FROM
To
ce
ace
DATE =
TIME
SUBJECT
ATTACHMENT
BODY
Geof Tate
srgrarGroris on acronis con ogeprhranta on betQrut,
geabarrogrambus,com,stark@rambus com, amb com, jreman@rembus. com,
Bimarket@rambus. com na
sansng07
12:38:57
samsung exec meeting and high stakes poker 12/15/07
{i met for 2.5 hours today with
~ywlee, ceo semi biz
lk tim, gm dram division
=98 choi, vp sales & marketing
+ ey chung, works for choi: dram marketing manager
(unfortunately oj chin, sr vp logic, called in the morning to say he
‘couldnt make the meeting afterall -'l have to get him in
Januaryffebruary).
they are very pessimistic about the near term outlook for the dram business
= DC says pc growth in 98 will be less than in 97
+ supply >> demand
« the growth in MBytes/PC is slower than expected as more PC's are selling
at lower asp's with less memory than expected
= maybe 99 wil be ok, but 98 will be very bad is their conclusion
they asked about cirus bugsifixes/sales.
=i said they are dropping 2-channel but they wall do integrated multimedia
chips using rambus forthe low end instead
they fixed the clock problem for HP and are working on the ix for 6OOMhz
‘concurrent: should have in qi
{or total graphics controllers sales in 98 using rambus | said about
4MUnits for cirusichromaticls3.
{told them we expect to sign with amd soon for main memory deal and are
‘negotiating with cyrix.
concurrent
“they said they had to scrap a lt of wafers
= they tink they are the ONLY one developing it
+ sald neciigMyundadoki have working 600mhz concurrent now and toshiba
is developing sl; bu samsung is probably ahead on production ramp
is not alone
{ asked about sdram pe-100 ac problems - hk lim sald theyd be'd fited very
‘000,
they asked about subtie problems with controlle's/rdram's using rambus and
\What are we going to do about t
agreed there have been problems.
when a system customer has a problem they always call us and our job is.
to.quickly identify which component doesn't meet spec and then work with
that company to fix the problem
+ fa controler has a problem then for the time til they cam fix it it
‘might be they can use certain rdrams and not others because maybe some have
‘more margin than others
«i said our plan isto invest more effort in valdation/checking specs at
the module/component level RH 387808
R 235314
OUTSIDE COUNSEL ONLY
‘cx0981-001Booy
pointed out that with samsung we have had a hard time even geting
‘rams to check for specs: dr. im indicated they might be wailing fo be
‘more cooperative f we have a program for reducing sysystem problems by
‘working with all component suppliers
=i pointed out that these kinds of probiems exist (a lt from what i hear)
TODAY with sdram, especially sdram-100, but atleast with rambus the
systems companies have someone to resolve the issue: who does the system
customer go to for sdram problems?
i flagged them on rdram that their die size is >>10% over sdram and they
‘should be working on their shrink to get <10% through changes in aspect
ratio, m-2 pitch, etc (and that allen would have more details).
they said nec and others were same/bigger. i said there were others <10%.
they assumed they must have big die. 1 said the ones smaller often had
‘smaller die for dram and sdram both. and that the reason they were
‘smaller were better (in retrospect) choices on layout/aspect ratio and
better m-2 pitch among other things.
tus out they interpret our push on die size 2 a push on their price. yw
lee said rambus must be worth more since tis $0 much faster. agreed
‘and pointed out the market sets the price, not rambus. sof there isa
market rice, samsung wil make MORE MONEY rf they have smaler die
‘overhead than if they Rave bigger. and i was worried they were rejecting
Cur arguments that a smaller die overhead is achievable. i think maybe dr
lim took some notes to go work on die size reductions. he asked f we
could tll them spectically why they are at 15%: | said maybe, but they
‘needed to give us specific layout information on the dram AND the
Comparable scram s0 we could help them allen, you should follow up on
this at your meeting.
{for 128Mbit drt thinks rambus overhead would be around 10% (vs 15% for
64M),
{95 choi thinks that in 1999 for production the 64Mbit will be cheapest
Ccostbit. 128mbit won't be cheaper til 2000 he thinks.
(Gr lim mentioned that even now their sdram yield is 3-4% worse than edo).
| encouraged them again to do rambus logic. they said they had an option
in the contract. i agreed but said they needed to EXERCISE the option and
{get going. they asked what royalty rates: | said we'd give them a good
‘deal like everyone, 3.5-4%.
{reviewed our analysis of ddr and said it won't work at 100mhz for main
memory. they did not disagree. yw lee seemed to be agreeing when itaked
‘bout the problems even 10Omhz sdram is having. my sense is yw lee sees
ddr a8 underwhelming. i said its absolutely no competition for direct
rambus. i said maybe some companies would add “ddr support” to chipsets
butt wouldn't work at a system level. they did not disagree. 9s choi
said they have been Keeping a “iow profile” on ddr “ke we'd agree" at
‘our last meeting. i pulled out the double bunny ad and said they have
NEVER run a rambus ad, this sure didn't look low profile to me, and they
wore wasting their money promoting a solution thal won't meet what they are
saying.
POKER PHASE: playing some poker at chaminade with mattled/davelete was,
helpful. basically chol was chief negatron and bitching about everything
R 235315
rH 397809 (OUTSIDE COUNSEL ONLY
‘cx0981-002so0y
‘about rambus: tons of $ spent, no return, no market, ete ete. rather than
trying to appease choi i decided to react frustrated and that we couldn't
work wi imsul marketing and had gen up doing so: i. cling his
| gave them the same warrants offer as all the dram partners (and told them.
itwas the same as we are offering everyone).
{9s choi then launched off saying he expected us to come back at this
‘meeting with better deals on royattyetc to motivate them,
{said (and believe) that we agreed last meeting that HE was supposed to
send us a counter.
he went on and on and on about how could we expect them to get motivated if
they promoted rambus and every dram comapny benefited, why should they
promote, etc. etc.
‘told him the reason is that itis what the customers want and it offers
highest prica/performance. NO commodity dram exists where samsung has any
Sort of motivation lke what he is asking for for rambus. if they promote
‘séram, the whole market benefits too.
the real advantage for being early isto gain experience, be a strategic
supplier, get early quals at major accounts, etc. same for rambus as for
any dram type.
told him we have given up on trying to motivate samsung marketing group.
we have spent years trying to motivate them and we always get the same
rejection. s0 we've decided itis hopeless. not even 0% royalty would
‘motivate him. 3+ years ago we gave them the best dram deal in retum for
best efforts marketing and have received ZERO marketing support so far and
have concluded we should expect nothing more in the future.
‘50 we are putting our effort into winning designs at major customers and
doing very wel. and in co-markeling with those dram companies that are
helping us, not hurting us.
after much more railing by gs choi he said he understands we don't want to
change our royally. i agreed and said we've given them the best deal and
cant give them a better one. and that even at 0% i don't think fd make
him happy 80 why bother anyways.
| asked him to make me a specific proposal. finally he talked about
intets warrants of 1MU - they are 90% of mpu's; samsung is 25% of drams,
‘80 we should give samsung a pro-rated amount of warrants at least, say
250K-iah. i said i understood his logic, BUT a HUGE difference is that
intel has committed to switch their customers 100% from sdram to direct
‘dram whereas samsung has made no commitment and seems to be working
against us. i asked him if they would consider making a commitment to drop
‘ddr and announce it and stop promoting it; and drop planslactvities for
any competitive dram ike sidram; and switch their customers from sdram to
ambus period. he said they would consider.
i said maybe the board would be motivated by his proposal and that fd
‘check and let him know ealry january; and that they should decide f they'd
do itor not.
R 235316
RH 387810 (OUTSIDE COUNSEL ONLY
¢x0981-003