International Marketing and Staying Ahead of TechnologyInternational Marketing and Staying Ahead of Technology

International Marketing and Staying Ahead of Technology Transcript:

Announcer: You’re listening to The Kula Ring, a podcast made for manufacturing marketers. Here are Carman Pirie and Jeff White.

Jeff White: Welcome to The Kula Ring, a podcast for manufacturing marketers brought to you by Kula Partners. My name is Jeff White and joining me today is Carman Pirie. Carman, how are you doing, sir? 

Carman Pirie: I’m doing well, you know, and per normal, I’m excited for today’s show. But, you know, for a different reason this time. What I think is interesting about, like, never let it be said that The Kula Ring doesn’t try to follow the trends, you know, and AI is on the tip of every marketer’s tongue and you could probably I don’t know if you printed out all the blog posts or podcast transcripts or whatever written about how it’s going to impact marketing. You could probably ride that trail to the moon and back, you know. But we’re not going to talk about any of that today. So we’re like.

Jeff White: No but we’re talking about AI.

Carman Pirie: Right? It’s kind of fun anyway.

Jeff White: Yeah, no more from a manufacturing lens. 

Carman Pirie: Yeah. So let’s get into it.

Jeff White: Great. So joining us today is Abby Guo. Abby is the marketing director at Graphcore. Welcome to The Kula Ring Abby. 

Abby Guo: Thank you, Jeff, thank you Carman for having me.

Carman Pirie: Abby, It’s awesome to have you on the show. I want to, let’s start by understanding Graphcore, and then I want to know how you ended up there. But first things first tell our audience What Graphcore is and does.

Abby Guo: Yeah, actually Graphcore is an AI semiconductor company. So now I think everybody is super familiar with like ChatGPT. ChatGPT started activity in December 2022 and then all of a sudden everyone’s like you know what is AI? So basically we build chips for our customers. So they put chips and our systems in their data center to facilitate, to train their models, like ChatGPT, be like maybe before ChatGT with all those traditional, slightly smaller our models and then after the training they can also deploy their models on our chips to do the inference.  For example, after they train the models like ChatGPT and then they deploy them so people can just log in, register and then start to use them for every request you send to the model and then they give you answers against the process inference. So our chips can do both training when you build the model and we can do also inference when users start to use a model on request. Yeah, that’s basically how the product works.

Jeff White: Yeah. So these are really specialized like computer graphics, not computer graphics chips, but computer chips that are used specifically for large language models and nothing else.

Abby Guo: Yeah. And it can be used for like, all the models considered now like AM models. So now like becomes only like I think the dominant models in the markets since 2022 and ChatGPT was launched so before actually along years there has to be a lot of different types of models people are using is just like people don’t I to maybe if you are not in this industry that a few people have heard about this before, for example, we support like larger level models, like equally larger models and also some CV models. For example, if you are driving a car and you see all these kinds of images on the road, right, you’ll see people walking by other cars. And then like behind all these there are also some models we call computer vision models. So we kind of like up the data from the images and then do computing from, you know, just on the edge of the car. So that’s also a different type of model called CV. Yeah, So there are actually a lot of different models doing different types of workloads. So people’s minds are occupied by large kinds of models right now.

Carman Pirie: Which I think is going to be part of the challenges as a marketer in this space. But how did you end there? Let’s talk about that first.

Abby Guo: So that’s very, every time I share this story like people were like ‘wow’ because Graphcore was funded in 2017. That’s the year when all the famous, now famous AI startups were founded like OpenAI also like there are other AI chip companies, startups in the U.S. also founded the same year so that’s why that’s 2016 we consider it as the year of startups. So actually we were first funded in Bath in England, then later on we moved our headquarters to Bristol, England like very big hub of AI semiconductors, which very few people know about. There you can see also other semiconductor companies have been there for years. I joined Graphcore in 2019 when we launched our first product. So because before you have a product this kind of like don’t really need a lot of our go to market right and then in 2019 when we had our first generation of chips our product I joined them and start to lead the Asia Pacific market. Yeah, and it’s funny because back then we were quite well-known in the Western media. They all say, Oh, Graphcore was the only rival to Nvidia and Graphcore is the only AI chip unicorn in Europe because all the rest are in the US. Yeah. And so we were high profile in the Western world but like, back in Asia actually not like very few people heard about it before. That’s why I joined and started to kind of like boot everything from scratch really, from nothing to really a kind of like an AI chip superstar in Asia to get to now.

Carman Pirie: And then you’ve moved since moved and basically taken over the US market as well.

Abby Guo: Oh yeah. I was also for the Asia market especially since China is a big market for us. So I’m leading right now the Asian Pacific market in China, Japan, Korea, Singapore and Southeast Asia market and a little bit of India and Australia. Yeah like the East Asia market is a big one and at the same time, I support my CEO and CMO, on the Indian and American markets, including Europe and the Middle East of these areas and also America market. Yeah.

Jeff White: You touched a bit on kind of starting around the or you know, kind of competing within Nvidia and I want to come back to that. But before we get to that, let’s talk a bit about what Graphcore’s customers are like, who are you selling to? It’s not sort of the typical, you know, it’s a B2B audience, but you’re not selling to other manufacturers necessarily or at all. You’re kind of targeting a very different group of people.

Abby Guo: Yeah. So right now because we are not as big as Nvidia, we are still like a unicorn, like a startup. So now we focus, mostly we focus on the customers in consumer Internet and also mostly on like application internet or software layer. So for example, order I can’t name them but like older tech, like companies, big tech companies you can think of, they are the type of customer we are focusing on. And also at the same time, because these customers, they’re big, right, they’re building applications, they’re building like e-commerce service. they are building an e-wallet or something like that. So they’re basically providing applications to end users and because they are big, they have their own data center. So for this type of customer, we just sell to their data center directly so they can use our chips in their data center. However some small and medium customers, don’t have the budget, and these are expensive to build. So they would just use the cloud, from public cloud companies in the market. So for these customers, we mainly sell them through working with cloud companies. So cloud companies, work as our partners to build cloud services based on hardware. Yeah. Also, some of our customers buy directly from our server OEMs, so because of our accelerator. So some of our server OEM partners like Dell, also have like they also sell together with our product as our integrated solution. So basically we have three different approaches to customers, for different types of customers.

Carman Pirie: Seems to be an incredibly dynamic environment, Abby. With a lot of different, you know, very kind of leading-edge customers that are, I would assume, innovating in very rapid cycles. And meanwhile kind of like, you know, can the silicon change that fast? You know, is that how does that talk to me about the pace of play there? How do you deal with that?

Abby Guo: That’s a very good question. Yeah, that’s actually what always depends on us because our customers develop products super quickly. You can see the how features all that. If you use the like e-commerce or social media or any type of application or website like you can see they just launch new features all the time. Right? But like, how do we keep up with this? our customers? Because we are a hardware company, right? Like we have to. If we need to do a new generation of product, we have to start from the silicon level where like if you start basically from the sand, you know, it’s about something that you use to build the intelligence for like in the machines. So it takes a really, really long time. I even personally think Graphcore already runs really, really fast on this, but still, it’s difficult to keep up with the innovation or updates on our part of applications. Right. So I think that’s kind of like the biggest challenge for our marketing team because how do you keep your company on the same page as and pace with your customers? And how do you always kind of like keep tracking what your customer wants and what your customer requires in the near future? Or in the further future. That’s really quite challenging. I think the marketing role isn’t like shouldn’t be limited, just like one way to bring the product from labs or factories to the hands of our customers. It should always go in both directions. So I think marketing teams really need to stay resolute with the whole team. Like, especially the team that’s doing maybe demand or customer-like marketing. While on custom marketing, you really need to stay very close to your customers and hear what are our customers developing. What is your part of the pipeline part product strategy for your customers? So you can bring this feedback or bring your insights of your customer back to your product team. So the product, when they become the product team, always designs future products so they can add these features and the customer may want to learn new products. So that’s how you kind of it’s kind of like dance of both sides, right? And then at the same time, I think it’s also your need to really, for example, here I want to give you an example because like before ChatGPT, I think no one’s noticed. Like I think people in the public only know about ChatGPT after their launch. Right. But actually, if you keep listening to your customers working really close to your to the product team of your customer and then you actually, you know, they started to do it like maybe two years before they launch to the larger models. So you need to know they are already doing that before they launch it. So you can just tell your product team, okay, maybe our customers are doing like our model, maybe it’s the trend maybe you should consider to support this. To be able to support this larger model for the customer. So that’s, that’s what I think. So like, you always need to like, plan ahead and always need to do the work beforehand before you launch a product or before you launch something.

Carman Pirie: And I can certainly appreciate that duality and needing to stay in touch with that customer base and understanding and anticipating their needs in addition to just selling to them. Do you pull on any other groups to try to even get ahead of the customers, if you will? You know, I’d be curious if you’re, you know, how you kind of think about the market maybe more broadly.

Abby Guo: Yeah. So actually, it’s so in my, you know, actually in Graphcore our engineering team, which is like a different team from the product team, right? They are quite strong. So we, Graphcore are quite famous in academia and research area because of the product we, the chip we produce. Our design is called IQ Intelligence Processing unit, right? So actually it is designed only for machine learning. And also our software called Poplar is also designed from scratch together with the chips to be delivered to a customer to use only for AI. So actually our approach is that we keep ahead of a customer. How do you keep ahead of our customers? We work with the top researchers and we work with the top universities like a university in the US, Europe and Asia too. We support their papers. We do the research and get together with the team. And so we got to know and also we are building the train together with the top researchers in the world. And normally it happens like this. So the innovation happens first in academia or in the research area, and then the commercial companies will follow. So the commercial companies are actually our customers. That’s how you stay ahead or how do you lead the voice or lead the trend in the market? So that you can like in this way, so you can also lead your customers.

Jeff White: Almost a Firma, kind of model where you’re working in conjunction with deeply with the education and researchers and, you know, the people who are truly kind of finding the absolute cutting edge of these things and then looking to commercialize it. And at that point, they already know your hardware. So they’re comfortable with it. They’re probably going to move ahead with that.

Abby Guo: Yeah. Yeah. And also a part of that is like because we really need to. So what we are saying, is it’s not like you sell TV or sell something like you can just you think there are to your customers house and they start to use. So actually we need to spend a lot of time and make a lot of effort to educate the developers. So we are a B2B company and we also need to have this mindset of B2D with like a business to developer. So because all the NGOs are actually developers in the companies, right? So we need to spend a lot of time tutoring how to use our product and at the same time make our product line, and especially software super easy to use and very mature. So I think by doing this we kind of are also allocating the whole market from firstly from the universities firstly from research labs and then gradually to commercial companies. So people will switch their habit from using maybe a later CUDA to other software or programs to do things. That’s a long-term effort. You really need to put years of work into really transferring or allocating your end-user developers to use your software instead of your competitors.

Carman Pirie: Abby, help me understand the role that, uh, the industry analysts, uh, maybe the investor community, VCs, etc. How are you? How are you leveraging that knowledge base? Or are you?

Abby Guo: Yeah, so because we still are in, it’s kind of like in the startup mode, like we will be here for like seven years. For eight years. Yeah. But like, we’re still a very young company, like as a hardware, especially from a company because all the big semiconductor companies you can see on the market like, like pretty old, like decades of years old already. So we have, we are relatively young still very young. So actually in this kind of slightly small team and also in the startup are all kind of like vibe. So all the like all the team is kind of like have multiple roles. So for me, it’s a marketing role. But what I do is like everything you consider from between the labs and customers, revenue is like, I cover all of this. So that’s actually the question is like, like really this a really good question. As you know, that allows me to share what kind of marketing role is in a start-up like this. So now we are still like a VC-backed company. So a lot of times, we need to consider the, if you call it investor relationships that we don’t really call that much. But like some sometimes you really need to influence the investors as well. And sometimes you also need to influence government officials as well. Because right now AI is becoming more and more like national security level technology. So all the governments are trying to put their hands on a lot of different areas in AI, especially hardware. So we need to kind of like first of all, we need to make sure everything we do is legal and then we need to make sure everything we do is supported by. Not necessarily supported by the government.  But like it will definitely go along with the policies and especially on the executive level, we need to be fully aligned with the government policymakers.

Carman Pirie: Yeah, that is an interesting dynamic I would think with the uh, emerging in the emerging regulatory framework that you’re working within. And yeah, that would be kind of interesting. It’s a double-edged sword because it’s coming at one,  on the one hand, there’s a level of control being applied to the industry, but on the other hand, there’s a dramatic level of funding being supplied by the American government to expand in the semiconductor space. So it giveth and it taketh away, I guess.

Abby Guo: Yeah, yeah. All of the government is like I think it’s trying very hard to kind of invest in semiconductors because basically it’s a brain for all the intelligence like, like machine intelligence or if you don’t build a really like powerful like computing brain for that for the applications I think the whole AI application layer platform there will be like push back a lot by hardware that’s why I think you can see now on that. Also, a lot of new startups are strong startups that have started to show up in the market and a lot of investors are looking for new start-ups. In AI chips to you know to invest in yeah that’s yeah I think on the whole industry as the whole VC and AI start-up kind of like industries like becoming really I think interesting and it’s changing very fast as well. It’s very, it’s very hot like a lot, a lot of stuff is happening right now.

Jeff White: I think a lot of people, you know, you mentioned the Nvidia example of a company that’s been quite successful in this and how you’re looking to compete with that. And, you know, potentially consumers may not necessarily know about the meteoric rise of NVIDIA. And I remember buying their gaming chips for computer games 20 years ago, but I don’t even know if they understood what was going to happen with that. But now they’ve been incredibly successful and, you know, kind of leading that charge somewhat. It makes it interesting for a startup like yours, even if you have been around seven or eight years to compete there. How do you go about that? And what are some of the strategies you’re employing for kind of going after a whale like Nvidia?

Abby Guo: Yeah, it’s like, um, I think, yeah. I have to say you have really like very good knowledge about like what’s happening here, how it comes from like gaming chips to how to AI chips, right? So that’s how we started. I like the way it just started because if you consider like AI workload is more like image processing rather than just like the traditional workload. So like then on the market, like just multi skills and GPUs like thing and then GPUs and actually they’re better than CPUs to like for AI workloads because they do all image processing. Right? And then but like so they still like not designed for AI workloads so it comes from image processing. Right, So I think of for young startups like us, like as a new company that wants to exist, like want to grow and thrive in this market, I want to challenge Nvidia. It’s like I think firstly we need to have a very unique position of the company and of your product, your core product especially. So you need to have like just one unique positioning for the product of your company. And also you need to have several different positions at different dimensions. For example, what I said because I mean they are like so our, our positioning for Graphcore IP is, is that we are designed especially for AI workloads and both our hardware and software. So that’s a unique position in that big competitor like Nvidia, they don’t have like they just naturally become like really good and dominant right now, right? But like at some point they are evolving as well. They are just making more and their hardware more and more powerful for AI workloads. That’s true. And also I think so because like at different maybe dimensions or different fields, you also need to have a very good positioning for your product. So firstly, if you like, in some fields, like for example, your competitors are so much better than you, they have been there for years and they are very good at this of use like a certain different type of tomorrow’s and you need to really have a good position over your product as well. Like how you can stand out in this field of like comparing to our competitors. You can always find out some points that you are better than, your competitors and you have to develop the whole message for this positioning. Let your customers let the market know. Secondly, like are also for some fields may be or in some models, my cases some models other customers use maybe some Nvidia GPUs and our Graphcore IP is kind of like you build the performance. And then for this in this field, you also have to like figure out how to position yourself. And so you can like tell the story or deliver messages to our customers to win at this type of field or models. In my case. So certainly like of course in you are like at some fields you are definitely better they or they are competitors’ products so you need to position yourself really well and to stand out, complete this because this is where you kind of like in on your customer and they start to trigger an interest in your product and then they can try to explore different models. Maybe you are less like you are like not as good as compared to what they would or they would also want to would willing would be willing to try. Yeah, since you already knock on the door of your customer.

Jeff White: Man, I’ll have to put a dollar in The Kula Ring swear jar by bringing up an apple comparison. But this has a real kind of Mac versus PC sort of feeling to it, where you control the hardware, you control the software, the experience, the brand that you’re bringing to market, which is pretty interesting. I think the way that you’re visually going to market is as well when compared to Nvidia. So I think you’re, you’re doing some interesting things in that way.

Carman Pirie: Basically you’re saying, Jeff, that you can be more focused. Um, yeah, for it to be more focused.

Abby Guo: Yeah, Yeah, definitely. I think because especially now we don’t really have a big team, right? We need to be really like focused on the point we intend to win over our competitors and then but that’s, like worked out like, even like you are really good at a certain like field, but you also need to be able to compete with your competitors in some areas and they are good at as well. So it’s kind of difficult. And also I think one another that’s really important is you really need to kind of like nurture your own community and especially for like I think now more and more people hear about this overall, the developer relations now it’s like a big, very big role in the marketing team, like to help develop, like to maintain relationships with developers because for instance, like in my case, because CUDA is like really strong, like a lot of my, a lot of the developers or users of chips, they actually they are so used to program in CUDA. Actually, the big challenge is how do you because they, they don’t care about they don’t really care about hardware. They program the hardware in a software environment. They don’t really need to see what the chips look like. Right? And then so you need to really build a community where you can keep your developers or users stay in the community and gradually you migrate your democracy, your, you know, competitors like community to yours and keep them staying in your own community. That’s how you make more and more like users and developers. So I think it’s like the final one all comes from these constant small wins over developers. Yeah, like people now realize that and they are trying to really make a lot of effort to kind of the community because this is how you stand it is how you flourish.

Carman Pirie: Yeah we can you know, we can talk about this for probably several hours in the interest of trying to bring it to uh to a bit of a close I’m curious Abby, have you had, you know, in a space that’s incredibly obviously very tech forward, the innovations that you’re bringing to market what you’re able to do, the features you’re able to enable, etc., you know, that’s all very important. You know, the innovation side of this is a critical component to being competitive. And, you know, often when you’re dealing with a developer, communities, etc., um, the, thought is that the focus is on the meat the some of the performance. Um, so I’d be curious as we bring this to a close if we, you know, where do you find the role is for the um you know, kind of the software side, the role of branding of uh, the creative presentation of the firm. Um, uh, does that have a role to play here? Are you finding that that kind of initiative takes the back seat to the more, uh, hard-nosed sales benefits, as it were?

Abby Guo: So I think on, the branding thing is like because it’s funny because there are two actually in this, especially in like really technical industry, like the AI semiconductor is like it, there are two different kinds of a mindset to do marketing. One must is like engineering a mindset like you’re going to see like Intel or the blue colour or the design of the website is like very engineering thinking. They literally just like do everything from like a more engineering mindset. And I think in our like what cause like grinding areas and some more like marketing oriented branding against it and we have like the pink colour like different like a really, really nice in bright colours and also that we all, we have our own font like typeface like that’s only like owned by Graphcore. So it’s kind of like the difference between Android phones and Apple. Like it’s two different ways. It’s like Apple phones are more like marketing and youth-oriented design. And then the Android phone is more like engineering, like the mindset, like orange is like. So I think that’s so firstly because now of all our customers like and the developer that are very young like is there, it’s like they’re super young on divide process. So, how do you approach them and make them start to have this, each has to really look at your website and remember you. So that’s actually also very important, especially I think, in, in an environment like, like now, I think all the kind of like the key developers, they, they just like they don’t want to order things, they want something so that like a little bit like you need a little bit like, you know, both. So that’s why we have won some awards for our branding? Also, we got really highly high comments from our all customers on all branding stuff. And then from that then you need to present like your part of and to kind of like, oh, technology caught them, we’re really well and all this whole branding guidance so people kind of from emotional point of view, they already like you and then they go really deep down from rationally they also buying the stuff, the technology you are delivering the product you are, you are selling. So I think this for both hands you need to be really like good at and then you need to keep your like also need to keep like updating or evolving so you can keep the customer stickiness all the time so they don’t leave you. So actually, like for this product, the customer journey is super long, like in each of about, like maybe anyone contacts with a customer, they may drop you. So all I think is as a marketer, we should always keep in mind how to keep our customers interested as first time and also stay with you all the time.

Jeff White: I will say I did notice the custom font as a type guy, so well done on that front. I love it thank you so much, Abby. It’s been wonderful to chat with you.

Abby Guo: Oh Thank you. Jeff and Carman.

Carman Pirie: All the best.

Announcer: Thanks for listening to The Kula Ring with Carman Pirie and Jeff White. Don’t miss a single manufacturing marketing insight. Subscribe now at kulapartners.com/thekularing that’s K-U-L-A Partners dot com slash the kula ring.