Timing is everything: how SiTime drives innovation in modern electronics
43:01 Share

Welcome to today's episode of Lexicon. I'm Christopher McFadden, contributing writer for Interesting Engineering. In this episode, we sit down with Piyush Sivalia, Executive Vice President of Marketing at SciTime, to explore how their cutting-edge MEMS-based precision tiling technology is revolutionizing modern electronics.

Piyush shares how SciTime solves the toughest challenges in the 10 billion timing industry, from AI data centers to electric vehicles and wearables. Before getting into today's episode, here's something to elevate your 2025.

Level up your knowledge with IE Plus. Subscribe today to access exclusive premium articles enriched with expert insights and enjoy members-only technical newsletters designed to keep you ahead in technology and science. Subscribe now. Now let's continue with today's episode. Piyush, thanks for joining us. How are you today? I'm very well, thank you. Pleasure to be here. Our pleasure. And for our audience's benefit, can you tell us a little bit about yourself, please?

Sure. My name is Piyush Savalia. I'm the EVP of marketing at Sidetime. I have responsibility for all aspects of marketing, but also responsibility for the VUs that we have. And they're reporting into me. So we have an aerospace defense VU, a comms enterprise data center VU, mobile IoT consumer industrial VU, and finally, basically an automotive VU. So

responsible for that too. I've been in the industry for about 34 years now, pretty much a semi guy all the way through. Started as an apps engineer, applications engineer, moved over to the dark side, as they call it, marketing about 30 years ago and loved it so much that I've been there. And I

Quite unusual for Silicon Valley. This is only my third company in my 34-year career. I kind of like to go in early and then help build companies. So I came to Sightime in 2008 when we were almost a pre-revenue company. I think we had shipped a little bit. And here you go. That's where we are right now. We're going to be doing...

We haven't announced yet, so in the range of give or take $200 million is what the financial analysts have us at last year. With a market cap of close to $5 to $6 billion, clearly we're being valued for growth, and so that's my job. How do we build the customer relationships? How do we get the products out that basically help us drive the growth and innovation forward?

Excellent. Very varied career then. Excellent. Going to the first question then. What's the key innovation behind SciTime's new timing solution and how does it enhance AI workloads in data centers? So let's step back for a moment and let's talk about timing and what innovation we have there and then we'll move to data centers if you don't mind. So all modern electronics needs timing as a reference signal.

Literally, timing is considered the heartbeat of electronics. If the timing signal isn't there, electronics does not go. And just a reference for timing, when you buy a PC and it says the processor is running at 1.8 gigahertz or something like that, well, that's your timing signal that's going into the processor. And basically, the faster the speed, the higher the frequency of timing, usually the faster the system. Usually.

And so timing plays a very critical role in electronics. There have been numerous instances where timing got interrupted or something bad happened to timing and the whole system stopped working. So anyway, that's what timing does. For the last 70, 80 years, the technology for timing has been quartz crystals.

So literally, you take a quartz crystal, used to be natural previously, now it's all artificially grown. You cut it at different angles to get different frequencies out of the device, and then you marry it to an electronic circuit. So you take the mechanical resonant vibrations of the quartz crystal, and you convert it to electrical energy through a circuit, an analog circuit, and that's an oscillator circuit.

And so the ports industry has done a pretty good job of actually delivering these devices to the timing industry. I mean, you pretty much open most electronics today and it will have multiple timing devices in there. We estimate it's about a $10 billion market today. We think it's going to grow to about $20 billion over the next half decade to decade.

And we are focused on a particular aspect of this $10 billion mark, which is what we call precision timing. So the reason for precision timing's existence is that today's electronics is different than the electronics of the past. It's a lot faster. It's basically always connected. And it's a lot more intelligent, meaning you're pushing the intelligence closer and closer to the customer.

Alexa devices and Siri and all these devices, that's one form of intelligence. But there's a lot of other things that happen in these devices that's happening in the background that you don't even know about, but it's there and it's making certain decisions based on this. So now in this context, for a device to be always connected, to be always working, to have the processing power for this intelligence, to have the

the communication, when you communicate data, decisions, whatever to the network, you got to have a very good and clean timing device, actually a clean timing signal. And that has to come from a device. Now, here's the other interesting part, which has changed in the last three, four, five years. This electronics is getting subject to harsher and harsher conditions. So let's give a few examples. Think about your cell phone.

5, 7, 10 years ago, there was a cellular protocol in there and there was a Wi-Fi signal in there. Then we added Bluetooth. Then we added NFC for contactless payments. Then we added radar for various kinds of functions in some phones. You've got now at least five different wireless protocols inside a cell phone that are going to cause interference with each other. A timing device has to operate and deliver a clean signal in the presence of these.

Another example is a GPS signal. We've heard instances of how the GPS signal is spoofed or it's jammed and that causes all kinds of issues. Airplanes being off course and things like that. But what if you imagined a world where the GPS signal

remained true in spite of all these disturbers, whether artificial or natural, or whether man-made disturbers or natural disturbers. And if that was the case, you would always have a good GPS signal. Therefore, you could always know where you were located.

A third example is basically you look at a car. The amount of electronics in a car has gone up dramatically as you get to ADAS level 2, level 2.5 support today, and you strive towards ADAS level 4 or level 5 in the next half decade to a decade.

And, but a car is not a pristine environment. I mean, if you think of the PCs from 20 years ago, they were sitting in air-conditioned offices with a constant temperature on someone's gas. Not the same anymore, right? I mean, you've got that power in a cell phone and you're carrying it around with you and it's going to drop, it's going to do stuff. Same thing with a car. It's moving. It's going through different temperatures. I mean, it's running hot in the engine. It could be in the Arctic and it could still be 125 degrees inside the engine.

or basically you're going through a bad road and there's a lot of vibrations going on and shock. In all of these environments, the timing has to operate reliably because if it doesn't, you're going to lose whatever processing and connectivity you have in the car. And so that's the backdrop for what's changing in the electronics industry. And that's where we come in. Our core technology is

allows us to create our devices to be much more environmentally resilient than the porch devices of the past. And we're talking orders of magnitude more resilient. And so, for example, a very common example, suppose there's a timing device. There's multiple timing devices in a base station, but suppose there's a timing device that's driving the cellular connectivity, and the base station is mounted on top of an electricity pole that's next to a railway line.

Freight train goes by, that electricity pole is vibrating, that base station is vibrating, that vibration is going to couple into the timing device because ultimately the heart of the timing device is a mechanical vibrating resonant element. And so it's going to couple and it's going to cause problems with the timing signal. We eliminate that.

So whether it is vibrations, whether it is shock, whether it is airflow, which causes rapid changes in temperatures, whether it is electrical noise coming from multiple protocols or things like that, our devices are more immune and more resilient to these disturbers than anything else. And that plays in really well with today's electronics.

So that's what we do. How do we do it? So there's three components to our device. One is what we call the MEMS resonator. Resonator is basically a mechanical element that vibrates at a resonant frequency. The mechanical vibrations of this resonator are converted to electrical energy, a clock signal by means of an analog circuit called an oscillator.

We design our own MEMS resonators. We design our own oscillators. That's different than the codes industry where they design their own codes resonators, but typically by the analog from semiconductor companies. Because we design the two together in-house, there's a lot of co-optimization that happens. And when you put those two devices in a package to create an oscillator, that co-optimization, we believe, helps us make 1 plus 1 equals 3.

we get some value add from the co-optimization so that we can deliver better things to the customer. At the heart of the resonator, the benefit of our resonators is that they're extremely tiny. Our MEMS resonator is 0.5 by 0.5 millimeters in size. The mass of our MEMS resonator, and compare that to a quartz crystal, the smallest quartz crystal you'll see out there is basically...

1 by 0.8 millimeter, so basically 0.8 millimeter square versus our 0.25 millimeter square. And the mainstream in quartz is down in the 1.6 by 1, so 1.6 millimeter square size. So you see the size difference, and what that translates into is a difference in the mass of the device.

Classic physics forces mass times acceleration. Whatever the acceleration is, whether it's vibration, whether it's shock, whatever that acceleration factor is, if that's constant, the lower the mass, the less energy couples on your device. That's the benefit that our MEMS resonators give. They're much smaller in mass, so less energy couples. That's how we are able to achieve the mechanical resilience. The electrical resilience we are able to achieve by a very careful design of the analog devices.

We've got some of the best analog designers in the world, and these guys basically make sure that our devices are immune to all these outside disturbance. So that's the heart of the technology that we put in our devices that allows us to deliver what we deliver and change the world of timing. And our focus really is on solving customers' problems with innovation.

Customers themselves don't know what kind of problems they're going to face in this new world of intelligent connected electronics.

And our job is to anticipate to the extent we can and solve them. And not just that, but if a customer runs into a problem halfway through the design cycle, we are there for them because our devices can be configured very easily. They're programmable. And so we are there for them. We can do all kinds of stuff in our devices to give them a solution that will work and not delay their time to rev it.

So that's what we do. And that's the heart of the technology that we build. Lots more behind it. I mean, we spend the most in R&D of any timing company. I mean, I think we are like at least two to three times more spend in R&D than anybody else, even companies bigger than us, because we are so focused on the innovation, so focused on solving the business problem. And

I mean, I've had many customers tell me, "Man, it's just a timing device." I'm like, "Yeah, but you see the amount of technology that goes in there and a little bit of black magic that goes in there because all this analog stuff is not easy to solve." And so there's a lot of technology that we ship into our devices. Fantastic. It always surprises me when I hear how analog tech is still required, even in highly sophisticated devices. Incredible.

That's just how it is. You're operating in the real world, so it's not the digital world. It's the real world. So you have to have some kind of

I learned that a long time ago, debugging systems in the lab. You go to an analog engineer who's been around the block a few times, and they'll give you the best tips. They'll tell you how to solve your problems much, much faster than anybody else. Yeah, I bet. That brings us on to the next question. How does the MEMS tech differ to existing quartz-based solutions? So...

Many differences, but at the heart of it, the MEMS is built with silicon. Now, silicon is the most abundant. I mean, quartz is quartz crystal. It's silicon dioxide. So the MEMS is built in silicon. The silicon industry, I mean, to me, and I've been in it for 34 years, anytime it touches an incumbent technology, it always wins.

You think about it. I mean, we used to have vacuum cubes that moved to transistors that moved to ICs, and now it's all ICs. I mean, yeah, there are vacuum cubes in specialized applications. For the most part, the bulk of the industry is ICs. You look at gyroscopes, mechanical devices. When a MEMS gyroscope came out, suddenly the market just expanded. And now you have MEMS gyroscopes in your phones, in cars, and a bunch of different places.

So why does silicon take over and become the incumbent and displace the previous incumbent technology? It's because silicon gives you more features. It gives you higher performance. It gives you smaller size. It gives you better reliability, lower power, potentially lower prices. So all these different things are encapsulated in the silicon industry. That's why it's the hottest industry on the planet today. And so that's what we do. We build it with silicon.

So what does that mean? I mean, we do batch processing. When we build a batch of men's wafers, we basically have like 25 wafers, each wafer having 100,000 dye on there that's built all together and processed together.

in a silicon pap versus the quartz side where it's a different manufacturing process. What you have to do there is you have to cut the quartz crystal at the right angle to get your frequency that you want. And then this device has to be optimized. There's a lot of sputtering that goes on. There's a lot of machining that goes on on this device. And of course, the quartz guys have got it into an art form today because they're shipping like 30, 40 billion units of this every year.

But still, it's a very mechanically intensive process. Lapping, machining, etching, etc. Versus in silicon, it's all done in batches. So what that does at the end device level, you get a lot more reliability on a silicon. You get a lot more predictability on a silicon.

And so that's how it's fundamentally different. What does that translate into in terms of customer benefits? Our devices usually have a lot more features to customize than ports. Our devices are orders of magnitude more environmentally resilient, mechanical, electrical, all those compared to ports. Our devices basically deliver higher performance under all of these environmentally stressful conditions.

Our devices are in a way more easy and flexible to manufacture. So if customers suddenly have an uptick in demand, it's easier for us to meet their demand because it's a matter of starting more wafers, expediting them through the fabs, expediting the packaging, things like that. Versus in the case of ports, you might actually have to install a whole new factory.

to do this. We use the existing semiconductor infrastructure. Our wafers are manufactured at TSMC in Taiwan. Our MEMS wafer, the analog wafers at TSMC, the MEMS wafers are actually manufactured at Bosch in Germany. Our packaging is basically standard plastic packaging for the most part that is used by the entire company

half a trillion dollar semiconductor industry at places like AAC used by basically everybody in semiconductors at places like Arsene at places like HANA

things like that. So that's what we do that is different than quartz, and that's how we are transforming the industry. Okay, then. So it's not quite what you're saying, but with quartz, although they put it down to a finite cutting the actual crystal, it kind of has to be adapted around the crystal a little bit because it's a bit more manual work into integrating with the crystal itself. My understanding, Barry? Yes. All right. Yes.

And that manual work, if you think about it, that manual work results in issues. See, reliability is a statistical measure. When we've looked at our reliability, we're talking about a billion hours in TBF, meantime between failure versus the false guys who are at 20, 30, 40 million hours. And you may say...

Man, it's millions of hours. I mean, who cares? No electronics last that long. Well, the problem is it's a statistical value. You have to take every component in the electronics device and basically calculate statistically how reliable it is. So for example, our devices, over 10,000 units, you'll not see a pod fail with quads. You might see four to five pods fail. Right.

I didn't realize that. That's interesting to know. Building on that then, so what have you found are the primary challenges faced in developing precision timing solutions, especially for applications like artificial intelligence, Internet of Things, and electrical vehicles? So I think...

The key is innovation, right? I mean, we have to innovate very, very quickly. You look at the data center industry, and that's a couple of hundred billion in size today, and they're talking about a 40% growth rate on this 200 billion size. And you dig a little deeper. I mean, the mainstream Ethernet connectivity inside a data center was 400G Ethernet a couple of years ago. It's 800G Ethernet today.

It's going to be 1.6 terabits per second Ethernet a couple of years from now. So the pace is incredibly fast. Now, to go from 400G to 800G, your timing has to be twice better. To go from 800G to 1.60, your timing has to be twice better. And we've got two years to deliver this. So the rate of innovation has to be very, very quick.

And I think that's one of the key challenges that we overcome. A second key challenge is that customers are learning about this industry as they are building and deploying. So the problems are unanticipated. And so how do you respond to problems real time when they are not anticipated right at the beginning? You build a lot of flexibility into the device. So an example, I mean, Ethernet, the frequency for Ethernet is typically 156.25 gigahertz.

However, because of the way the designs are done, about 10 to 15% of the designs will experience bitter rates at this frequency. And you have to modify the frequency to 156.25391. Yeah, exactly. I mean, the classic, it's like, seriously, I mean, it's like,

Q5 versus 25391, but that's what decreases the bit error rates. So what did we do? We built in programmability into our devices. So you don't have to build a whole new MEMS resonator right from the beginning to handle this. What you do is you basically just program the device differently. It's like a 10-second jump.

To program insert 156.25, you deliver 156.25391. We had a customer who called us on Friday evening at 5 o'clock saying, guys, we're having problems. We've got to shift this product next week. We're having problems with this frequency. Can you program us some devices and see if the 156.25391 works? Friday evening at 6 o'clock, we got them the samples. They were down the road.

And Tuesday, they call back and say, okay, device works. You're in production. We're going to give you orders for 100,000 units like in a week. That flexibility, I mean, it applies to a smaller set of customers, but that flexibility is huge. And not only is it beneficial during prototyping, but it's also beneficial during these production runs. And so pretty much all our customers take advantage of the flexibility during prototyping.

And then about 10% to 15% of the customers who are in production run into your issues, and they want to solve problems, and we help them there too. So that's another thing that we build into our devices because the problems are unanticipated. So coming back to that, the rate of innovation is one big challenge. What flexibility do you build into the device is another big challenge. At the market level, we've been around for 20 years. Quartz has been around for 100 years.

Getting customers to understand that we offer them benefits, especially when they are risk-averse, especially when they don't want to try new stuff because they're already doing a whole bunch of new stuff on the AI processor and this and that. That's a challenge. And so we have to do a lot of evangelizing, give customers data so that they can actually decide for themselves which solution is better. So on the market side, that's a challenge is how do you get out to these customers, all these customers?

There's 100 AI startups out there in the world. How do we educate all of them? Actually, let me be more specific. There's 100 AI processor startups out there in the world. How do we get to them? How do we educate them on the benefits that we offer? How do we show them that by using us, you will be able to cut your time to revenue or meet your time to revenue versus unanticipated problems that might come up with the other devices?

These are the problems that we are solving. We solve them. That's our reason for existence. So we've got a team of people, engineering, business, et cetera. We get into it and we go solve it and we've done pretty well. I mean, our revenue last year, like I told you, will grow significantly over the year before. Even at the financial analyst expectations of 200 million, we were 144 million the year before. That's not significant growth.

And so that's what we do. That's how we solve problems. By solving problems, we take it benefits. I time it benefits the world. Absolutely. And word of mouth from existing clients will help as part of your marketing as well, wouldn't it?

Sorry, you're dropping in and out. So can you repeat that question? Yeah, word of mouth from your existing customers as well. Absolutely. That plays a huge role. That plays a huge role. And we've also found that basically customers, once they use SightM, generally they don't go back to course. Because they see all the benefits and then the word spreads inside a customer. We've got some customers today, large customers today, who we first shipped revenue to them in 2009.

Wow. Okay. They've had like annual design upgrades since 2009. And we are on every one of those designs. Superb. Dream. That's the dream. Right. You've answered the next few questions. I'll skip over them. The next one. How does SciTime's approach to precision timing contribute to sustainability in industries like automotive industry and data centers?

It's a great question. And I don't, I mean, I have an answer today, which is just beginning to scratch the surface of this. Right. Okay. There's so much more work to be done out there. And there's so much more learning that we get and there's just work to be done. Sustainability. Let's look at manufacturing. All our devices are RoHS compliant.

And whatever the industry standards are around elimination of potentially dangerous materials, we are at the leading edge of that. We absolutely adopt those things as quickly as we can. Our suppliers do, for that matter, and we drive our suppliers to adopt them as quickly as possible. So there's that part of it, which to an extent, just being ahead of the curve is a good thing.

So that's one. The other part of it is basically how we help our customers do things differently that helps the environment. So one is, for example, power reduction. I mean, data centers run, obviously they run, they consume a lot of power. And so how do we get lower power into that system? And again, we're just beginning to scratch the surface here. We think we have a long way to go there.

there's a lot of different benefits. Because so far, everybody's just been in the mindset of let's just deploy and then we'll figure out all these things later. In power consumption, I mean, we just introduced a device last week which combines the function of two different devices into a single device. Obviously, it consumes less power because you're not consuming the power of two devices. It's just a single device. That's one way we help. There's another thing that we do, which is that...

In a data center, you have a bunch of different applications, whether it is the processors, the GPUs, the CPUs, whether it is the switches, whether it's the NIC cards, whether it's the active electrical and optical cables, whatever it is, all of these have to be time synchronized. Meaning, actually not all of them, the key parts of them have to be time synchronized. So basically the NIC card has to be time synchronized with the GPU. What does that mean? It means that basically...

In time, what the CPU thinks of time versus what the NIC card thinks of time, there should be a very small error between the two. It's kind of like you go into your kitchen and you look at the microwave, you look at the oven, and you look at your hand, and guaranteed those three times are different. Guaranteed. That's not time synchronized. And in fact, the microwave and the oven are probably using bad timing components, which is why they're never on the same time as your watch or your phone.

So there's a protocol that is run on these devices and these high-performance networks that makes sure that the difference between these devices in a matter of time is of the order of nanoseconds. Why is that important? Because a training task, basically, you have to break up the training task to go get it processed by multiple GPUs, and then you have to reassemble it at the end of it. So it's a massive parallel processing effort.

How you break it up, how you route it to each GPU, how you assemble it is very heavily dependent on the time synchronization between all these devices. Today, I mean, there was a study done that said that GPUs are idle 50% of the time, up to 50% of the time, because the network is not fast enough and the network is not synchronized well enough to get data to the GPU, to get data optimally to the GPU.

Now, if we could play a role in increasing the utilization rates of the GPUs, that's a benefit. And that helps in terms of consuming lower power. Again, like I said, a lot more work to be done here, but that's a benefit because we can more efficiently... For the same training task...

you're using less GPU time because you're basically making sure that it's up more of the time. It's not either. And you're doing that by routing the package to it in a timely manner and inefficiently. Of course, there are other things that play a role, but this is just the timing part. So therefore, you're getting power efficiency out of it. That's one example. Another example is in the reliability part of it. So for example...

A cloud service provider has publicly stated that they lose, for every 24 hours of an AI cluster being down, they lose like $2.4 million of gravity. A ton of money. So now, basically, but they have to factor that into their business models, right? So they might end up deploying more AI clusters. So if we could help them with reliability...

Maybe they need to deploy less AI clusters. Maybe. It depends on what their growth rates are and what their capexes are and all that stuff. But that's another place where we could potentially help by increasing the reliability of these systems at the uptime so that basically the resources are not wasted. I guess that's the best way of putting it. So these are some of the ways in which we help.

But again, like I said, we're just beginning to scratch the tip of the iceberg. So I'm using multiple metaphors here, but we're just beginning to start here. Absolutely. And especially the growth of things like cryptocurrencies and blockchain, that's going to be really important too, isn't it? Yeah. Absolutely. And it's going to run on all these systems. And so how do we impact that? Like I said, we have to go figure all that out. Okay. Excellent. Yeah.

Can you share us some examples of how CyTimes timing chips are used in everyday devices or systems that consumers might interact with, if any? Absolutely. Is it a secret? Absolutely, not a problem. I won't name names, but I'll name applications. Okay. So for example, wearables. I mean, everybody is wearing smart watches or carrying some kind of wearable to do some health monitoring. We are in many of those devices. Okay.

We are in some of the most accurate of those devices because we offer a timing signal that is very accurate. So that's one example. Base stations and small cells that are used for distributing cellular signals, we are in those devices. We are in industrial farming equipment. For example, automated industrial farming equipment

that basically uses a GPS signal to run on its own and do whatever agriculture work it needs to do. We are in those. We are in cars. We actually are in many, many electric vehicles. In multiple instances of our devices, in electrical vehicles, things like, I mean, the cameras,

the surround view cameras, the ARES cameras, the computers that sit in there, the Ethernet connectivity that's in there. We are in those kind of applications. We are in sensors. So you have some sensors that are sitting. One of the most interesting ones, not very high volume, but one of the most interesting ones was we were in a sensor that's sitting at the bottom of the Mariana Trench to detect earthquakes.

So, I mean, there's an early earthquake detection system out there in Japan. We have been, I think we still are, in some of the sensors that sit on the ocean floor. Now, they don't replace those very often, so there's not much volume there, of course, but it's a really interesting application.

We are in a bunch of industrial sensors. We are in motor controls, servo motors that are used everywhere. We are in motor control of that. On the networking side, I talked about base stations and small cells, routers, switches. Pretty much every application in an AI data center has multiple instances of site and devices in there.

I mean, we've talked about or we've done the math. We have several hundred dollars of content into an AI data center rack, things like that. So, I mean, the gamut of applications, I mean, in aerospace defense, GPS equipment, communications equipment, I mean, vehicle mounted equipment, things like that. I mean,

Literally, I mean, we have a list of about 300 applications that we are in, and that list keeps growing. So from the bottom of the sea to outer space, basically, everywhere, in between. Everywhere. That's a nice way of putting it. If you don't mind, I'm going to use that. Check. Invoice in the post. There you go. Fantastic.

It kind of brings on to this. You kind of answered this as well, but I'll ask anyway. How does SciTile ensure the scalability and adaptability of its solutions across different high-growth markets, such as networking infrastructure and personal mobility? So as the industry moves faster and faster, forecast predictability is usually compromised. Fair enough. I mean, people just want stuff that they didn't think they needed, and they just want it.

And so we routinely get this. I mean, we routinely have customers coming in saying, okay, I had a forecast for this, but I need this now. Can you ship it? And so that's where this whole semiconductor infrastructure helps us. I'll give you a very vivid example. When we started in 2020, when we started the year 2021, our forecast was X. By the, when we finished 2021,

I think our forecast went up. Our actual revenue went up to 1.3, 1.4, 1.5x, something like that. So within the year, we had to ship 30, 40, 50 percent. I forget the exact number, more than what we had started the year out with. Wow.

And that was obviously a time when there was a lot of demand for semiconductors. It was just beginning to ramp up and all that. And it was multiple things playing out. How do we do this? We do this because we use the fabulous semiconductor infrastructure. So we use the same infrastructure that's already in place for a half a trillion dollar semiconductor industry.

And so for us, it's a matter of starting more wafers. For us, it's a matter of, and of course expediting them, and then working with our partners to cut the cycle times. That's what we do. And that's fundamentally different than the quartz industry where there are captive factories which are basically machining these quartz devices and then packaging them. And so we've been able to meet these upsides

we believe, a lot better than our competitors have been able to. Because of the nature of the fabulous infrastructure, where we need more wafers, we start more wafers at TSMC or Bosch. Yes, there's a cycle time. Of course, there's always a cycle time. We need more packages. We just basically go and assemble more at our packaging houses. And that flexibility of the fabulous infrastructure has allowed us to scale

as needed. And when 2023 was a down year, down as needed. Okay. Sorry, I'm writing notes in trying to catch up. Excellent. And that brings us nicely to the last question then. So what are Sidetime's goals for 2025 and beyond in terms of predicted market share, innovation, and impact across industries? If you're able to answer. So cannot talk numbers, obviously, because we are a public company. But

See, we want to solve difficult timing problems for the customers. And sometimes even they don't know what they are. So when they find out, how do we respond to it quickly? It's all part of our gene pool, so to speak. I mean, it's all part of our DNA. But how do we go about solving that? And so it starts with innovation. We got to innovate. And in our case, not only is it

What everybody knows is engineering innovation. How do you deliver better products faster and all that stuff? There's also business innovation. How do you talk to customers about re-empting some of the problems that they might have or at least educating them on that so that they are ready for it? How do you go to market? And then, of course, for all of this to work inside the company, we have to have cultural innovation.

We basically got to figure out a way to serve the customer, the end goal of solving difficult problems for customers. And in a team that is going to grow, how do you make sure that new members feel part of the team very, very quickly? So there's cultural innovation. So we look at innovation as being the heart of the company.

and multiple facets to it, the engineering, the business, and the cultural part of it. And that's, I think, going to help. That plays a very key role in our future growth trajectory. So obviously, over time, we'll introduce new products. We'll introduce products

new applications where we may have made a difference and we'll talk about that. Things like that. I mean, that's basically what we are trying to do. Ultimately, it's all in the service of the customer in terms of how do we solve the problems. Fantastic. Best of luck to you. Put your hands forward as anything. Like I said, that's all of our questions. Is there anything else you'd like to add that you think is important we haven't touched on?

No, I think we've touched on most things. One thing, I mean, engineers always like more information. Go to our website. Lots of information out there. We've tried to make our website as informative as possible, so go there. If you want to buy devices or prototypes, samples, things like that, they're available through our e-commerce platform on the website. So we can do that. Some of the newer parts are not there yet, but we've got a lot of parts in production which are already there.

models, application nodes, other kinds of collateral. It's all there. So go visit the website, go check it out and keep SciTime in mind for any kind of timing requirements that you have. Excellent. We'll share any links in our copy, basically. It goes live. But other than that, thank you for your time, Piyush. That was very, very interesting. Thank you. Thank you for having me on. Our pleasure.

Also, don't forget to subscribe to IE Plus for premium insights and exclusive content.