David Harold from Jon Peddie Research sat down with Imagination Technologies chief revenue officer at CES 2025 to ask about the shrinking mobile GPU market, the threat to automotive semiconductor incumbents from Nvidia and Qualcomm, and opportunities for the company in AI.
At CES 2025, we sat down with Imagination Technologies’ new chief revenue officer, Jake Kochnowicz, to talk about the health of the company and its plans in 2025. Imagination ended development of RISC-V CPUs in 2024 to focus on GPUs and AI. To that end, we wanted to explore what opportunities and challenges the company sees in its key markets.
Congratulations on your promotion to chief revenue officer (CRO). How was the change from your previous product role?
Kochnowicz: Stepping over and owning the whole thing end to end has certainly been interesting. It’s been good because I think there’s a lot of areas where trying to make it more coherent is something I’ve been able to do because I do know the products quite deeply. That helps me understand where there are particular challenges, and I’ve been able to make changes in how we operate, which has helped in terms of confidence and stability and what we’re saying in forecasting and all that sort of stuff. It’s definitely been a lot of work.
Everybody has a different view of the world picture today depending on where they are—there’s a lot of geopolitics still going on in the industry. What is the best market for you these days?
Kochnowicz: Geographically, I would say we’re still fairly evenly split across the world. It is pretty equitable between what we would consider the three main geos: US, China, and then everything else. Maybe recently a little bit more in the US, whereas I think this year we’ve been more positive on the APAC side, so it kind of varies.
What about the general challenge of trying to sell a mobile GPU? Because it feels like the market is quite thin for mobile GPU now. Where does it go?
Kochnowicz: The market for mobile is still there in the sense that there are new chips that are designed in every year. I would say many companies have either got their own GPU or have partnered up pretty closely and are going to continue with those partnerships. But we do still have a footprint in mobile.
You don’t have to say anything about it, but obviously from our point of view, it feels like it’s all leaked, that you are working with Google. Those are real devices with real benchmarks.
Kochnowicz: I can’t comment, but we do see mobile as still being meaningful for us. It’s definitely difficult because you’re in a world where every kind of chip manufacturer is basically either designing a custom GPU, and they own it, or they’re licensing. MediaTek is the dominant Mali shipper, and, therefore, the Mali development would be quite heavily driven by them. Qualcomm has Adreno, Samsung has AMD now. And they’re all kind of doing their own thing, so it makes it hard to get in there unless you do something a little bit special, which we are always looking at.
And then you look at the other markets: Automotive is definitely still the strong one for us. We continue to see traction there. We continue to see OEMs look at building their own chips. I would say they want something a bit different, a bit unique to them. The cookie-cutter Nvidia or Qualcomm product is just not quite it. So, they’re looking to do something more unique, and that for us as an IP company is an opportunity.
Does it feel like the Japanese and European big semis supplying to the automotive Tier 1s are feeling threatened by Qualcomm, Nvidia, MediaTek, etc.?
Kochnowicz: I do think that the incumbents feel threatened. But what I’ve witnessed is that we have gone through waves. Initially, it was only the traditional automotive chip players, then Nvidia came in with a bigger, higher performance thing, and it was the thing that the premium brands had to have. But within one generation, they quickly realized they don’t particularly want to be stuck to this particular wagon. So then Qualcomm came in, and they were more cost-effective and had leverage from mobile. And now I’d say we’ve got to a position where all the incumbents kind of retrench, kind of step back. I almost feel like we’re in the third wave.
All the OEMs realized we went from Nvidia, being stuck in that camp, and now we’re stuck in the Qualcomm camp—why am I in prison again? And so I would say [with] this generation, I’ve actually seen a genuine mission from incumbents to want to go and take up the fight because the OEMs have basically come and said to them, ‘We need something else.’
We’ve gone through an Nvidia wave, a Qualcomm wave, and now we’re getting to a more competitive state.
Who do you think is doing something good in the automotive segment?
Kochnowicz: Automotive is quite broad, but I would say you’ve got Renesas launching their R-Car Gen 5 platform recently, and that’s a beast. You know, it’s really, really big. I would say other companies are TI and NXP.
What do you think of Hailo?
Kochnowicz: There’s been a lack of out-of-the-box AI solutions. I think because it’s very hard to deliver, as we know from our experience delivering a standardized AI IP.
So, there’s actually not that many AI pure play IP companies now because the software burden is very, very hard. And so I think what’s happened is companies that haven’t already developed their own in-house are looking for partners. I think companies are looking for proven AI companies that have got scale, or at least have got something behind them. That aren’t just vaporware.
It’s all far from solved. That’s largely because the software keeps evolving. And the thing with hardware is, it’s fixed. And so there’s definitely a tension at the moment in the industry between those who are realizing that a fixed-function accelerator—which is really, really good at what it does, but not very good at what it can’t do—and then something more flexible, whether it’s a GPU or something else like a DSP, that getting the right mix of those particular elements is becoming more important because the future-proofing is an issue.
Do you see GPUs continuing to play a major role in AI, especially at the high end?
Kochnowicz: Yes, absolutely. GPUs remain central to many AI solutions. Interestingly, we’ve had customers ask for ‘headless’ GPUs—essentially GPUs without graphics capabilities—because they only need the compute functionality. However, creating headless GPUs isn’t as simple as it seems due to programming models, testing requirements, and how GPUs interact with memory and other systems.
What about software challenges in automotive AI?
Kochnowicz: Software is a big challenge. We’ve partnered with companies like CoreAVI, which specializes in safety-critical markets, to provide reliable driver-side solutions. Higher up the stack, things get more complex, especially when you factor in explainability and certification requirements for AI systems.
Some companies are adopting dual-processor setups—using one processor for primary tasks and another for redundancy or cross-checking. This approach avoids full-stack certification issues while maintaining safety.
Do you see opportunities in areas like Arm-based AI PCs or Windows on Arm?
Kochnowicz: It’s an area we’ve been investing in, particularly in DirectX support, which sets us apart from other IP providers. While there’s interest in this space, it hasn’t fully taken off yet. Many companies are waiting to see how Qualcomm performs before committing.
That said, we’re seeing interest from OEMs experimenting with smaller-scale projects, which could lead to broader adoption in the future.
How do you view the Arm-Qualcomm legal disputes?
Kochnowicz: The fallout from those disputes has made the broader semiconductor ecosystem wary of Arm’s behavior. For a smaller company like us, this presents opportunities to engage with companies exploring alternatives. While Arm remains a major player, their actions have prompted some to look around, opening doors for us.
Anything else you’d like to share?
Kochnowicz: We’re focusing heavily on innovation, particularly in GPUs, to offer customers flexibility and advanced capabilities. We’ve redeployed some of our best talent into GPU development, and early feedback has been very positive. We’re excited about what’s ahead and can’t wait to share it with the world.
LIKE WHAT YOU’RE READING? INTRODUCE US TO YOUR FRIENDS AND COLLEAGUES.