集微網消息，在往期的集微訪談欄目中，愛集微有幸采訪了從事汽車行業工作的Mathew Vachaparampil。他現任 Caresoft Global CEO。為全球主要汽車 OEM 及供應商提供數字李生和競争評估服務。集微訪談就關于特斯拉、汽車晶片市場和車用晶片提出了一系列問題，并收到了一系列有關領域的專業答複。
問：電子電氣架構和軟體将如何影響汽車供應鍊？
答：是以我們在行業中看到的是，特别是像特斯拉和Lucid這樣的公司正在推動行業發展，他們正在從基于域的架構轉變。在以前的架構中，可以分别進行動力控制、車身控制、制動控制、轉向控制等，而且由不同的供應商提供相應的元件，如德國博世（Robert Bosch）、大陸集團（Continental）和日本電裝（Denso）等。架構朝着分區或中央化的方向轉變，而在新的架構中，車身控制子產品能夠承擔多個功能。
例如，特斯拉已經将大量的軟體工程引入公司内部，并且他們自己設計軟硬體，将印制線路闆（PCB， printed circuit board）的制造外包給深圳的供應商
我們認為，随着時間的推移，原始裝置制造商（OEMs，Original Equipment Manufacturers）将逐漸将軟體、車輛、電器和電子架構等方面帶回公司内部，這将有助于降低線束成本和電子控制器單元（ECU， Electronic Control Unit）成本等。
0到1級供應商将失去對汽車電子技術架構的控制權等。而隻是軟體的一個方面。軟體還有其他方面，比如完整的自動駕駛系統，還有人機互動和娛樂系統。自動駕駛系統中的雷達、不同類型的燈光等都是不同的系統。子產品數量将增加，電子裝置也将增多。而在人機互動方面，供應商需要與微信、蘋果、谷歌等娛樂系統和技術進行整合，而這是一個獨立的領域。
然而，總體來說，到2030年，每輛車的軟體成本将從每輛車約300到500美元增加到每輛車約1000到1500美元。同樣，電子裝置成本也将從3000到5000美元增加到10，000到15，000美元左右。因為我們想要保持對成本的控制，原始裝置制造商不願意将其交給供應商，因為一旦交給供應商，他們就無法控制成本和價格。
問：在未來電子電氣架構有什麼發展趨勢？
答：正如我剛才告訴你的，每輛車的電子成本和軟體成本都将增加。我認為，在汽車的電器和電子架構方面，會出現融合現象，因為每輛車将擁有更多的功能，是以電子控制器單元的數量将減少。其中一張幻燈片顯示了特斯拉在過去六年中如何從特斯拉Model X到Model 3再到Model Y将線束成本降低了550美元，而且他們減輕了10公斤的重量，減少了1公裡的電線數量。這是由于體系結構的融合，他們減少了控制子產品的數量，是以我們現在看到這種情況發生在進階駕駛輔助系統（ADAS， Advanced Driving Assistance System）和内部娛樂系統上。由于出現了新的領域，電子裝置的使用肯定會增加。我認為這将是汽車行業未來發展的方向。
問：在 Caresoft Global，lceberg 資料平台是如何收集和分析車輛基準資料的？
答：Caresoft是一家工程公司，專注于為汽車行業進行基準測試。Iceberg是我們的資料平台，每年會對超過35輛車進行基準測試。Caresoft擁有一種獨特的技術——高能掃描技術，這是一項擁有專利的技術，可以通過CT掃描車輛來擷取整個計算機輔助設計（Computer Aided Design， CAD）和計算機輔助工程（Computer Aided Engineering， CAE）資料。它可以通過進行碰撞分析、各種仿真模拟來評估虛拟性能，同時也能夠評估成本。實作性能的成本。我們每年在世界範圍内對來自美國、歐洲、中國、日本、南韓、印度等地的超過35輛車進行基準測試。我們在世界各地擁有超過七個技術中心，這些資料幫助我們的客戶了解市場上最新的技術動向。
以及成本是多少？此外，我們還有一個名為DeltaCosting™的平台，它可以分析車輛設計和成本，并提供如何以最低成本獲得最佳性能的建議。是以，這就是客戶使用我們資料的原因。Caresoft極大地專注于幫助我們的客戶改進技術，同時也降低成本。
問：AI對幫助客戶進行成本分析的前景如何？
答：當我們已經有了資料和觀點，尤其是我們有虛拟仿真：我們可以模拟最後能獲得多少性能，我們有資料，同樣也有成本資料。我們正在研究将人工智能應用于這個領域，以便找到最佳解決方案，就像今年早些時候推出的ChatGPT一樣，在人工智能方面取得了巨大的成功。随着我們獲得更多智能資料，包括設計、性能和成本等參數化資料，我相信人工智能有巨大的應用前景。請繼續關注我們，并期待未來會有哪些新進展。
問：您能否預測未來汽車主機廠的并購趨勢？
答：這個問題非常具有挑戰性。原因是車輛、電子電路和軟體系統的發展方向都非常清晰，并且向内部內建。就自動駕駛輔助系統而言，我對美國的情況非常熟悉，我們有更多，是以我們有自動化巡航等等。但是，實作L4、L5自動駕駛非常困難，我們不知道誰将成為赢家。另外，特斯拉在這方面也是一股強勁的競争力量。在中國，百度、華為等各種公司也在研究自動駕駛系統，是以很難預測誰将在這個領域獲勝，因為現在玩家太多了。從原始裝置制造商的角度來看，原始裝置制造商将購買解決方案，但他們需要充分整合軟體和電子架構，以使其無縫運作。這就是我能告訴你的。此外還涉及到與人機界面和客戶技術的整合，例如微信、百度、蘋果等等。控制資料是一個重要的方面，這些領域仍處于開放狀态，我們仍然無法确定其中的規則。
問：晶片短缺如何影響汽車供應鍊？
答：在過去的兩年中，由于缺芯的發生，許多原始裝置制造商不僅僅隻看ECU，他們關注每一顆晶片。例如，他們在T-BOX、 ECU上面向博世和電裝這樣的公司采用雙重采購政策，但卻發現這些公司使用了相同的晶片，這意味着晶片沒有采用雙重采購政策，這就存在風險。是以，現在原始裝置制造商正在考慮整個晶片的生命周期，并試圖確定有雙重采購政策。正如我之前提到的，他們正在嘗試更多地獲得軟體和電子裝置的控制權，因為他們不想外包。這對于作為原始裝置制造來說是其未來發展的核心。如果公司想要生存下去，就必須了解技術，不能隻關注車身、底盤或者機械加工流程，而忽視了電子、軟體和技術。這是汽車的關鍵所在。我認為原始裝置制造商會在這個領域上掌握更多的控制權。
你談到了福特F-150 Lightning由于晶片短缺所導緻的問題。正如我所解釋的，他們希望降低電子和晶片領域的風險。在吸取教訓後，許多原始裝置制造商現在正在内部開發軟體和設計電子裝置，以便在零部件方面擁有多個供應來源，不再依賴供應商，這就是我對此事的看法，而他們也正在密切與晶片制造商合作，簽訂長期合同等等。
問：如何評價特斯拉持續降低産品成本？
答：例如，在車身、電子電路、控制子產品等各種系統中，我們看到特斯拉正在大力推動成本下降。并且特斯拉宣布他們想要減少汽車生産成本的50%，他們正在采用子產品化的生産系統，我們認為他們可以實作自動化生産線，并減少塗裝工廠中的房間等投資規模。
同樣地，他們正在探索各種技術，例如采用甯德時代（CATL）的磷酸鐵锂電池，使用碳化矽晶片在動力系統中。我不知道特斯拉減少75%的碳化矽使用這一說法是否正确，但我可以告訴你，特斯拉非常注重保持汽車性能不變，力求以更低的成本提高性能。是以，就性能與成本而言，特斯拉确實是當今行業的基準。
Q：as the ee automotive architecture gradually become centralized. Software programs are clearly becoming more crucial to the car， still believe the law of gears， too， is evolving a lot.How will Electrical/Electronic Architecture and Software Impact the Automotive Supply Chain？A：So what we are seeing in the industry， especially with companies like Tesla and Lucid， they are moving from as a domain based architecture， where you do power train controls， body controls， brake controls， steering controls separately， and you have different suppliers， steer ones. It could be Robert Bosch， Continental and Denso supplying these components. To an architecture， which is moving more zonal or central， where body control modules are doing these multiple functions.Tesla， for example， has brought a lot of this software engineering in house. And then they design both the software and the hardware and they are outsourcing the manufacture of the PCB boards to suppliers in Shenzhen We feel over time that when we look at the software or the vehicle， electrical and electronics architecture， OEMs are going to bring that in house which will reduce harness cost， ECU cost and so on.And 0 to 1 suppliers will lose control and so on. Now this is one part of the software， but there's other parts of the software， too. There is a whole autonomous system. There's also the HMI and entertainment systems. The autonomous systems are， do you have radar？ Light are different systems there. The modules will increase， the electronics will increase. And then HMI you have integration with Wechat， with Apple， with Google， with all these entertainment systems and technology there. That is a separate avenue to go.However， overall， the software cost per vehicle is going to go from around 300 to $500 per vehicle to around 1000 to $2，500 a vehicle by 2030， same way， electronics is going to increase from 3，000 to $5，000 to maybe 10 to $15，000. The OEMs， because we want to keep control of this. Because if they give it to suppliers， they have no control on the cost or the price.
Q：Previously， the number of ecus was limited and they were all isolated from each other. Each ecu has one specific function of which features were simple and basic. But now the new features and new safety countries require a lot of additional using vehicles. Could you estimate how the cost of milk will be effective with an increase of the ccu。What is the development trend of Electrical/Electronic Architecture in the future？A：As I told you， just now， the cost of electronics is going to increase per vehicle， the cost of software is going to increase. I feel in the electrical and electronics architecture for the vehicle， there's going to be convergence， so the number of ECUs is going to reduce because you're going to have more functions per vehicle. One of the slides that I will show here， show you， shows how Tesla has reduced the cost of harnesses by $550 in the last six years from Tesla X to 3 to Y. They have reduced the weight by 10 kilos. They reduce the amount of wire by 1 kilometer. That's because of the convergence of the architectures， and they've reduced the number of control modules， so we see that happening now on the ADAS and the entertainment inside. Because there are new areas， electronics will increase for sure. This is the way I think that the industry will head.
Q：At Caresoft Global， how does Iceberg data platform gather and analyze vehicle benchmark data？ Caresoft Global ，Iceberg What’s the methodology of Iceberg regarding gathering and analyzing the benchmarking data？
A：So Caresoft is an engineering company， but we do a lot of benchmarking for the automotive industry. Iceberg is our data platform where we benchmark over 35 vehicles a year. Caresoft has something unique. We have something known as a high-energy scan， which is a patented technology where we can see this scan the vehicle. From that， we can get the entire CAD and CAE data. This allows you to do evaluate virtual performance by doing crash analysis， doing various simulations virtually. That allows you to also evaluate cost. Performance was this cost. We do around the benchmark over 35 vehicles a year around the world， United states， Europe， China， vehicles from Japan， Korea， India， and so on. We have over seven technology centers around the world， and this data helps our customers with the what is to understand what is the latest technology in the market.And also what is the cost？ we have a platform called Deltacosting which looks at the vehicle design and the cost. And they can also see how can we get the best performance for the lowest cost， so that is what they use our data for. Caresoft is focusing tremendously on helping our customers improve technology， but also reduce cost. And speak slowly.
Q：What is the promise of AI for helping customers with cost analysis？
A：When we look at all our data and the intelligence， and especially since we have virtual simulation， which is we have performance， what's your performance？ We have data， and we also have cost. We are working on some technologies to integrate AI into this so that we can find the right solution for like ChatGPT earlier this year was huge in terms of launch from an AI perspective. As we get more data and there's intelligent data where we have parametric data which looks at design， performance and cost， I definitely feel there's a huge application of AI and you can stay tuned to us and see what's going to happen in this space.
Q：Could you forecast the M&A trend of automotive OEMs in the future？ Thank you. So presumably， the large of the boy to the old hands with many resources will design the software in house and develop our relationship with the peer ones for all the basic of the most automated driving agency. Could you forecast the mergers and accusation trend of a motive OEM in the future？A：This is a tough question. And the reason is there is a clear direction on the vehicle， electrical electronics and software systems. It's in house. I've explained to you， on the ADAS systems， I'm very familiar with what is there in the US we have way more， so we have cruise automation and so on. They are finding level four， level five， very difficult. We don't know who's going to be the winner.Then there's Tesla. Here in China， you have various Baidu and so on， Huawei all looking at autonomous vehicle systems， so it's very difficult to predict who's going to win this strategy， the platform strategy here. So I don't know the answer to that， but there are many players here. How does it affect things？ From OEM perspective， OEM will buy the solution， but they need full integration with their software and electronics architectures so that it functions seamlessly. That is what I can tell you. And then there's a whole technology place of integrating with the HMI and customer technologies. Be it. Wechat be it， Baidu be it， Apple. Controlling the data is a big piece. So that's a whole piece that we don't know it's still the wild west in these areas.
Q:Thank you. So did you have your ratio of automotive cheap shortages？ Automotive OEM are driven to work with the team makers more close？ Only before， because possibly their honorable could be faced at all . by a tiny MCU how they built across the evaluation system is influenced by such a phenomenon， okay？How is the chip shortage affecting the automotive supply chain？
A：With the chip crisis that happened in the last 2 years， a lot of OEMS are not only just looking at the ECU， they're looking at every chip. For example， they were a dual source to T-box， ECU， to say， Bosch and Denso. But they find out they use the same chip， but then there's no dual sourcing strategy. There's risk. So OEMs now are looking at the whole chip life cycle in them. And they are trying to make sure there's a dual sourcing strategy. And as I mentioned earlier， they're trying to get more control of the software and electronics， because they don't want to outsource. This is core to the future of the company as they be as OEMs. If you're to survive， you have to be understand the technology. You just cannot build a body or chassis to the mechanical process， the electronics and the software and the technology. That's the crux of the vehicle. I feel that OEMs are going to take more control of this area.You talked about the F-150 Lightning that is because of chip shortage. As I explained， they want to de risk the electronics and chip area. Having learned their lesson， a lot of OEMs are now insourcing the software and also the design of the electronics. So that they have multiple sources for the components， and they are not relying on suppliers. That's the way I see this and so on. And they are also working very closely with the chip manufacturers for long-term contracts and so on and so forth.
Q：Thank you. And tesla revealed our plan for reducing， as I see devices in our train by 75 % without any more illustrations. Is there any substance or spencer goal behind what they are claiming and how to interpret the number of 55 %？What do you think of Tesla Continue To Reduce Cost of Existing Products？
A：Okay. We are seeing， for example， in various systems， in body， in white， in electrical electronics， in control modules， Tesla is really pushing the envelope to take cost out. And with Tesla making the announcement that they want to reduce 50% of the production of the car， now they're going to a module bullet system where we think they can automate the assembly line and reduce the size of the investment for paint shops and so on.Same way， they're looking at various ways in technology like they move to lithium-ion-phosphate batteries with CATL， they are looking at silicon carbide chips in power train. So I don't know whether the 75 % is correct or not that I don't know. But what I can tell you is Tesla is extremely aggressive about keeping the same performance， increasing the performance for a lower cost. So Tesla is really the benchmark in terms of performance versus cost in the industry today.
![C# - 方法的可變參數-params 050[圖]](data:image/gif;base64,R0lGODlhAQABAIAAAP///wAAACwAAAAAAQABAAACAkQBADs=)