. For NeoFuse, the licensing and royalty revenue grew 21.4% and 136.8% year-over-year, contributing to around 42.1% of total revenue. For PUF-based security IP, licensing revenue increased 443.2% year-over-year, about 0.4% of total revenue. For MTP technology, the licensing and the royalty revenue declined 42.6% and 15.2% year-over-year, accounting for 4.6% of the total revenue. If we break down royalty by 8-inch and 12-inch wafers, 8-inch wafers, which accounted for 64.1% of the royalty, decreased 1.2% sequentially but increased 22.7% year-over-year. Royalty from 12-inch wafers contributed 35.9% of the royalty, decreased 1.2% sequentially but increased 109.4% year-over-year. Regarding customer product tape-outs, 115 product tape-outs were completed in Q2, up 10.6% from the previous year. This shows the persistent design activities of customers with our IPs. We will provide more information in the management report that will be released later today. In the next section, I would like to address our future outlook. We expect the growth of the revenue to continue in the second half and beyond. For licensing revenue, the main contributors to our revenue are still NeoBit and NeoFuse. Due to the increasing advanced technology platforms and a more comprehensive PUF-based security IP, we anticipate that licensing from NeoFuse and the NeoPUF will continue to grow. For royalty revenues, we expect 8-inch wafers royalty to regain momentum due to the 2 to 3x PMIC content increase in 5G smartphones. In addition, we have developed various automotive-grade process and expect the EV automotive sector will drive further growth momentum of production royalty for 8-inch wafers platform. 12-inch royalty will continue to grow as more than 300 tape-outs for various applications in the pipeline ready for production, which including -- which include TDDI, OLED, Tcon, ISP, Bluetooth, Wi-Fi, TWS, switch, set-top box, video processor, SSD controller, IoT security processor, digital power, DRAM and others. For new application development, our new applications are centered around the development of PUF-based technology. As mentioned before, the PUFrt was adopted by customer in IoT platform, AI training modules, blockchain, FPGA and industrial automation applications to ensure data security and increase attack-resistance ability. Apart from PUFrt, our PUFiot is also well-developed, and engagement with customers is ongoing. Aside from our development efforts, our collaboration with ARM to embed our NeoFuse in its security platform, which already has customer adoption and has entered the production verification stages mostly, contribution from the development mentioned above will be seen this year. For the new technology development, in addition to 5-nanometer technology platform, MRAM, ReRAM and ARM security IP platform, which we mentioned in previous quarter, we are developing 6-nanometer and 5-nanometer plus technology with our leading foundry partner and have already demonstrated 6-nanometer silicon results successfully, while the 7-nanometer has been adopted into automotive application. Second, we are developing new security IPs. After the completion of PUFtrng, that is the PUF-based true random-number generator; and the PUFrt, that is the PUF-based root-of-trust, our PUF-based IoT security solutions, security elements and the hardware security module IPs are also under development. Lastly, we are in the process of creating an open platform. This open platform focus on PUF-based hardware security by integrating OTP, PUF, security-function IPs and the cryptographic algorithm IPs to provide total security solutions for AIoT and the 5G applications. Despite the global pandemic, the need for increased performance, yield and the security continue to create strong and sustained demand for our IPs. I would like to thank all our employees for their efforts in continuing success in penetrating new markets and maintaining our position in existing markets. With that, I will conclude my presentation on our operating results and the future outlook. Thank you for your attention.

Thank you, Dr. Shen. Now I would like to invite Chairman, Dr. Charles Hsu, once again to give an introduction of NeoPUF's quantum tunneling mechanism. Dr. Hsu, you may go ahead.

Okay. Okay. So in the following session, I'm going to tell you a little bit about the mechanism of quantum tunneling to form our NeoFuse and also by using the NeoFuse to form our NeoPUF. Please turn to Page 23. Here, I will explain the mechanism of our NeoFuse. We invented the NeoFuse transistor. The programming mechanism is to apply voltage just high enough to generate a tunneling current path in the gate oxide, and we call this the current path quantum tunneling. And as shown in the figures, we take the -- this NeoFuse transistor, and we apply the voltage. And on the right-hand side of the figure, you can see the path from the gate to the substrate and -- after the programming. So if we want to -- and we call the cell after programming, the data is one and the result programming is 0. So only the cell with the programming voltage applied to it, you will see the tunneling path. But on the cell without apply the voltage to it, it will -- it does not have the tunnel impact. So by doing this, we can differentiate 1 and 0. And so when we want to store the data into this -- our NeoFuse, then you can apply -- some of the cell, you apply the voltage in, and some of the cell, you don't. Then you will get your data stored in our NeoFuse. And in the Page 24, I will explain what does it mean by quantum tunneling. Okay. So in the Page 24, on the left-hand side of the figure, it is the physical structure of the oxide with many defects in the upper figures and without defects in the lower figures. The figure on the right is the corresponding energy band diagram. And from semiconductor physics, we know that defect in the oxide is a trap, and in the corresponding band diagram, it is the quantum well. So if you have many quantum wells in the oxide, it will facilitate the electron to tunnel from one well to another such that the electron can tunnel from the substrate to the gate and contribute to the tunneling current. And for the -- in the low figures, in the oxide without defects, it is very difficult for electron to tunnel from the substrate to the gate. And if you look at the energy band diagram, that one has to surmount a huge energy barrier in order to get over the oxide and reach the gate to contribute any current. So in such case, consequently, it generates very little tunneling current. So -- and once the trend or the defect is generated by the high-voltage stress, it is very difficult to recover. And from literatures, it shows that the broken silicon oxygen bond needs at least more than 600 degree C to recover. Therefore, the program number created by this mechanism is very stable. It is resilient to voltage variation, temperature variation, noise and the aging effects. So that's why we have so many customers use our NeoFuse in their chips, and none of them has any issues because once the OTPs they use -- NeoFuse they use in their chips, there's not any issue occur because the NeoFuse, once you store the data, pulling the data into it, it is very stable. Okay. After explaining the tunneling mechanism, let's turn to Page 25. So as I mentioned that the quantum tunneling is a programming mechanism in NeoFuse. So it is known for its reliability. So data will stay there much more than 10 years. And also because its mechanism is [ complementary ], so it is not visible and it is not traceable because it is very difficult to trace what kind of the bond breaking in the second oxide. So it will become perfect for the secure storage because it's invisible and untraceable. In contrast to the eFuse, which is some of -- it's the traditional method for the onetime program or storage. And you can see that the mechanism of eFuse is when you want to program the data into the eFuse, you have to burn up the fuse to establish one. So in such a circuit, you can see in Page 25, on the right-hand side, you can see that we compare the unprogrammed eFuse and the programmed eFuse. You can see that in the programmed eFuse, this very clear mark, the broken metal lock there, so it is the -- it is traceable and also visible if you do the reverse engineering, and that will cause the data leakage. So it is not safe to store the data in the eFuse. And please turn to Page 26. So now if we take the 2 NeoFuse as a pair, okay, and apply the voltage on their gate in parallel. So when the voltage is high enough, we will see the oxide with more defect will generate -- with more defect will -- with more defects generated will begin to have higher tunneling current, and the oxide with fewer defects generated will have less tunneling current. So every time when we choose a pair of NeoFuse transistor to compare the tunneling current during high-voltage stress, we will always see one of them, either the left one or the right one, having higher tunneling current. So it is like tossing a coin, you never know which side will turn up. So by looking at the Page 26, the figures on the -- in the middle, so you can see, if we apply the voltage to appear for NeoFuse, you will see the probability of tunneling occur in the left-hand side is the same as the probability occur in the right-hand side, also 50%. Okay. So in such case, so when the tunneling occur on the left-hand side, we will define it as 1; and the tunneling occur on the left-hand side, we will define it -- right-hand side, we will define it as 0. So as our NeoPUF was based on the nature randomness of the oxide quality, so we amplify the variation of the oxide and transform them into digital signal, so -- which become the unique fingerprint for chip itself. So in summary, our NeoPUF is built on the foundation of our NeoFuse technology and shares the same quality of reliability, invisibility and non-traceability. Since we have many NeoFuse platforms already, so that means we have also many NeoPUF platforms that are available and ready to be used. So NeoPUF is the one and the only one produced using quantum tunneling mechanism, and this is something no other competitors or any newcomers can do and to catch up with. So I think our NeoPUF technology will dominate the world's PUF technologies. Okay. So in the -- next, I will explain how we use the NeoFuse and the NeoPUF to build the hardware root-of-trust, because hardware root-of-trust is the most important fundamental element in the hardware security. So you can look at Page 27 and 28, okay? A root-of-trust must ensure that the secret key is stored in a manner that cannot be detected and is not susceptible to reverse engineering, and -- such that we can guarantee the safety of the data. And the root-of-trust is, therefore, the key component that protects the storage of system data and maintain its integrity. So our PUFrt solution, PUF root-of-trust, PUF-based root-of-trust solutions, leverage our NeoPUF and the NeoFuse technology to establish a robust root-of-trust. So how do we do it? First, we extract a unique number from the NeoPUF for the chip, okay? So first, we extract the unique number from the NeoPUF for the chip. So that becomes the unique ID, and essentially, it is a digital fingerprint for the chips. And this unique ID is then stored within our NeoFuse, so -- where it remains securely and free from detection and the threat of reverse engineering. So for key generation, okay, any key you want to generate based on this -- our root-of-trust, which is the key generation is crucial for coding and decoding sensitive data. Coding, decoding means decryption and encryption, okay? Decryption and encryption can be executed through the combination of the unique ID and the true random-number generator. And the keys are securely protected from the physical tampering in the embedded secure NeoPUF -- NeoFuse OTP, and this help solve the major security problem that chip designer faces. Okay. So in summary is we extract the secret from the NeoPUF, and because it's a random source, so nobody knows what the number is extracted. And then we install this number. We treat this number as a unique ID and then we store in the NeoFuse. And the NeoFuse, it is untouchable and it is invisible. So this is very -- it's the root-of-trust. And then for the future, any keys you -- based on this root-of-trust, you generate the additional keys, okay? But -- so when you use the key outside this region is those keys, which is not the original, unique ID or the fingerprints, okay? So the root-of-trust based on NeoFuse and the NeoPUF has many outstanding features, in particular, ease of use, high speed and low power utilization and low cost. Before the availability of our root-of-trust solutions, our customers need to incorporate at least 3 IP vendors' IP, for example, the OTP from one vendor; platform, another vendor; and the trng from -- true random-number generator from another vendor. So these 3 IPs into their design and -- all use very expensive external hardware random number generator. So consequently, we have a strong conviction that our PUFrt will become the future market leader due to its overwhelming competitive advantage. Okay. So this is about my introduction and explanation for the quantum tunneling PUF.

[Operator Instructions] We have a question from a guest. Please provide some example of NeoPUF customers and which specific type of NeoPUF product is driving the development.

Okay. Regarding the NeoPUF customers, here, I would take 2 cases as an example. One of our customers use NeoPUF for the NB-IoT. That is for the interconnect application. The -- another customer use the NeoPUF for the FPGA application. So I think that is a very good example for everyone to know that the NeoPUF can be used for interconnection and also for the FPGA and the other security-related applications. So for the -- another question is which specific type of NeoPUF product is driving the development. Well, I think the NeoPUF just provide the random numbers for security IPs. So the security IP-based on NeoPUF can enable some use cases. As we mentioned the other day, PUFrt, that is the root-of-trust, that is based NeoPUF to generate the UID and key storage-related usage. And now we also directly PUFiot so that we can embed the CPU inside so that customer can use it based on PUF to encrypt and to decrypt the data. So we think -- there are various applications which will drive the NeoPUF usage in the near future.

I can add some more comments. We have the customers use NeoPUF alone, but also we increased -- we create some additional function to create a so-called PUF root-of-trust. That's our second IP related to PUF. And for PUF root-of-trust, as I mentioned, inside the PUF root-of-trust is NeoFuse, NeoPUF and true random-number generator. I think, basically, customers very interested on using this one as their -- for their solutions because it provides root-of-trust functions, okay? So currently, a lot of customers request on this. And we also have a promotion, we call IP GO, which you can -- customers can freely download our PUFrt and to do their simulations, okay? So we do have a lot of requests on this. And another third IP is we call it the PUFiot. So PUFiot is PUFrt plus all the cryptographic functions. So the cryptographic functions -- because PUFrt is to generate the key, right, and the cryptographic is to use the mathematical method to -- by using the key generated by PUFrt to encrypt all the data or decrypt all the data and also do all other signature functions, authentication functions. So that's our third PUF-related IP. And of course, it's -- we just recently completed this IP, and it's already have customers interested in it. And another one will be -- we call it the PUF SE, is a PUF secure element, which is under development. So I think we will have various of customers. And I think the requirement from them is basically based on our -- their interest mainly on our product, but they would like to also help us to give them the solutions. So actually, in the future, we are not the only IP providers, we're also the security solution providers.

Another question related to the PUF. It's regarding what -- how will the revenue model work for NeoPUF-related new IP.

Okay. Regarding the revenue model?

So we -- basically just additional revenue stream. For example, one customer, we license the new IP, like the PUFrt, which we will have 2 royalty -- okay, one royalty through the [ Ench ] company, which is like [ royalties ] through the chip company. And when they go to production, we will have the royalty from the fab, which is the foundry. So this creates additional royalty stream for [ NeoPUF ].

[Operator Instructions] As there are no further questions, we will now begin closing comments. Dr. Hsu, please proceed.

Okay. So with 20 years of innovation, eMemory continues to develop the foundational semiconductor technology, which help our customers to speed up their technology development and improve their chip performance. We hope to use our innovation to ensure all online activities are secure. And this is our company's vision: embedded everywhere and secure everywhere. Once again, thank you for your patience and the support for eMemory. Thank you.

Thank you, ladies and gentlemen. Thank you for joining us today. You may now disconnect. Goodbye, and have a good day.