Latest / Technology Now / How could AI revolutionize supply chain?

Transcript

How could AI revolutionize supply chain? · Technology Now

• MARK BAKKER First of all I had three double espressos followed by a mini croissant, a little muffin and that was about it. And then after that a couple more espressos. But that's the usual, that's nothing special. MICHAEL BIRD That was Mark Backer who's our guest for this week, Aubrey. He was telling me what he had for breakfast as you can probably guess. Quite a lot of caffeine. AUBREY LOVELL Oh my goodness. Was he buzzing from all the caffeine? MICHAEL BIRD not quite, but it did get me thinking, where did all his breakfast come from? And I don't mean, you know, the cafe down the road. I'm talking about the food itself, the ingredients, the everything. AUBREY LOVELL it is good question, right? I think we take for granted a lot when we have food, whether it's grocery store food or going out to dinner, restaurants, et cetera. You we don't really think about the sourcing of that food. MICHAEL BIRD Yeah, it's pretty amazing, isn't it? So many things need to come together, especially for Mark's croissant; flour from wheat, butter from cows, which needs feeding, sugar, salt, yeast. You can tell that our producer, Harry, is a cook himself. Now, and the different supply chains which have to come together for it are really quite remarkable, aren't they? AUBREY LOVELL It is remarkable and I'm going to hold you to it, producer Harry. I would like some croissants if they can be shipped to my house. But obviously we're talking about supply chains today, right? MICHAEL BIRD Yes. Yes. We are talking about supply chains. I’m Michael Bird MICHAEL BIRD It’s pretty amazing isn’t it. So many things need to come together – especially for Mark’s croissant. The coffee is one ingredient, but the croissant needs flour from wheat, butter from cows which need feeding, sugar, salt yeast. The different supply chains which have to come together for it are really quite remarkable AUBREY LOVELL So we’re talking supply chains today? MICHAEL BIRD We’re talking supply chains. I’m Michael Bird AUBREY LOVELL I'm Aubrey Lovell MICHAEL BIRD And welcome to Technology Now from HPE. MICHAEL BIRD Right Aubrey, so as we talked at the top of the show, supply chains are pretty complicated. Everything is so interwoven into our modern society that it's almost more of a web than a chain. AUBREY LOVELL And of course, when it comes to technology, the supply chain really is particularly challenging, right? Because it includes the semiconductor industry and having to work around the literal laws of physics when building microchips. MICHAEL BIRD Yeah, absolutely. And when you think about the supply chain of technology, you have to consider the supply of the silicon for wafers, rare earth elements for semiconductors, plastics, ceramics, metals for wiring. And on top of all of that, you have to supply the people who are doing the actual work. Now, according to journalist Edward Humes, a smartphone could have a transportation footprint, that's the footprint of all of the components combined of well over 240,000 miles or 386,000 kilometers before it even reaches the customer. That's the distance to the moon and even a significant part of the return trip too. AUBREY LOVELL REACTION That's crazy to think about. MICHAEL BIRD Yeah. Yeah. Now, like it has in almost every other industry, AI is disrupting the world of supply chain. And we are thrilled to be joined this week by Executive Vice President and General Manager of HPE Operations for HPE, Mark Bakker. And Mark is going to tell us more about how the world of technology supply chain is changing. AUBREY LOVELL But before we get to Mark, we're heading back to the birth of civilization itself. It's time for Technology Then. AUBREY LOVELL Okay, so have you heard of the Silk Road, Michael? And if I had to press you for when it started, when would you say it started? MICHAEL BIRD okay, I don't know. 10th century. I don't know. Something very, like really, really, really, really, really, really, really, really old. Very long time ago. AUBREY LOVELL Generally we view the Silk Road as having started in the year 138 BCE . So that's like way back there. We're truly going to ancient times with one of the oldest supply chains known to history, which is the supply chain of obsidian. Now obsidian is a bit of a wonder material if you're living in the Bronze Age, right? It's incredibly sharp edge. and the ability to chip it into usable shapes like knives or spear tips made it incredibly desirable. But here's the problem, right? Obsidian is volcanic glass. So it's not found everywhere. And if you don't live near a natural source, you're gonna need a supplier, right? But the truly remarkable thing about this story is when it begins because obsidian was so important and so useful that supply chains for it stretched back to some of the earliest civilizations we know of. Evidence of long-distance trade of obsidian in the Fertile Crescent up to 12,000 years ago show obsidian source from Armenia Ethiopia and Eastern Anatolia in Turkey being found in locations all over the Near and Middle East and the Fertile Crescent with supplies of the material traveling up and down the Levantine corridor say that 10 times fast. But it wasn't just restricted to the Middle East, right? As societies popped up around the world, obsidian supply chains were sure to follow. A report in the Journal of Archaeological Science notes that obsidian artifacts that were all found in a single location in Oaxaca originally came from multiple different sources across Mexico, highlighting ancient supply routes and Mesoamerica . people didn't have cars or planes to transport product either. Materials could take years or even generations to get from their source to the end user , not like nowadays, right? So it's kind of bonkers when you really think about it. So like, Aren't you glad that we don't need to wait that long when we order something online nowadays, Michael? MICHAEL BIRD Oh my goodness. I don't know if I could survive without stuff taking more than 24 hours to turn up. But Aubrey, can still take a while to arrive if they're made to order, especially with the complexity of modern day supply chains. So... To tell us more about it, I spoke to Mark Backer, Executive Vice President and General Manager of HPE Operations for HPE. And he started off by giving me a quick rundown of the supply chains we have to deal with here at HPE. MARK BAKKER The supply chain at HPE obviously covers the end-to-end execution of a supply chain, which starts with collaborating with the sales force and our product management community and planning the business . You need to have a plan, a forecast, a demand plan that drives signals into how much material and components we need to buy from our suppliers. It brings us to the next part of the supply chain, which is actually engaging with our suppliers in negotiating pricing, buying the parts, and getting those parts into our manufacturing locations. And from the manufacturing locations, once they receive the parts, they wait for orders from customers to drop. And the orders come in, manufacturing partners build the product, ships it out through our logistic service providers who pick it up and deliver it to our end customers, which is then the moment which we are able, once customers receive the product, we can actually invoice and recognize the revenue. It's amazing. MICHAEL BIRD How quick is that from the point of being like, okay, yeah, we think we need to order a certain number of components. Like, are we talking hours, days, weeks, months? MARK BAKKER No, it really depends on what components and what parts. There is, some call it, you know, the higher end complexity components like semiconductors. That if you look at the process to actually make a semiconductor from all the way from what they call a wafer start to the end product is a process of about nine months. So if we as an industry do not process our demand signal early enough to the industry there, we will end up in shortages there's something, an element which is called lead time. Every component, every product has a standard lead time. How long does it take from an order placed with a supplier until it gets delivered to the destination? . Some of the simple parts can be a matter of days, some of the more highly complex parts can be weeks, sometimes months. MICHAEL BIRD So every supplier has got different lead times and there's different complexities. Why don't you just order loads and bulk and just let it sit in a warehouse? Because it's not like apples where they would go off. Presumably screws will just sort of sit there. MARK BAKKER It's interesting you mentioned that. They're not necessarily apples, but we do use the reference sometimes that parts can become brown bananas. Because, you know, we live in a technology world and there is continuous innovation at the technology level and new generations of products or updates and so on happen. That's one. Second part is, that putting stuff sitting in a warehouse is very expensive. Yeah. I need to buy it. I need to pay my supplier for it. And if I have it sitting in a warehouse for months without generating the revenue associated with that, that's cost for the company. So managing inventory is the key part of a supply chain execution. You need some inventory and some buffers in order to make sure that you have sufficient to be able to execute on orders, but too much is a problem as well. Having a warehouse full of little screws that cost less than a penny each, it's okay. But a semiconductor or in the latest case when you think about accelerators or GPUs that cost $100,000 each, for them to sit in a warehouse and do nothing for anybody, it's not helpful. MICHAEL BIRD Yeah, that makes a lot of sense. How many countries does this supply chain span? MARK BAKKER Right. We as a company, HPE, delivers our products in practically all countries around the world, But we don't necessarily build, manufacture products in every country in the world, right? We centralize our manufacturing in key locations that are close to end markets so we manufacture in about 12 locations around the world Europe, North America there's quite a bit of activity in Asia because that's close to the supply base and then from there we deliver it to our end customers or in some cases our channel partners where they distribute it to the end customer they have. MICHAEL BIRD So what are some of the threats faced by today's supply chains? MARK BAKKER Well, the biggest threats continue to be, and many people know about this, right? In recent years there's simple phenomenons. Supply chain disruptions have been, can be weather conditions. can be hurricanes as an example, or there have been situations where there is a fire that happens in a factory that manufactures large volumes of certain parts that all of a sudden is no longer existing. Earthquakes, we have seen, a simple example in recent years was when a certain ocean vessel going through the Suez Canal goes sideways and blocks, you know, like a traffic jam of, you know, ocean vessels with containers full of products that cannot move. Those are some of the things. And then obviously there is everything that is considered geopolitical tension. MICHAEL BIRD So are you sort of constantly keeping on top of the news? Yes. Because actually you're thinking, okay this thing happening in another country, how is that going to impact the supply chain? MARK BAKKER That's right, We've s een many disruptions. And we have this saying as well, right, where we say, we know the next disruption is around the corner. We just don't know what it is, where it is, what it will be. MICHAEL BIRD Not an if, but a when. MARK BAKKER Right, and then using information that gets published on a daily basis, I get daily alerts of what's going on in the world with all the locations, et cetera. And if it's meaningful and material enough, you can actually take that information and feed it into your planning environment. so, know, leveraging that kind of platform and the data that comes with it, and then, you know, from there now with, you know, the ability to do generative AI and predictive analytics and all those things, it becomes much easier to anticipate those disruptions and take the right actions to be you know, proactive about it. MICHAEL BIRD well let's let's go on and talk about AI then so from your perspective how has AI changed the way that supply chains are thought about or how has AI changed the way that you think about supply chains? MARK BAKKER it's, it has changed it a lot and it will continue to change it even more, right? As you know, the different AI opportunities manifest itself, materialize, it becomes, you know, more and more impactful. Yeah. In the supply chain it's such an extensive ecosystem of different parties. It has customers and partners. has suppliers. It has manufacturing partners, logistics service providers, ourselves and all those entities transact together in that ecosystem and generate enormous amounts of data. So it's essential, first of all, to have a data management, a data strategy, as where we started. And from there, you can actually apply all sorts of analytics, prescriptive, predictive analytics, that have helped us to actually improve on our forecast signals. MICHAEL BIRD Presumably if you're using lots and lots of data sources and many of those aren't HPE data sources, there must be lots of different formats. How do you manage that process of making sure you're getting good data from your suppliers, from customers, from salespeople, from anyone really? MARK BAKKER You know the interesting thing is when you think about in a supply chain in these different domains, different parts of that value chain use different units of measure. A customer will think about, hey I placed an order with you; My component supplier needs to know from me, hey, how many units of what parts do you need? When we get to the process where we're actually building it and shipping it and invoicing, it becomes the order needs to have a dollar value. And when you get into the logistics environment, they don't care about any of what I just mentioned. All they care about is how many trucks, how many ocean vessels, how many containers, how many pallets, They don't care what's in the box. And so those different units of measure, you need to connect all those things as well. And apart from that, like you said, it's an extended ecosystem with lots of data. So first of all, what we worked on is what we call a unified data platform . Because it's like you said, in order to any of the AI related stuff, we need that data, we need it to be as clean as possible and as accurate as possible. MICHAEL BIRD Yeah, yeah that makes sense. Okay so there's a shock in the supply chain. A boat has blocked the Suez Canal. How is AI being used to manage that and how is that different to what it would have been like 10, 20 years ago? MARK BAKKER It's like I mentioned, we have this technology we use, this platform we use that gives us the alerts, that gives us the triggers. We link that again into the data management platform where we can actually track and see the material that is stuck. And then from there, you use the data in a much more automated way, digitized way, and with the right tools in your planning system, in your inventory management system, it actually helps. Before, would have an employee to go sift through all the spreadsheets and the data and try to connect that piece of information with what orders do I have in the system and now the system and the AI related algorithm can actually tell you the answer. You are going to have a problem with this order, and here is a way to solve it in the old days, all of this stuff would be manual. In some cases, the employees in Europe would not even have visibility to what's in the warehouse in North America. MICHAEL BIRD That would have been incredibly manual. That would have been quite a stressful day for everyone, I imagine, because like the traveling salesperson problem, there's so many different data points all connected.so we're able to use AI to maybe think about all of those options all at once, to some extent MARK BAKKER with the evolution of AI capabilities and tools and different variations of it, it makes it much easier. We will never be able to prevent the disruptions from happening, but we are much better prepared as a result of what we learned. MICHAEL BIRD I mean it's sort of almost creating like a self-healing supply chain. So I mean is there a human still in the loop at the moment? it's not fully automated? MARK BAKKER I think, elements of the process that probably ultimately can be fully automated. And I think the evolution with agentic AI will get us to a point, and with the next generation of technology, which also allows reasoning and all that kind of stuff, the agent, as part of agentic AI, will get smarter and will be able to respond, et cetera, and actually make decisions. We're not there yet. And I think we're also mentally not yet prepared to allow it just to go because there's too much risk, right? You mentioned earlier the importance of data. And in my opinion, we're not emphasizing the need and the importance of data enough. We're all excited about agentic AI and the agent is going to do all is stuck. But if the underlying data is not right, then, it's going to hallucinate. And you take the human out of the loop and the agentic AI agent starts hallucinating and is going to tell your customer certain things or is going to make certain decisions. That's quite scary. The outcome of that is it can be very, very bad MICHAEL BIRD Do you think you will ever get to a stage where you have a fully automated or highly, highly automated supply chain? MARK BAKKER Well look, it's hard to predict the future on that. I certainly believe that there is opportunity to get very, very close. Because we're also embarking on, we talked about agents as part of agentic AI. But we're also talking about the robotics aspect of it, right? And the physical robots and the physical robots that leverage the same thing they're learning in it to get into the more physical space of this. The thing that is important to emphasize in my opinion from a supply chain perspective is that supply chain has a lot of real hard physical elements to it. There's parts, there's products, there is boats, planes, trucks, pallets, stuff that gets delivered, forklifts involved, etc. So when you think about all those things and you say, fully automated, you need to get to a point where you can do all the, let's say the soft stuff, automated the processing of things. But at some point it gets to, now it needs to drop into a factory, which is a completely automated production line, with robots that actually do whatever and you know AI related stuff that has cell feeling that does inspection of what rolls off the production line so it's possible but it's gonna take a while MICHAEL BIRD I don't know about you Aubrey, but or coming out of this interview, think it made me realize that supply chains are significantly more complicated than I expected., like I just don't think about how something turns up at my front door. I don't think about the laptop that I'm recording this on or the microphone that I'm holding. I don't think about the individual components. It just sort of magically appears. And I think by extension, I don't really think about what happens when something goes wrong. Like when Mark talked about the ship that got stuck in the Suez Canal, there was some disruption and that did have an impact, but that's probably the first time I'd ever really thought about that. AUBREY LOVELL Yeah, it's true. It usually takes some sort of major, you know, inflection point or disaster, right? Like, for example, or even COVID, you we talk about COVID a lot and the pandemic and how that changed so many things fundamentally. Remember when we had supply chain issues with toilet paper, right? MICHAEL BIRD Totally. Yeah, totally. And I think the pandemic highlighted just how fragile our supply chains are. I was fascinated when we talked about data. mean, what this really highlighted to me is that in order to make your supply chain as efficient as possible, you just need such good data, like data, data, data feels so important. And the point that he made about how actually different people care about different elements of data. For example, if you said to your logistics supplier, I've got 20 servers, he's like, what Mark was saying is like, they don't really care, they just wanna know, is it a pallet, is it a ship, do you need a plane for it? And it just shows how many individual stakeholders are involved in the process of getting that one item from point A to point B, and how actually it just requires so much coordination and cooperation. AUBREY LOVELL absolutely. think, you know, one thing that kept coming across my mind too, and I know we didn't really touch on this, but we were talking about all the different elements of trying to be predictable and using data to have more efficient routing or how to do, you know, just predict different things that could happen, different scenarios. But one thing I remember, when I was, you know, doing work in supply chain, we had an issue with theft. And I think you could literally do a whole conversation on that around cargo theft and supply chain theft, because that is another area where you have to be, you know, predictable in terms of like using, how can you use data to help prevent some of those issues that happen, whether it's people, you know, getting into your systems or it's people that are actually going after these, you know, cargo shipments ? MICHAEL BIRD Wow! I thought Mark's conversation around how AI is being used to improve supply chain efficiency was quite interesting. Again, it relies so much on data. And I think this is one of those scenarios where agentic, really will really make a big difference across global supply chains. But yeah, the sort of then flip side to that of actually if you want to fully automate a supply chain, the way he talks about how it's so physical, know, warehouses and pallets and boats and aircraft and boxes And actually, there's going to have to be an element of robotic automation to make the supply chain fully automated, AUBREY LOVELL Yeah, and I think the balance of automation versus human interaction as well is important. Like, we're not saying, hey, when we say fully automated, it takes the human out of it, because I do think there's still a greater good that you would want human interaction. Like, obviously, you need humans to govern the process It's not like you're just giving everything over to AI, but there's a lot that they can help with that, again, frees up us to do other things in our roles. MICHAEL BIRD Yeah, and so having the human in the loop, I think you're right, is quite important. think Mark also talked about that. Anyway, so Aubrey, we talked a fair amount about the importance of data for AI when it comes to supply chains and automation, but as Mark told me, it isn't just data which needs to be considered. MARK BAKKER I think the other piece that is really important is the standardization of the process within the full end-to-end value chain. Because if you have non-standard activities happening in the process, it's very hard to automate and to get to your question, right? The full automation end-to-end. It needs to be standard. If you have variation and it makes it hard, think of a production line in a factory. If you produce a million things, they're all are the same. Same dimension, same color, same whatever it is, it's easy to automate that. If you have orders that drop into that factory that every order has something different on it, it needs to have a gold screw and the other one needs to have a blue label here, then it becomes very hard to automate something. MUSIC STING AUBREY LOVELL Okay that brings us to the end of Technology Now for this week. Thank you to our guest, Mark Bakker, And of course, to our listeners. Thank you so much for joining us. If you’ve enjoyed this episode, please do let us know – rate and review us wherever you listen to episodes and if you want to get in contact with us, send us an email to technology now AT hpe.com , subject line: Neolithic Obsidian, and don’t forget to subscribe so you can listen first every week. MICHAEL BIRD Technology Now is hosted by Aubrey Lovell and myself, Michael Bird This episode was produced by Harry Lampert and Izzie Clarke with production support from Alysha Kempson-Taylor, Beckie Bird, Allison Gaito, Alyssa Mitry and Renee Edwards. AUBREY LOVELL Our social editorial team is Rebecca Wissinger, Judy-Anne Goldman and Jacqueline Green and our social media designers are Alejandra Garcia, and Ambar Maldonado. MICHAEL BIRD Technology Now is a Fresh Air Production for Hewlett Packard Enterprise. (and) we’ll see you next week. Cheers!