BRIAN SANTO: I’m Brian Santo, EE Times Editor in Chief, and you're listening to EE Times on Air. This is your Briefing for the week ending August 23rd.
BRIAN SANTO：我是EE Times总编Brian Santo，你正在收听的是EE Times全球联播。这是截至8月23日的一周新闻概要。
*AMS decided to purchase Osram. The former specializes in sensors, the latter in photonics; together they’ll chart an intriguing technological roadmap. We’ll find out what the combination will mean for the market.
*The Hot Chips conference was held this week. We’ve got a rundown of one of the hottest, an unconventional wafer-scale AI processor from secretive startup Cerebras.
* Hot Chips会议于本周举行。我们从神秘创业公司Cerebras获得了最热门的非传统晶圆级AI处理器。
*Also today: smart water bottles. These are computerized water bottles complete with display screens and WiFI connectivity. The bottles run apps designed to entice your children to drink more water. You might be asking yourself: How smart is a smart water bottle? Well, at the moment, they are being bamboozled by six-year-olds.
ECHO ZHAO: Well, instead of water, kids can put Coca-Cola in it, so they can outsmart their parents.
BRIAN SANTO: We’ll get back to so-called smart water bottles in a moment.
First up today… Austrian sensor manufacturer AMS said it's purchasing photonics specialist Osram. It was actually the second time AMS tried to buy Osram this year.
Nitin Dahad wrote the story of the merger of the two companies and their intriguing technological roadmap. Here he is with International editor Junko Yoshida, who asked him about the financial intrigue and how the two companies are likely to proceed once they combine.
Nitin Dahad撰写了两家公司合并的新闻及其引人入胜的技术路线图。我们的国际主编Junko Yoshida与他一起谈论，向他询问了两家公司的财务状况以及合并后可能会如何发展。
JUNKO YOSHIDA: Before getting into the deal, tell me: What does AMS do, and what does Osram do?
NITIN DAHAD: AMS is very focused on sensing ICs. Sensing ICs technologies and solutions around three categories: that's optical, audio and imaging. And a key component of its business is actually consumer, around 71% in 2018. And from what most people understand, Apple is the major customer. Osram, as we probably all know, is a lighting products company, and it's got three divisions, which are automotive, optosemiconductors and digital.
JUNKO YOSHIDA: Got it. So what actually caught my eyes about the story is that there is a bit of flip flops on the part of AMS potential buyer, but also on the part of Osram. And it turns out that there were also some others involved in a bidding war. Can you break it down for us, Nitin?
NITIN DAHAD: Yes. Without making this into a documentary, let's go through this.
So AMS previously bid for Osram in May, 2019. At that time, Osram said the probability of the deal going through was low, but it would allow AMS to carry out its formal due diligence. So following that, in July, AMS said it wouldn't be pursuing the deal because it didn't see sufficient basis to proceed with the offer. You may interpret that as being they couldn't raise enough finance or I think some might have suggested it's been overpriced.
Soon after this, Osram's board received and recommended an offer from US investors, Bain Capital and the Carlyle Group, which it said was an attractive offer. So this actually came at the time of Osram's earnings announcement, when they said business was bleak and they were very pessimistic about the future. And then it goes on.
So two weeks later, on the 11th of August, AMS put in a counterbid with a higher offer, and just to give you perspective here, the AMS original offer was 38 Euros, or 38.5. Bain and Carlyle put in at 35 Euros. And the AMS counterbid was again 38 Euros.
JUNKO YOSHIDA: Back to the original price.
NITIN DAHAD: Yes. Actually, the Bain and Carlyle offer was rejected because I think one of Osram's larger shareholders, Alliance Global Investors, claimed it was too low a bid.
Since then, and the latest as of last week, is AMS and Osram have both said they are in constructive discussion on a business combination agreement and hope to start the tender period for its software before the end of the 5th of September 2019. Now if any of you remember the soap opera "Soap" back in the 70s, this is kind of that going on at the moment.
JUNKO YOSHIDA: But at the end of the day, it's clear that both sides, both AMS and Osram, now have this will to merge. Is that right?
NITIN DAHAD: Yes. Without going into sort of the technical sides of all the arrangements, yes. I think they both have the will.
JUNKO YOSHIDA: Good. One of the things about this deal, when you come to think of it, AMS is a much smaller company than Osram. So what motivated AMS to buy Osram? And why was Orsam on sale?
NITIN DAHAD: AMS, just to sort of put it in context, as you say, AMS' 2018 revenue was 1.4 billion Euros, which Osram did 3.8 billion Euros in 2018. The official line is that they want to create a global leader in sensor solutions and photonics with revenue of about 5 billion Euros. It would appear to make sense from both companies' perspectives. Because AMS is very significantly exposed to a large customer (i.e., Apple) for its consumer business, as we already talked about. And they say the deal would diversify its revenue mix and not make it so highly dependent on a key customer.
JUNKO YOSHIDA: That's important.
NITIN DAHAD: Meanwhile, I think we already said, Osram has been quite pessimistic in its forecasts, especially with automotive markets being weak. So it obviously was a key reason for it to look at this offer.
JUNKO YOSHIDA: So from the AMS standpoint, they are expanding its own revenue base by acquiring a much bigger Osram, right?
NITIN DAHAD: Yes.
JUNKO YOSHIDA: And that's the bottom line?
NITIN DAHAD: Yes. And I think they're keen to position it-- although it's a takeover-- they're keen to position it as a business combination, as most companies usually do.
JUNKO YOSHIDA: All right. Let's talk about some of the technologies involved in this MMA. So your story tells us that AMS is keen on expanding its automotive business. How big is AMS' current automotive business and what do they make? In contrast, what sort of automotive products does Osram make for the automotive market?
NITIN DAHAD: AMS, in its 2018 revenue breakdown, said automotive was just 29% of total revenue. Its automotive sensing solutions around position sensing, lidar imaging, driver monitoring, battery management and error classification. And as an example, its VCSEL... and in case we don't actually know what VCSEL stands for... I know many people use it... but it actually means Vertical Cavity Surface Emitting Lasers. So its VCSEL arrays and technology provide sensing and high-powered light sources for lidar.
Now Osram I think is clearly known for its lighting products for automotive. The rationale, as I said, for the combined business, is to boost automotive revenue. And they together want to be a leading player in optical semiconductors with solutions for 3D sensing, in-cabin sensing, industrial imaging, assisted and autonomous driving, and the list goes on. Automotive-human machine interface and automotive digital lighting, which obviously comes from the Osram side.
JUNKO YOSHIDA: Did AMS say anything about how big they envision their automotive business will be after the acquisition of Osram?
NITIN DAHAD: So, yes. Their statements say that they want to significantly sort of diversify the revenue composition of the combined business so that it's around 35% exposure to consumer and 45% to automotive. So automotive is obviously clearly a jump, but also when you look at the combined business, it's almost half. They want to be less dependent on car volume growth, but driven by market penetration of new features in the automotive business. So in short, it's an increase in the automotive business.
BRIAN SANTO: Nitin’s story on the acquisition is on the web site.
The annual Hot Chips symposium was held at Stanford University earlier this week, and the participants this year did not disappoint.
Some of the most anticipated news was from Cerebras, a secretive startup that finally revealed what it’s up to. The company introduced what it is calling its Wafer Scale Engine. It’s an artificial intelligence training system designed to stick a dagger in Nvidia’s heart.
There are no words that can adequately convey how big this thing is, so we’re just going to say – it’s big – and give you the numbers. Each system consists of 84 tiles built in a 7 by 12 array on a wafer. Each tile has 4,800 cores for a grand total of over 400,000 cores. Each core has a sliver of dedicated memory, but add all those slivers and you get a total of 18 gigabytes of memory. All of that together represents over 1.2 trillion transistors.
Now waferscale technology is exactly as hard as you’ve heard tell. The company notes that the largest waferscale device that ever succeeded was 840 square millimeters; the Wafer Scale Engine is over 46,000 square millimeters. 840 versus 46,000. Like I said – big.
The 84 tiles are connected in a mesh network, meaning the entire system is pretty much self-contained. It also means Cerebras had to devise a way to create that mesh communications network on the wafer – or on-chip, since the company considers the entire thing a single, giant chip.
Andrew Feldman is the CEO and a founder of Cerebras. He wouldn’t comment on the cost, the design or the roadmap for the rack system that Cerebras plans to sell, but he was very clear on what it can do – he said the box will deliver the performance of one thousand Nvidia GPUs while requiring just 2 to 3% of its space and power. Power consumption is absolutely critical for data centers. If Cerebras’ Wafer Scale Engine works even half as well as 1,000 GPUs, the reduction in power consumed alone will guarantee success.
Interest in this thing is keen, however. Feldman said, “Initially, we thought there would be 200 customers in the world, but we’ve revised that estimate to 1,000. Everywhere we go, we find companies with large data sets they don’t want to keep in Google Cloud where a single training run might cost $150,000.”
We invite you to eetimes.com to check out our story on the Cerebras device, along with other stories from the Hot Chips conference, including one about Huawei’s ambitious plans for a large ecosystem of AI products based on its new DaVinci cores, and another on IBM’s innovative suggestion for a replacement for DDR memory to support its Power 9 CPUs.
An astonishing percentage of children aren’t drinking enough water every day. According to a study published in April in The Journal of the American Medical Association, one kid out of every five doesn’t drink any water … in any given day. They might be drinking milk, or they might be drinking juice or soda pop or some other sugary drink, but not water.
As if there weren’t enough things for parents to worry about.
And, just like so many other sociological problems, there’s a technological fix. It’s the smart water bottle. These are water bottles with display screens. Why do you need a display screen on a water bottle? Why, to run apps on – apps designed to entice children to drink more water. One app, for example, connects young friends who can compete with each other to see who can drink their water first. That’s right – the kids connect – it’s a water bottle that requires Wi-Fi. Are you surprised that some of these water bottles can cost up to 140 smackeroos?
But even before the study on children’s water intake was published, there were already companies preparing smart water bottles. For example, Gululu, based in China but presenting itself as a Silicon Valley enterprise, ran a Kickstarter way back in 2016 asking for $100,000 to develop a smart water bottle. It received double that.
Smart water bottles are now coming available worldwide, but one of the markets that embraced smart water bottles the quickest was China. EE Times has a sister publication, called ESM China, which ran a story about the phenomenon. Echo Zhao is one of our colleagues there; International editor Junko Yoshida invited Echo back to EE Times On Air to talk about it.
JUNKO YOSHIDA: AspenCore China team recently wrote a story about so-called smart water bottles. In essence, from what I gather, this is just a water bottle. But it's connected to an app and mobile phone. It's designed to track and record how much water your kid drank per day. Just like a smart watch can track how much exercise you did today, that sort of thing. So in my opinion, this is a classic IoT device story, and yet this particular smart water bottle designed for kids because this is really targeting children. I think that makes this story uniquely Chinese in my opinion.
ECHO ZHAO: Yeah.
JUNKO YOSHIDA: So tell me a little bit about... It does say that smart bottle comes with an electronic screen. Tell me a little bit about what this display is used for, and how is it designed to entice kids to drink more water?
ECHO ZHAO: Okay. Well, through the screen you can check battery of the cup, or how much water did you drink? It also enables social interaction, music or story playing. And interactive game function. I feel it sounds like a toy for kids.
JUNKO YOSHIDA: Yeah, definitely.
ECHO ZHAO: In order to entice kids to drink more water, for example, some cups have an animated character on the screen. Every time kids drink water, sensors will make the character grow up a bit. Or they could get angry when the kids forget to drink water.
JUNKO YOSHIDA: Jeez.
ECHO ZHAO: Also some cups can make friends, so kids can compete with their friends on who drinks more water.
JUNKO YOSHIDA: Oh my goodness.
ECHO ZHAO: Parents can monitor how much water did their kids drink with an app on their phone.
JUNKO YOSHIDA: Wow. Talk about kids on a leash, right? How much water kids drank. All right. So tell me a little bit about, beyond WiFi connectivity and a small display, can you tell us about what other basic technology building blocks are in this smart water bottle.
ECHO ZHAO: The function was realized by using microcontrollers and Bluetooth or WiFi and sensors and touch screen and battery. And I think that's enough.
JUNKO YOSHIDA: Yeah, that's pretty basic IoT building, IoT device building blocks, right?
ECHO ZHAO: Right.
JUNKO YOSHIDA: All right. So according to this story, though, Echo, some of these smart bottles cost more than a hundred dollars, US hundred dollars. But apparently the smart water bottle is not necessarily as smart as it's cracked up to be. Tell us what the smart water bottle DOESN'T do, making it less smart.
ECHO ZHAO: Well, instead of water, kids can put Coca-Cola in it, so they can outsmart their parents. I believe that some people misunderstand "smart." Smart makes life easier, not more complicated.
JUNKO YOSHIDA: Yeah, that's true. I mean, this kind of tracking device in the first sight, it looks kind of cool. But it does get your life that more complicated I guess, right?
ECHO ZHAO: Yes. I remember that several years ago I was given a smart cup as a gift. It was very cool. But the cup actually smells of plastic. So I never drank water in it. Just used it as a toy.
JUNKO YOSHIDA: So wait, wait, wait. So the cup was made of what material was it you said, did you say?
ECHO ZHAO: Plastic.
JUNKO YOSHIDA: Plastic. Okay. So it made you feel a little nervous, whether it's safe to use that or not.
ECHO ZHAO: Yeah. Yeah. Exactly.
JUNKO YOSHIDA: Interesting. Okay. So tell me how popular this smart water bottle is in China today. Do you think this is just a fad that's already fading away? Or is this going to stay?
ECHO ZHAO: In my opinion, it should be each market or a fleeting shill, but given China's large population, in GD.com, that's an e-commerce platform in China, we found there are still many brands of smart cups, and some have thousands of shipments per month.
JUNKO YOSHIDA: That's a lot. Wow. I did see that there are a lot of brands. How many brands do you think that they have? A dozen? More than a dozen brands do you think?
ECHO ZHAO: I believe it's more than a dozen. And the interesting thing is, you think it is a unique market in China, but some cups, as they showed on their website, that they are originally designed in the US.
JUNKO YOSHIDA: Really? Wow!
ECHO ZHAO: Yeah.
BRIAN SANTO: The original story on smart water bottles was published by ESM China, but we’ve got an English translation at eetimes.com. It’s called, “Is a $140 Smart Water Bottle for Your Kids Worth the Price?”
And now if you’ll just step this way, I'll introduce you to the anniversaries of technological innovations from years ago this week.
On August 22, in 1955, the first computer user group formed. A group of people in and around Los Angeles who were all customers of IBM’s new 704 computer were trying to learn how to use their new machines, but got frustrated when they discovered they already knew more about the 704 than the people who sold it to them. They got the list of customers, called everyone on it – from a phone booth – Hey, remember phone booths? – and invited them all to a meeting. The meeting was held in a basement conference room at the Rand Corporation.
On August 18th, in 1947, after 8 years of working together, William Hewlett and David Packard finally got around to incorporating their company.
On August 17th, 1982, Royal Philips Electronics manufactured the world’s first Compact Disc printed with music produced specifically for commercial sale. It came off the line of a Polygram factory just outside of Hanover in Germany. The CD was “The Visitors” by ABBA. Although pressed later, copies of Billy Joel’s “52nd Street” hit the market first.
On August 21st in 1888, William Seward Burroughs patented a calculating machine. He and some associates formed the American Arithmometer Company to sell it. That first machine didn’t always provide the correct results, however. As you might imagine, kind of a bad thing in a calculator. But three years later, Burroughs got a patent for a revised version that was consistently accurate, and that’s the one that sold millions for the Burroughs Adding Machine Company. The company eventually evolved into one of the leading computer manufacturers of the 1960s and early ‘70s. It merged with Sperry in 1986; the combined company renamed itself Unisys. Here’s a clip from an industrial film produced by the company in the 1950s.
AUDIO CLIP: This was the Burroughs organization, a little machine shop in the city of St. Louis. Mr. Burroughs had worked in a bank and we know what prolonged mental computation took out of a man. Now, he had devised a machine which computed totals with cool and unchanging accuracy. He believed his machine had a big future among banks, that they would eventually buy as many as 8,000. There were then 8,000 banks in the country. Back in the 1880s, this was considered an estimate of majestic dimension.
BRIAN SANTO: For some of you, the name William S. Burroughs might have rung a bell. For others, it might have rung two bells. The founder of the Burroughs Adding Machine Company had a grandson named after him. That was the Beat Generation figure and author of the book, “The Naked Lunch,” William S. Burroughs.
And that’s your Weekly Briefing for the week ending August 23rd. This podcast is produced by AspenCore Studio. It was engineered by Taylor Marvin and Greg McCrae at Coupe Studios. The Segment Producer was Kaitie Huss. A transcript of this podcast can be found on eetimes.com, complete with links to the articles we've referred to.
We'll be back next Friday with a new episode of EE Times On Air. I'm Brian Santo.