good afternoon. good afternoon and a happy new year. you know it should be a vacation day but wehave a special treat so i hope it's okay to be out here on chinese new year. so welcome, distinguished guests, students,faculty and friends to this very special event:
agilent stock price, the 4th annual ernest s. kuh distinguishedlecture. each year, the kuh lecture brings an outstandingleader in technology and engineering to our campus. we're honored to have launched the series witha talk by intel co-founder, andy grove, followed
by bill sullivan, ceo of agilent and continuebuilding this tradition of excellence with our 2014 kuh lecturer, sehat sutardja, andour 2015 kuh lecturer today, wayne dai. this year's student cosponsor for the kuhlecture, is the berkeley chapter of the national electrical engineering and computer engineeringhonor society, eta kappa nu, those are the folks in the engineering shirt in the back,so... thank you all very much. [applause] this wonderful lecture series wasendowed through the generosity of ernie and bettine kuh.
the kuh family is here today, ernie is recuperatingand is not able to be with us here today, but bettine is here. ernie was dean of the college from 1973 to1980 and he's been a national leader in engineering education throughout his career. he's professoremeritus of eecs and a true trailblazer in the design of integrated circuits and systems. it's an honor to be ernie's berkeley colleaguesand i thank him and bettine for making this splendid occasion possible. we are delighted today that bettine, ernie'ssister susan, his sons, tony and ted are here. and i think tony will deliver a few remarkson ernie's behalf in a moment.
we're also delighted that some of wayne'sfamily: his wife joanne and daughter tiffany are here. tiffany is a grad student in bioengineeringhere at berkeley, continuing this tradition and she just finished her quads. is that correct? so congratulations, tiffany. [applause] okay, i'd like to welcome tonyonto the stage to say a few words. please, tony. well, thank you shankar.
it's really a pleasure to be here. my dad, he's been here almost--he first cameto uc berkeley in 1956, so it's been almost sixty years. so, my parents talked about this and theydecided to endow this lectureship series and let me just say a few words about this. so, you know, i talk to my dad quite a bitand, you know, on many occasions, he tells me, well "tony you need to look at a biggerpicture." for a while i didn't quite understand whathe meant by that. but then, you know, over the years i've seenand grown and i see everyone's so busy these
days, especially in the college of engineeringhere. we all have to solve a number of importanttasks, some these are small tasks but they're very important to do and we sometimes losefocus about what the bigger picture is. so whether you're a student: you're crammingfor an exam, you have to learn some difficult equations or you have to understand some complexalgorithm; or even as a faculty member: you're working on a paper with a student or colleague,you have to make sure that the results are presented accurately and correctly and youhave to make sure that the presentation is appropriate and properly nuanced; or you even--a university administrator: you have daily tasks that you have to accomplish with yourstaff and other faculty.
so, you know, we lose sight of sometimes thisbigger picture. so i think this lecture helps us to maybepaint part of this picture by looking -- taking a step back. exploring a little bit about, you know, whatis going on in the broader context of things. so these distinguished lecturers provide viewsof their experiences and give insights and also maybe provide some inspiration aboutdifferent topics in engineering and technology. so we're very pleased today to have dr. waynedai talk. wayne was a student of my father's and weare close family friends. so i also like to give special thanks to shankarand his staff for arranging and organizing
this event and finally, i'd like to thankthe audience, especially the students for coming here today. so thank you. [applause] yeah, it's my privilege now to introduce wayne. you know in this job, it's really a privilegeto meet many remarkable people and business leaders from around the world. few, however can rival wayne's vision andskill as a ceo of verisilicon holdings. wayne is founder, chairman, president andceo of verisilicon holdings. he was the co-chairman and cto of celestrytechnologies which was acquired by cadence design systems in 2002.
he was elected as one of the top-10 venture-backedentrepreneurs in china in 2005 and was elected as one of the top 10 talents of science andtechnology in china in 2005. in 2007, he was recognized as the ernst & youngentrepreneur of the year in china. before starting verisilicon, doctor dai wasa professor in the baskin school of engineering at uc santa cruz. he's published over 100 papers in technicaljournals and conferences and he was also an nsf presidential young investigator in 1990. he currently plays a leadership role in theberkeley shanghai club and has also been an unwavering supporter of the college's activitiesin shanghai and throughout china.
i'm sure a lot of us here have actually beenguests of his at the berkeley shanghai balls and those are really spectacular events. i believe wayne is one of the most passionateand dedicated alumni we have. so thus it's with great gratitude and delightthat i welcome wayne dai to share with us his far-reaching ideas. wayne. [applause] thank you dean sastry. today actually the new year day. in mandarin, "yang" means "horned animal." but i don't know if you know: is this theyear of sheep, goat or ram? now, i think in usa today, they call the yearof sheep, in the wall street journal, they
call year of goat. but daily news simply call the year of anyruminant horned animal. actually, this morning, president obama actuallygave a greeting, he didn't specify, he said it doesn't matter which is year of. he didn'tknow is it sheep or goat or ram. so let me take a vote, see how many people think it'sthe year of the sheep? yeah, about 15 people. how many people think the year of goat? we seem to have more people. how many people think year of ram?
oh, that's surprising. but i think the correct answer probably yearof goat, because in a certain part of china, i don't know if you know, the goat is, theycan climb very well. so this is because in the history of china,this is how they define this. so i think probably most likely in china. you probably didn't know that they have veryspecial skill. so starting a company is like climbing a rock. as you know, i got a tenured full professoruc santa cruz. is only two secure job in u.s., federal judgeand tenured professor, right?
president have term, and the ceo can be firedany day. so giving up tenure a big deal. so why i do this? it has something to do with the china market. the china ic market, at the time i started,they're only about 6.7% worldwide market, not big. then 2005, already exceeded u.s., 2010 exceededu.s. japan combined, and this year it's three times the u.s. market. so that's huge.
however, china import more ics than oil. look at the supply -- only 10% or so. so this is the market. if you have market but don't have technology,technology will come. you don't have money, money will come. so that is a key thing in this market. truly, in 2001 i started a company. this is how the semiconductor grows. and of course on the way there's lots of governmentpolicy.
and recently, next five years, china's governmentalready allocate hundred billion u.s. dollars. nowhere in the world have this kind of fundingto invest china semiconductor manufacturing design. so that's key. this is number of fabless companies in china. this year actually 2014, already 1600, i thinku.s. probably 200, taiwan 200. lots of them maybe small, maybe not go anywhere,but the talent pool, nothing wrong with [??]. that means over the last 10 years, there's a hugetalent pool develop, combined with the capital china going to put in next five years -- that'sthe golden years for the semiconductor china.
here see when time we started, were 200. our company currently have 500 employees,70% dedicated in r&d. see, we don't have fab, we don't do manufacturing,we do design, and based in shanghai, in this building, in the [??] park. by the way, this park has more than 2,000companies, probably the one closest can't duplicate us anyway. we're 70% based in china, however with 70%revenue come from outside china. this you look at, we're actually offices ineurope. so now, i'm kind of not talk about specific technology, maybe talk business model, i call business 1.0.
everyone knows when you sell product, company is the product. probably i will notstart company just sell one product. so product, people will say tam total availablemarket, market share, asp average sale price, and the gross margin. this is very traditional. when every company, you first probably askwhat product you sell. however, recently there's some change, obviouslyon top of 2.0 and 3.0. now this is the top 10 fabless companies in2013. and the percentage means, spending as percentageof revenue.
now they grow from 19% to 25%. some high-growth company move 30, 40%. what that mean? if your gross margin below 45, 40% you cannotsurvive. that's why they need the 50% gross margin. those the companies no fab, no manufacturing,just [unintelligible]. this is changing, getting worse on the right,because the situation only invest more. whole semiconductor growth single digit, sowall street say what's wrong with that industry? you have mature industry, that means you needget more money back, right?
we're immature, but you don't invest more. [unintelligible] so this is the problem. so this is really a thing we see is the currentsituation with semiconductor costs: the r&d costs so high, even without manufacturing. this happened 20 years ago, when the manufacturingcosts more. then people say, taiwan say, ok, not everyonebe the fab, i be the fab, [unintelligible] and you just doing design. so that's called fabless. like last slides i show for fabless.
so those are found with a foundry, englishyou call fabless, no fab. they're actually significant contributor wholeindustry, like broadcomm, qualcomm, marvell, all those companies. now right now, even without fab, design veryexpensive. so we need also outsourcing design. so we call next wave is design-lite. we don't say designless, we tell people, fabless,designless, they don't have anything to do. so design-lite means you focus on core competency,you don't do everything yourself. so this is probably the business models.
and we call silicon platform as a service. so you know saas, software as a service. what do we say, is no-fab, of course, butalso we don't brand the product, is a service. beauty of service is you don't invest, it'sother people's product. and also, you have limited inventory risk. now this is our, the chip we design in themicrosoft xbox, in iphone 6, 6 plus, and also galaxy, chromebook, chromebox is video conferencefrom google. but we never brand those products, we onlydesign and handle manufacture, but that's a new model, we never brand them.
now this is i call the xaas . we have softwareservice, platform service, infrastructure service, even security as service. service, right? let me give you some examples. one example is microsoft xbox kinect, videosensing, you don't need -- like wii, you have a string attached -- no string attached, itcan play the game. but also if you change your clothes, you goto shop, immediately, the video sensing, your body can put on without going to changingroom, all those things. so this is the [??] company.
when they started they only raised $9 million. these days, 10 years ago silicon valley fundedmaybe 50 angel investors for ic companies; now maybe 5 -- one-tenth. to get money for startup companies not easyas 10 years ago. so they only raised $9 million. if they do everything themselves, there'sno way. so from day one they outsource to verisilicon,all the design, handle all the manufacturing. so finally they use this 9 million all theway with microsoft. they ship actually very well, and also 2 or3 they sell to apple 360 million.
this is the model of design-lite -- doingvery well. if they everything doing themselves, you cannothave this outcome, probably run out of money. i give you another example, this is with google. how can you work with google, not an ic company? not like apple, right? but google not apple, has 2,000 people into everything themselves. google's different. you don't see google do the bigger thing for ic. i give you some background.
youtube is very good, video is very important-- 1 billion users, 50% more, youtube. the video form, actually people collect royaltiesh.264, h.265. google don't like that. you don't want collect royalty because yourformat video format. so they bought a company, they create a formatcalled vp8, vp9, this called webm. so right now the youtube 4k -- 4k is high-definition-- only support vp9, not support h.265, because they need that royalty. this ces, many tv companies actually support that. now i just give you this background: vp8,vp9 is the google response to h.264, h.265
without royalty. google wanted royalty-free. google is not a company doing ics, but youneed ics. so that's why we will work with google. they have a chrome box, this small box has16-side high-definition video conference, right now using intel i7 with the fan, usesoftware to decode, encode. so google said, can you replace that? that's too expensive. i give you specs, one page, no details, say,can you do this -- replace that? within 6, 7 months we did design, finish,we replace.
now when we do this kind of job, this service,you do everything. they don't tell you how to, you just say,can you do this? so we complete from spec, and we finish wholething, and the week before that we have google software team have to port their software. so this is the new model. our customer is not a chip company, it's not like someone says, i'm busy, can you do this job for me? is really the internet company. this is called turnkey model. [unintelligible] this first google-brand chipreplace all the direct verisilicon.
this is new model, in the past not like this. so this is good. basically what i'm talking, with this buildmodel we do service all the way, from spec, from anything, till finished chip. we have a general contract, i go to waferfarm, you go to patch testing, anything wrong is my responsibility. like microsoft xbox, we ship directly, youhave to ship on time, with correctness. we tape out one chip a week, this is all inshanghai, 50 chips a year, with majority done in advance nodes.
and one thing be very important, we deal withall the foundries. many service companies only deal with one-- if you do tsmc, you cannot do umc, you do umc you cannot do tsmc. we deal with all the foundries, including[??]. at any given time, i have 15, 17 chips concurrently, because you can tape out onechip a week and not finish chip in one week. you concurrent do 15 chips, maybe four ofthem [??]. that's all done in shanghai. so the way, the engineering actually growsvery fast [??]. what that means, in this 2.0 ic, you should sell service sale path. look at the internet companies, look at alibaba-- they don't have product.
so i think the better way is sell service. so verisilicon, selling platform service,you sell service. now, can we do even better then? what's better than service and platforms? before i talk about that, before i talk tothe business 3.0, i'd like to mention one thing the abstract talk about: how you canmake this engineering. those people only last 10 years. how can they turn their design, when the multinationalcompany come to shanghai, give 30, 50% increased salary, why don't they leave?
otherwise you become scooped. after three years they all left. our turnover rate less than 5%. that's very, we are the first pace they wouldgo because our people doing different variety of designs. and also, we acquire a division from lsi 8years ago, dallas-based. you know, california company merging withtexas company, sometimes quite challenging -- the culture different. shanghai with dallas -- they sound bossy,are you crazy?
i remember on the first day with lsi exactly,when there is people looking in, no expression on their face, saying you from outer space. you from china, you buy ip? they thought china just steal ip, why youneed buy ip? are you going to relocate us to shanghai? so is very strange. i say, well, you cannot force people to stay. this is the dsp division, they say, they giveyou dsp, bye bye, we all leave. so we screw up.
i remember that time, my first talk is veryimportant, i cannot talk about technology. they say, you from china, china technology. these people from intel, ti, motorola, noone will respect. so i say we need to say something else. so that time i remember 9/11. so there's a parent from shanghai that wentto u.s. to visit a son, actually died on the plane crash. there's a jazz singer chris walker in texas. he's never been to china but somehow he sawthis news, he's so moved he wrote a song,
together we stand. and he says someday, he'd like to sing thesong in front of the relatives in shanghai. at that time we have very small operationin santa clara and a big operation in shanghai. they said, well, this person's quite good,cooperation between u.s. and china, so we sponsor concerts, we invite employees, customers,very obvious the concert very well. at that time i didn't know we going to somedaybuy a company in texas. so that day i didn't talk about technologyvision or something, i just showed them, see, we had a concert and someone you know. so that actually made some change.
so they say, well, this company a little bitdifferent. they think of a high-tech company, you havethis fun concert. now also i saw this ping-pong table there. i brought hr lady from santa clara, i say,"after my talk, anyone who can beat her, 10% raise." in the u.s. usually 3% raise or 5%, no, a10% raise. and the ceo, in first talk he cannot lie,right? so there are 30 people, they all practiceevery day. they didn't know our hr lady is the nationalchampion in china. so the u.s. open set, there is no way they can beat her.
but at least they think this is not a normalchina company kind of image. so actually in two weeks i want them to cometo china. some people don't have a passport, they'venever get out of country before. they [??] passport, they bring drinking water,instant noodles, and beef jerky to china, because they worry about if there's no meat. we start from that position, but no one left,and that's amazing. they're very productive, the dsp team. of course in bay area for last 10 years we have very civic [??], especially during the downturn in 2008. there's two things for a startup company: instant noodles and a ping-pong table.
the ping-pong table is easy to exercise, youdon't need [??]. so for the last ten years we sponsor that, we doing very well. so this i just show you the culture, how tointegrate. i had to go to christmas party in dallas, in santa clara, in [??], so you had three places to go, so we integrate quite well, the team quitewell. we also have office in nice, france, europe. last year is the 50th anniversary of normalizationbetween china and france. so we invite french artist to paint this 12-meter-high, long in one hour with the symphony, and also the pianist [??] educated in paris, and shedid a great job. we usually have this culture event.
after the show, someone come to me saying verisilicon is alibaba in semiconductors. i would never receive this high comment ifwe don't have this show. so this is some kind of branding. this shows, very interesting, our shanghaiengineers, where they come from, their home town. i cannot go to everywhere to recruiting. they go to top university from home town,you can see everywhere. only 12% from shanghai. our slogan, actually, is fair, care and shareand cheer.
we didn't say work hard; lots of china companies,you see when you walk to office room there's a big slogan saying you have to work hard. if you're happy, of course you'll work hard. we're very fair -- my driver and receptionisthave stock options. in china that's usually not the case; maybeboard and you don't have stock options. share is very important, because if the seniorengineer is not willing to train the junior ones, then we have big problem. usually our junior ones, two or three years,they tend to [??] their own design. and you know how we are recruiting, when wego today, i saw this shanghai jiao tong university
visit here, so usually my first thought is[??]. so usually i walk into, i will have microelectronics unit only, master degreeand above, we don't take undergrad, usually 300 people, sometimes more, in the room. i cannot interview, so i give a talk. usually i want to go to campus to talk. and then right there we have reading exams,immediately after my talk. reading exam means you want to do circuitdesign or you want to do logic design, four positions, there are four different tests,you form different row. after tests, you go home, have dinner andopen your cellphones.
so we will give you, before midnight, if iwant invite you next day to come to interview. maybe usually around 80 people come back forthe face-to-face interview. how interview? we have a whole team: the first-line manager,of course, and the vp, me and hr, the [??] test. by 5 o'clock second day, we give the offer. we usually take maybe 15 people. so this kind of screening very difficult todo in u.s. for example, analog circuit, for people wantto design, you need two-year layout first, before you do analog circuit.
so what that means, when you do analog circuityou barely feed us, if you're too busy you do [??] yourself. we have friends working for us. so this is very good quality. i think this is very important, but two examplesi give to you. we have paradigm shift. these days, everyone go around the internetcompany. if google say, i don't buy server from ibm,ibm have the open power -- they sell it chip. this is a whole paradigm shift.
so sometimes i think technology innovation important, but sometimes the business model is. this is the business model innovation. i don't know if the chinese new year the lastcouple of days, you know this wechat. they selected red envelope, huge, huge. so that night, everyone shaking, try to getthe monies -- this huge. you can tell this is the way the businessmodel changes. now, of course we have business 3.0. better yet, not only sell service and platform,you just sell experience. i don't know if you heard of xiaomi.
this only four years, not everyone heard this. they founded in 2010, so they already evaluated $46 billion u.s. dollars, the most expensive private company. is close to toyota. i don't think they havemany patents. how come you have the four-year internet company can grow that high? just one year they almost increase 4.6 times. look at the cellphone: last year they sell61 million cellphones. and $12.3 billion revenues, and the users, 85 million users, and theythink this year they can sell 100 million. so in china already no. 1; apple has certainfans, they probably more stable, but other.
so china brand ones, are like huawei, theyall come up. so this is very good point. and recently they introduce note, the big screen. they only sell internet, 3 minutes when opening, sell 2.2 million. of course no middlemen, no channel costs,of course they save lots of money. but you need fans, -- how can you have 3 minutessell that many? look at the spec -- actually not bad, $370. this spec is very close to the most, and thequality, the case designs, etc. how they do this?
basically, lots of things they learn fromapple: for steve jobs, the quality, whatever. but one thing different from apple is theuser has the sense of participation. they have hundred developer software. they release software every week -- can you imagine how you could release software every week? every monday they do [??], develop, tuesday,so friday they doing new release, tuesday some user give the feedback. they quickly fix something, really from allthe feedback. so you have the whole r&d team, you have thewhole qa, your customers are your qa. so you see 100 developers, they have 1,000maybe the more core people, those people very
experienced, do the early test, then you have10,000 -- this can't go like this. so people feel happy if i report a bug andyou fix, they don't complain. they even feel proud, or they have some suggestions,then of course you have to check. all the notes, they post those notes. maybe last four years the notes could be goaround the whole earth. there's lots and lots of them. this is different, the way, this huawei theyhave fans. because this kind of feeling you cannot dowith apple like this. so everyone participates, especially younggeneration, they like this.
basically you have all the [??], all the people,and you have lots of qas, lots of developers. so that's why can see how they grow over thefour years, they grow very fast, the base. now i give you another example: the wristband. recently they start to sell in u.s., $13. in u.s. probably $100, right? or $80. 30 days, batteries 30 days. look at the industrial design, showing verygood quality, very nice looking. i think they hire industrial design from thenest. [unintelligible] they not cut corners.
usually [??] cut corners, cheap, this notthat, this like they learn from apple. so you really need good quality. not only count your step, after you set astep a week you can win a free coupon to test, so this $13 can make money. some other thing like unlock the phone, becausein china the average you need 80 times tap the password a day. so this within certain very short distanceyou unlock, become your id. the reason they do this is not just sell thisband to make money. they want this become your id, not just countyour step.
for now when you go to shopping mall, youlook at certain item for long time, means you really interested, they immediately, bigdata search, are you the real good vip customer, not random person? if you are the vip customer it immediatelysend message for you, only for today only, give you 20% discount. right? this is real. so they pick an id, or you walk in with tv,they immediately jump to the tv station you usually watch.
so this becomes your smart home, this kind of ids. so they've acquired well, this one. and also, this is the ip camera, this is called smart homes -- about $30. very nice-looking ip cameras. they can put everywhere. i don't use this one because first i talkabout this, the experienced people, they really not use platform, also the user experience. all i need is a very simple inside. those need ics.
i think right now very important, when i cometo berkeley i did a 2-year physics in shanghai jiao tong university undergraduate, then cometo computer science here in graduate school. there we talk about computers. now we're not only computer scientists, thisis also data scientists, it's all about data. look at the generations: used to be nokia,very solid hardware. then you have the apple ones, the softwareapps. now i think data is more important. so we need data scientists, we need information,like computing [??]. so this is the key message, i say we need to look at the business model,not just talk about product, we need to look
at service and platform, like verisign does,and then sell the experience. so probably i will try to quick this, becausei want to leave some questions for people. so iot, i already, probably, yeah, i will,yeah. iot is internet of things. this like the pc, then smartphone, the nextbig things. however, we need to be very careful. for in the iot mainly two things. lots of watches are black; you cannot have black watch,you don't want to wear. so the display is very important.
qualcomm has technology, 1/10th the power,that's why all still is about technology. event though we talk about internet comingto well, the fundamental analog eye of design, those still the key. this, the battery you cannot input that much. this, the display technology is mems, 1/10th. especially, bright sunlight, they will reflect it. we also work company with this fingerprint. you know, the swipe is 1d; 2d is press, butyou can fake the iphone with rubbers, very easy. but this uses ultrasound; it's the smallestultrasonic machine in the world, [unintelligible].
because i think the key thing is security. this is the bigger problem. but we all depend on technology. [unintelligible] now, the last point i wantto make is, we know this is the big thing, but we need to be careful. the market is so fragmented, if we not doingright, the company will become an idiot. so you will see many companies will die. before [unintelligible] i can do 20 watches,30 watch, so we'll survive, i will survive. the last remark for the students here, i think is the key thing you need to do some reading every day.
right now is short memory, they don't do really study, they used to just look at the pictures. you can see the numbers; we are the engineers, we like to see numbers. you can see the difference: just make tinyprogress every day to reap a big harvest after a year. thank you.[applause] dean sastry: so we'd like to invite questions from the audience, and we're going to give first preference to the students. wayne dai: so we have the incentives --i brought some calendars. this is unique; we made a calendar have both china and u.s. holidays. so if you are interested in china, or youget back for this year, you cannot buy this calendar.
so whoever asks the questions, you got thiscalendar. dean sastry: well, you got a name, major and year, and then you get a calendar in return. audience question: my name is salvan patel,i'm a second-year eecs major, part of hkn, that's the shirt. my question was, you introduced a lot of theseinteresting products, a lot of innovations, that are meant for the general public. what do you think is the major hurdle fromintroducing these as sort of a research product or a novelty, to like having them introducedto the general public, and having everyone have a smart watch or a smart id?
wayne dai: that i say, the batteries willbe a big problem, because you cannot charge a watch every day or every week, there's nota way. you cannot have a watch black. so that's a huge problem, the display problem. and also, wi-fi. look at the wi-fi right now. even though qualcomm [unintelligible] so manyyears. but smart home, you don't need a high datarate. you optimize cost power with low data rateyou will be [??]. so you need to think about
it differently. if you have same problem [unintelligible]those guys. but maybe you can think about special thingsjust for wi-fi for smart home. so still have lots of innovation. audience question: hi, my name's karen, andi'm a freshman eecs major. you mentioned we should make tiny progressevery day by doing our own reading; do you have any recommended reading list?[laughter] wayne dai: i met my wife in the terminal room in cory hall. so i don't remember, i studied, undergrad,graduate, never went to parties at all.
so we don't read other major novels; my daughterknows literature well. that's why i have the council, all those things. you can be not necessarily reading technicalbook, of course you can read a book. but you need to broaden things -- the science,the art, also even the business side, i think we need, that's why i talk about business3.0. audience question: how did you go about putting together ... dean sastry: name, name, name... audience question: i'm john, i'm a 5th-yearph.d. student in computer science. i'm wondering how you went about putting togetheryour initial engineering team.
wayne dai: actually i'm the only returnee;my [??] team is all local. i think this first team we form, at the beginning i never recruit fresh graduates, because you have a base, you have the experience. so there's some multinational companies sendthose people to the u.s., sign a 2-year contract, if you leave we need to, have to find themoney. so i pay all the monies, i'd rather get avery [??] team; this is very important. if you start with a team, i have a theory. a team, usually we not hire c team; b teammaybe if they are more open with high a team, but b team, probably not hire b team.
b will hire c; c will hire d. so you startwith b team, so the company you know this [unintelligible]. so initially, the team had to be a team. maybe small is fine. you have to be a team; that's how we determinethe company future. so our first team, i remember, most peoplestill there. so that's very important, how to make. but also i think the mix, i had some othercompany, you cannot student only. you need some mix of students, ph.d. students,also have some people who have working experience,
so you have mix. so i made some mistakes there, all my ph.d. students start company, then probably cannot work well. audience question: hi, i'm alan, a second-yeareecs and bioe major. i was curious, your description of the internetof things, what your vision for it is, namely, do you think there will be a server in everyhome? is it going to be focused around watches inparticular, or are there going to be other large opportunities across the horizon? wayne dai: i think the first wave is wearable,like i say [unintelligible] identification, when you come to smart home you need id.
but second, smart home, there you have smartcars; smart cars are very important. the surveillance is very important, everywhere they have it, but you don't need it with charge, you want battery operated. half year to [??]. that's very challenging,how you can do that, that's lots of energy problem. so it's not just wearables; i think the wearables is maybe the start, with wearables, then next wave smart home, then smart car. audience question: my name is joey, first-yeargraduate student, mechanical engineering. you talk about internet of things; where do you see the opportunities between china and united states, especially in terms of hardwaremanufacturing? thank you.
wayne dai: china has built market. for example, they have wristband trackingkits. only two functions: gsm, talk; they cannottype. five years old, right? but they are one kit; they need social networking, too, so the parents get set, they only can talk to these five friends. so they will wear it, because otherwise theycannot talk. then you can track the location, for safetyimportant. china have on kid, two parents, four grandparents:6 to 1. they do whatever, if anything good for kids'safety and education. china has more kids than whole europe, probably.
that's the marketing, so this can drive. of course you need be low cost, and also veryfriendly. so those things, there's certain things thatshould only happen in china, probably not in u.s., this is the reason. but they need lots of technology from u.s. [unintelligible] use the adis, the mem sensor, the military says very expensive. they willing to pay because lowest power. they got this from adi. this is usually, it cannot be china technology, china costs, it will be u.s. technology, china costs. that's why we have offices all over the place. so we still have more calendars.
audience question: hi, i'm brett, a first-yeareecs student. so we're starting to run up against the physical limits of how much silicon we can actually jam on a chip. do you think that this is going to be a limitingfactor on all these internet of things devices and devices like that in the future, or do you think we're just going to find some way around it? wayne dai: this is a very big question. you know, below 28, moore's law is stretched,i'm not say stopped, because cost per transistor increases, not decreases, for the first time in the last 40 years. you can see we have trouble with the [??] narrowmeter, the finfet very expensive, by the way, the finfet is a great invention here.
we have some alternative technologies, like[??], the other things, low power. but for internet things, is not necessary. for the finfet, is not easy to do analog andthe iphone app. now another thing i like mention, people sayeverything integrate. for now finfet may be for gp or cpu, later2-chip right in 1-chip solution. you don't need to put everything in one chip. maybe certain things, better fresh, have analog,i have maybe another chip. maybe 28 is the most we want to go, ratherthan do finfet. so we need to think about not only soc butsip also, system in package.
more, you need to look at that direction,not just everything put on one chip is the best way. not necessary everything need a finfet, especiallyiot. audience question: hey, i'm miles, fullbright scholar here in eecs, and i'm a ph.d. candidate from the university of hong kong. i just want to ask, you say iot is the nextbig thing, but its cause has been there for like two decades already. twenty years ago, people started talking aboutiot, and there were times when people think that's just like completely b.s., it's notgoing to happen anyway. but now it's coming back again.
but it still is like in a development process. i'm trying to ask, what is the single problemthat you believe, the single most important problem in technology that has to be solved before we can see, really, iot prospering? thanks. wayne dai: anything you cannot do too early,you cannot do too late. some of the technologies -- the power, andall this, the if technologies -- it's not there yet. so now i think at the time, when refresh,if, the cost can drive to, you can make. but wearables, you cannot wear a big thing. front of a watch, the area is limited, youcannot too thick, the battery, the other things. there is certain times, certain things willhappen.
i think this is, even though we have challengesgetting there, we cannot do 5, 10 years ago. like i say, the wearable, the smart phone,of course, everyone knows, you have this smart phone. the watch, people waiting for apple come outthe watch, probably will set some direction. i didn't mention detail like health care isvery important. china has lots of elderly. chinese custom, you don't want to send thosepeople to some senior house. so you have the distance, in-house care. so lots of kids walk outside, you need tohave monitoring sensor, so there's huge applications. this is really the kids' safety.
we need to identify certain needs. the wearables, i think average only wear 9months the most. people don't wear anymore, they say i knowhow many steps i walk. so how to keep people wearing the wearables? this is a challenge. if they don't feel any incentive, why needalways wear them? so if they have the incentive. so those still quite a challenge, but i thinkwe can do that. that's why i think is very fragmented, iotnot easy to do.
some one company, many companies will die,very fragmented. for now the wristband, note, no one sells10 million units, only few million. i think the first company [??] the wristband,will sell 10 million units, if you look at careful how they do this. so this is the marketing point, not necessarilyjust the technology side. i think if you see the wristband that can sell 10 million units, must have some reason why people wear it. dean sastry: wayne, this has been a fantasticafternoon with you, as i knew it would be, a wonderful talk. on behalf of the college, i'd like to presentyou with a small remembrance of today's event.
it is going to be presented by, maybe youshould come in front so you can, get it recorded on. it recognizes our deep appreciation for your commitment to the college as the 2015 kuh distinguished lecturer. many thanks to you.[applause] thank you to eta kappa nu, our student co-sponsors, and most especially to ernie and bettine kuh,for providing this outstanding forum. the whole event has been recorded, and itwill be on the college website, and we'll send ernie a cd of the lecture. thank you all for being with us today, andgo bears!
[applause]
