Earnings call transcript Q4 2023

Due to licensing restrictions, you must log in to view earnings call transcripts.

Operator

Greetings, and thank you for standing by. Welcome to the GSI Technology's Fourth Quarter and Fiscal 2023 Results Conference Call. [Operator Instructions] Before we begin today's call, the Company has requested that I read the following Safe Harbor Statement.

The matters discussed in this conference call may include forward-looking statements regarding future events and the future performance of GSI Technology that involve risks and uncertainties that could cause actual results to differ materially from those anticipated. These risks and uncertainties are described in the Company's Form 10-K filed with the Securities & Exchange Commission.

been being advise XXXX, recorded request GSI also conference that at have of is call today, Technology. I asked you to this the Additionally, May XX,

President, Lee-Lean Douglas are Chief Sales. Financial Lasserre, the is With Didier Schirle, today and President call Executive Officer. the Company's Hosting Chairman, Vice and Officer; Chief him Shu, of

to go ahead, Please would like now Shu. to I turn Mr. the conference over sir.

Lee-Lean Shu

XXXX to our fiscal quarter financial Good results everyone, year and welcome day, and fourth full

earnings call.

filled was progress year fiscal many goals. positive and XXXX The partnerships, new with toward developments, our achieving

setbacks We unforeseen also experienced and delays on several

We a addressable APU. the with about year during the fronts Gemini-I can learned lot the market

given pursue our team, resources. our reasonably with limited

strides opportunities the APU’s we and proving resources resources leveraging of are third-party identify for increasing in awareness OEMs. capabilities. However, significant to us helping capturing These have valuable recently revenue identify in users, made tremendous help resellers, and

aperture and we satellites, our opportunities, We have and such Gemini-I to superior whom a markets We leverage understand with our synthetic help or support focus for solution. these sharpened and prioritize near-term also we have radar, know resources can SAR, our as offering. where and

for application APU has end, vector interest expressed enhanced solution. search Gemini-I engines, our plug-in prioritized this more where we customers performance. our To resources dedicated have that is demonstrated have and focus target Another the in leveraging

to plan have pivot working with SaaS we project search deliverables first to leading players space a and been on data the science team we busy our has Our with one in met provider, partner. once other the

use large fundamental networks, ChatGPT with search focus search future language reason words we trillions the will APU the engines memory This LLM for neural as Vector in natural are architecture today our as are Large an language produce learn on to appropriate or application and is ChatGPT. Looking or roadmap, another work that deep such we vector upon essentially of transformers, billions of a model language models ahead text. versions function and the part to for address is processing. doing APU. build

size are improving AI to and open-source, to recently use increasing sets players, support platform integration use rapid to optimistic by partnerships an with plug-in and launched, several the machine the vector Open driving go-to-market of from APU that about algorithms that fiscal are enterprise search. aspects this our complexity and modest data we search tool Encouraged Additionally, SaaS and The in develop we decentralized search in positive these we XXXX. strategy of key our year more business in We AI like generating for plugin SearchiumAI proliferation of all search engines. by are of engines. potential growth intend this our the reception engines learning other market revenue

Fast applications focus we Gemini-I and that SAR the achieve of have both Search, For just to aim year. we specific I Vector have set fiscal mentioned, goals revenue this

compiler Our progressed the quarter. past L-Python has in stack

is high L-Python either. delivering development compromising designed to while Python's stack offer compiler performance C's advantages Our without

version in a any use or year. require current and C-developer. Python on we demystify have release currently do to focus a beta version for Although our APU later compiler, will are not the production-ready applications track a L-Python this

tape-out for I complete excited that on to to are announce am track by we the Gemini-II this

bring calendar this first and the of end the aim year by summer We XXXX. evaluate chip silicon to

design of provide will market to in the XXXX. second Gemini-II’s significant solution half

power features and enhancements latency. with will consumption reduced performance These

strategic after Search, edge assistance directly. or addressable rather expand larger advanced would market with partner ADAS, LLM, we Fast a vertical than last such systems, driver the and markets Vector to the applications, APU future one the go for a as being

compared or with XX XX TSMC of times memory Meg-Byte second bandwidth associative Mega-Byte The memory data connected bandwidth more AI. of for format state-of-the-art distributed This inherent The X the advantage Byte X XXnm to versus operation, parallel to Gemini-II XXX an memory and is bit flexible X than process. other parallel is power times processing the times is contains of SRAM the Gemini Gemini-I. processors. which fully with built APU processing, density per allows with processor built chip Mega

by of that is affordable allows PCIe together, edge terms, tasks GenX it integrated devices demanding DDRX Gemini-II Gemini-II used the in expensive XX be to simpler a to interfaces network. at package by complete significant solution edge controller while in includes still to processing providing combines while being components and still and less Gemini-II handle used external a allowing for capabilities. a be enough In PCIe This GenX applications different X. powerful

edge. Put data done This the can that center means be brings Gemini-II applications intense another to locally. capabilities computationally way,

For robots, example, satellites. and UAV unmanned and vehicles ADAS, or drones, aerial autonomous delivery

that edge Another application reliable mitigate a wind can losses significant to be requiring operation, processes and example, disruptions. or for infrastructure Gemini-II or like modes farms, to would IoT applications lead operational critical financial for failure efficient

of applications a Fast combination data bandwidth Search, driven for large memory need accurate proliferation Gemini-II’s market and like best-in-class Vector fast power, AI growing big of and the solution and processing for low-cost high the a solution processing. with built-in tremendous provides by

is a granted new block processing high allow perform building in-memory Gemini-II power. were we adder, patent Recently, basic for a to full Gemini-II's to which

explore deliver with announce a early-stage discussions to Provider Cloud are foundational we Gemini-II’s advantages. that performance are to in Service architecture We very how thrilled top currently could

impact and Microsoft large ChatGPT, models this generate year, seen Bard. language, human-like of have that the like language and disruptive Google’s understand BING Just we

As this believe surface transformative impact the We Gemini-II boundaries pushed, versions APU. have Language models language future abundant the Natural have and the can numerous the of we of merely large Processing across in potential the scratched for that continue be of to and they fields. of we envision market opportunities

memory have models adequate high-capacity The as language memory are The on-chip. solutions to XD has memory and Large GPU large limitation but very poor is memory for get language bandwidth built-in very to memory progressing. memory memory attention address access. worse large large models’ state-of-the-art requires built-in bandwidth the requirement going

requirement. to large potentially has architecture the address chip is for add translates a Gemini Gemini magnitude into next natural bandwidth, large to it migration of better orders substantial inherently improvement This chip generation performance. XD memory memory the memory

As be in market, compete positioned the the strongly ahead expanding effectively leading of to we AI competitors. industry's a standing would rapidly result,

Our teams are we on resources probability capitalize high of focused and applications Gemini-I’s to where revenue generating have capabilities. a on

and to new target As market, Gemini-II we creating bring approaching will experienced streams. in more we be revenue customers

which We cater has are to are potential potential. the APU, holds future quite With markets, capability larger promising. APU versions, for opportunities to roadmap the and formulating tremendous the our the much

us exploring our shareholder committed for create in of driving with parallel ahead. We you forward various are and value. for you updating coming our Thank remain sustained fully to I Board In Directors, support options on to and look to innovation growth actively progress quarters. years today. joining in your we the the

further. over Didier, Didier. business the discuss go performance Please I'll Now will call who to our hand ahead,

Didier Lasserre

Thank you, Lee-Lean.

we As on sharpened a revenue focus Lee-Lean stated, have APU our few near-term opportunities.

strengthened Lee-Lean we science team job of addition, primary mentioned. the top a whose is platforms development search for accelerate our contractor data our engine that to with the solution In high-performance plugin

that using AI Gemini-I working begun We have solutions a Gemini-II. also company for and computing offers embedded high-speed custom, with

access for Another improve the APU adding our to market critical has distributors. been development

Radiation for the new added that we hardened Tolerant to market. have our are pleased European announce for distributor APU, a and Hard SRAM, We our and

platform enable is near-term potential on to opportunities. licensing the have our partners. and In opportunities, to focus work build partnerships stages, plan strategic we to us to very we pursue and in to approach do addition early before we a This formally

allow problem system conversations in on or few preliminary said, specific ensure have the communication into we to effective a us Gemini applications solve performance integrate to determining for solution. This a would had their platform. benefits of That partner’s is the required we identify what to another

response, security demoed commercial recently observation satellites designed The infrastructure that to access Gemini-I in for SAR high-resolution provide They customers capabilities technology. on-demand company satellite for provide specializing customers monitoring, and national worldwide. flexible, are imaging We and can a applications. imagery to government private Earth disaster

conduct market benchmarks that running generate we provided commencing SAR with and on revenue can recently year. anticipate of benchmarks. we we comparison this to is the They fiscal those one the Gemini-I are process datasets modest Gemini-I,

contract. Phase government Business be States or could projects needs. for small was R&D business government X recently that United SBIR, commercialized a is specific supports awarded a that GSI Small SBIR program Research, Innovation

For computing in with edge problem we the with collaborate space will address to this Air of contract, the and Space Gemini-I. Force

military Gemini particularly will already making I well-suited for a This as government APU’s is great for references similar and other Technology, is contract agencies and capabilities it radiation-tolerant, for GSI Force applications. milestone the showcase the missions. Space provide it for

submitted proposals and pipeline. We direct-to-Phase other have project proposals the X SBIR other a have in for

is applications. to to up a Gemini-II months as and awarded morning that $X.XX Once will and would achieved, we Development milestones GSI’s XX we two release recognized finalized we Missions contract, integrate timeframe which verbal are just press with to Force has been note, be On years. for this could worth issue that This received the Research and executed, million, Air the been revenue Space next-generation confirmation be have a details. full agreement a and typical

fourth fourth compared of shipments customer in $X.X fiscal net breakdown quarter million, sales prior XX.X% prior of shipments XX.X% compared shipments were In fiscal the quarter the revenues, in in XX.X% the period quarter of quarter switch XX.X% to same $X.X me XXXX XX.X% in and Let to of the in XXXX, million, Nokia revenues, compared SigmaQuad the and XX.X% year were fourth the to of prior period and XX.X% year now quarter. or comparable the of or or shipments Military/defense to quarter. net a sales revenues, the of $X.X ago product ago net were a and XX.X% of fourth in XX.X% the of to for quarter. sales fourth million, quarter.

like Go I'd over call ahead, hand to the Doug. to Doug. now

Douglas Schirle

Didier. you, Thank

of review diluted on $X.X million, in will XXXX quarter for on quarter Gross I fourth loss a summary, the compared or preceding fiscal reported $X.X XX.X% the net quarter of results. quarter loss and fiscal share, revenues fiscal full-year third XX.X% the compared revenues in of of of for GSI the of fourth results XXXX. with quarter $X.XX of fiscal per $X.XX fiscal quarter. a $X.X diluted in of net a share, million net loss XX.X% million, revenues net XXXX, of the of diluted and or million, for to the by prior-year the start or million third $X period was followed $X.X XXXX a per fourth the net to $X on XXXX margin share, of net million fourth $X.XX per

effect fourth the in million goods. of and quarter cost of quarter on Total were revenue decrease in the fiscal in of The margin the fourth quarter XXXX gross prior primarily quarter. prior-year expenses was $X.X in the and operating in of million the due prior quarter. $X.X the to our costs XXXX were compared lower of the XXXX to fiscal fixed and $X.X fourth development expenses to million $X Research million, in in period million $X.X compared million, $X.X the in

XXXX, million the operating to XX, administrative the million and in in in general million quarter $X compared expenses period operating previous Selling, prior quarter was loss the of March quarter to compared loss ended and quarter. XXXX $X.X million $X.X and the in fiscal the million quarter. an million were operating of $X.X an $X.X prior-year loss prior-year Fourth $X.X in

and included preceding interest expense tax fiscal $XX,XXX a net interest third compared income ago. $XX,XXX a and for of $XXX,XXX the included provision net same of and quarter, a $XX,XXX income a provision of of interest in other and loss period $XXX,XXX, and other loss quarter tax XXXX of to In $XX,XXX. year provision other tax Fourth the and

$XXX,XXX quarter. and stock-based was to pre-tax compared year the expense a ago compensation quarter in comparable fourth prior in $XXX,XXX quarter Total the $XXX,XXX

$X.XX increase in XXXX. ended March on share, primarily XX.X%, net XXXX on margin share, for net revenues the reported $XX.X to or XX, year. loss was prior a million, net XX.X% net revenues fiscal the or $XX.X fiscal $X.XX compared per For of gross in ended million, mix. to fiscal XX, margin the million, of year $XX.X compared year Company XXXX, was in per diluted Gross loss March of diluted to product The due of the $XX.X million a

in compared XXXX, and operating in fiscal compared prior $XX.X XXXX. year. fiscal million million, expenses the $XX.X to $XX.X were Research Total expenses fiscal million million were $XX.X in development to

interest was primarily compared Selling, in XXXX. net XXXX and in and November the fiscal $XX.X and in administrative million, to by cost $XXX,XXX, The a the included of compared fiscal general income $XX.X tax were loss in other benefit for of $X.X research XXXX Company The of expense year. a loss a prior XXXX. million expenses provision to expenses to operating in The $XX,XXX measures an the $XX.X announced compared decline and million operating $XX,XXX interest was fiscal tax and ago. development and loss $XXX,XXX of year to million due other reduction

million March Company had long-term no and $XX.X $XX.X million versus was $XX XX, as to March investments At fiscal no short-term XX, XX, of XX, the and million March compared at cash, capital as in XXXX. cash XX, compared of and XX, investments XXXX, the debt. $XX.X to and Working million investments in million ended XXXX. equivalents, XXXX, of XXXX, at Stockholders’ with cash, $XX.X March March as XXXX, million long-term in was equity million $XX.X March cash equivalents, $X.X investments, short-term year

Q&A. to call the open will we point, this at Operator,

Operator

with from question. proceed first [Operator question Needham. Gill Instructions] And Please Raji the line your of with the comes

Nick Doyle

Doyle Raji Thanks. volume outlook? in questions the and then expand specifically Gill. Nick Two and could ADAS And the out in the you're seeing Gemini-II. then of language anything costs for large current contemplated to in what traction This the using with testing models? kind on then Gemini-II, applications is that related you all tape on your Are production on

Douglas Schirle

Yes.

of Gemini-II. R&D spending most yes, we're spending, on terms is of what In today

the be we've in expenses We Sunnyvale about fiscal than what quarter. have the And a recent there Israel. in be half and to seen hardware software It probably the R&D the XXXX million. will the in tape-out most team Gemini-II. run here for $X.X of will in Other that, should first similar team

Didier Lasserre

And applications, Gemini-I cut regarding or Were the out. talking you Gemini-II? you

Nick Doyle

please. Gemini-II,

Didier Lasserre

Yes.

So Gemini-II.

technology, discussed far we that. that as we want most do models, discussed, there. as we is the feel technology applicable large So be address. in as to And will certainly the in use the we'll Gemini-II, Gemini conversation advantage we partner ADAS, before, something we But language the likely to as certainly the a

customized be whether it's And the with Gemini-II is or with Gemini-III so start a to it's if also determined.

Nick Doyle

then Okay. a hard EU? Is the map guys sense. if timeline? there timeline you product in you And That there did makes mentioned say you a Did road rad the one. quick just a is for -- the for

Didier Lasserre

hard and rad-tolerant SRAMs are today. rad available The

the years. It's boy, bit We testing was beam -- been very that the Gemini-I a have APU. testing. But We -- some came done did was X.X specifically little day. on off it's back at least favorable. the it

could So it do. was limited the test we

full testing of this actually are we half in complement the radiation So going second to the year. of do

we sending data will So all the have for requirements the that folks be it into space.

year. APU the this sometime end rad-tolerant be So officially, will by the of

Nick Doyle

Excellent. Thank you.

Operator

And the question line question. your Jeff next Cowen. of the with Bernstein TD with proceed from comes Please

Jeff Bernstein

guys. Hi,

Just me. for questions of couple a

heard Just make to wanted I right. sure

a that You Gemini-I. target helping brought for on Is that's applications consultant right?

Didier Lasserre

the out for some They're search are specifically write that interfaces of fast helping the us there. vector platforms

Jeff Bernstein

Got you.

right? getting some is said am there's solutions that high-performance into now AI Okay. custom that going and you compute then that -- for integrate I Gemini-I Is clients. And a embedded guy supplier, solutions to

Didier Lasserre

Partially.

limited applications they're ranging it And to of have Gemini-II. applications from Gemini-I. a search SAR There's to not It's rescue. it's and they sell of looking multitude different lot a to for. at potential So applications different and that marine, at Gemini-I

for of the those too these our be essentially boards. many for considered cases, Some ultra-small their use boards little be some will to boards big they'll But that of a developing our able cases, are Leda they applications own in applications.

both applications, Gemini-I multitude So will it's of and a and be different it for Gemini-II.

Jeff Bernstein

Got you.

I large just low is opposed Okay. run potentially. there's matrices as queries of train language me and quickly at large Correct queries kind far just two to as And of to run power. if these One as I'm and then wrong. model the applications, think

we to being Is efficient other do has do today as by one here day? right? of the and able more matrices not the And training learning. to making with light are redo over I some about And that again over guess you new talking having which

Lee-Lean Shu

Yes.

short the are don't can if the training Our okay? can we you the search okay. now primary shop you train set. the training, We help do to But which Like you search, dual on efficiently, or can training, training to even target will which need means path, is single increase data do path, okay? be

store you have item That's first And that you chip. different training, don't memory come our consuming. in whole can the it recognize again. have That's run the very, coming in time we traditional into security very whole very a to similar set, second from If recognize, right time, over we can training, Traditional okay? away. you data model,

can the do tremendously. training, shop have you do save that, can you the So part you if then capability training dual to

Jeff Bernstein

thank great, you. That's

Lee-Lean Shu

Yes.

Operator

Partners. Hirschman And from Orin with Investment line the next of AIGH comes question the

Orin Hirschman

are how you? Hi

Lee-Lean Shu

Good.

Orin Hirschman

world Gemini So you NLP a And ChatGPT is actually the and very one first at, tremendous all processing best. to architecture interest exactly whole got that good of things architecture has situations which changed, what wasn't similar the was language time, Gemini, it introduced processing. other and this this really in like do you've memory of things you the natural types when fits in where

it of drivers prior having So Gemini-I question of and the for the written actually use to code I guess from sounded on last the application? be able like certainly to that this you're and one comments Gemini-II optimize

GPUs, the like in on simple to where is simply because a something I NLP boards sell acceleration So terms on - order a where is GPU in they're not the ChatGPT. think you side AI that just is lot applications one could of where having an would signed this of difficulty everybody accelerate not using of

Lee-Lean Shu

What's the again? question

Didier Lasserre

Yes. What question?

Orin Hirschman

with partners Gemini a is you're APU? in far introduce that the a terms a do great It what or boards own isn't fact, able it to is, is, certainly - it's the for that in so application question the sounds great of be on to like call So your on application? fact, in working priority is to that acceleration

Lee-Lean Shu

Okay.

my I for model, The I discussed challenge think the this large two-prong. language on biggest are statement.

a First memory. very you one, large need

memory. The okay? has solution. things two a good market It's solution. to - second I achieve, one, this Those difficult think high-bandwidth very you in need a are at the market the nobody that very today

Okay.

memory exciting Gemini okay? So from have very to we forward, thing just bandwidth, see with can architecture a as cloud very, mentioned, it okay. do good - to provider have how call to can see help service discussion we how our I larger help this we already and a very a move We foundational we

memory in my times mentioned I okay? explore if the to why nobody capacity, can payroll the then very that's state-of-the-art So okay? are That's try I okay? which go today's So, okay in from inherent we to we this or processes, have we're GPU and here. And bandwidth excited. XX can over high now we advantage memory we statement one lead can where market see provide, say add We our this architecture,

Orin Hirschman

type Any ideas can to talking Gemini-I what that's be that the when When acceleration the maybe of something or customers some you will code So and code to with see interface is mentioned ready able until simulation gains something to the like even ready. a ChatGPT working a you - type that? be demo that we're about with for will on? benchmarking or Gemini-II coding actually of you're like

Lee-Lean Shu

The they Yes, yes, have yes. OpenAI, plug-in.

Okay.

then can to your utilize the plug put in. the can software you do basically, in plug and the So then you machine existing model main to and

it. we right are working on So now,

are and least working And Gemini-I follow to the on. you how well Gemini-II extrapolate those presently future. second, that the to And extrapolate can Okay. at from and

Orin Hirschman

Any we idea coming in some might when benchmark see months?

Lee-Lean Shu

to it? to or guys. tell How Maybe you something have will quarter a two, we was

Orin Hirschman

of even what number even movie, as intensive, same a able early belief for with talking level suited a I of And more a in to, anybody computationally like taking also language do. terms similar just question doing about in something beyond can of question in let's project what futuristic. but is and that? And to muve but on Okay. projects. off There's visuals it is anything well? processing doing doing well natural had be Is you guess do show my related you guys of And to you that terms an and visual fact, say, are there talk

early did current a But of you step of Obviously, people. today. obviously, you impressed that do demo, can't a that But people what have architecture. using lot even dreamed that's almost beyond which what

futuristic or question? Gemini-III Gemini-II? So and any on one that follow-up perspective thoughts a in from be a can then will And need last that that done

Lee-Lean Shu

whether to generation asking we future want in you're how do So is the that? -

Orin Hirschman

No.

almost Specifically, the like more on is in capabilities search terms of just important incredible part capabilities, me visuals. to visual search NLP

impressive to other amount You then movie And over is projects world early of all - on do massive some people experimental with the did experimental that demo starting and visual data. projects.

suitable projects thoughts you that very interesting visual anybody more a obviously for suitable movie or project? And NLP Any for as type that GPUs more that in you I Gemini? do other know search, futuristic, terms versus uniquely what using just or that's for like using bit - just matter. to of has know visual but it's search of done search comparative massive Any visual

Lee-Lean Shu

with our Yes. we at - solution we we I look advantage mentioned We - workload, look that's can than partner, be okay? exists saying, have enough be we have any why hey, extend our we there. the memory, architectural advantage So Gemini the if faster at architecture. today, We we inherent XXx one - are will the just - of with this

memory that's the okay? better than if why it, enough thing enough solutions. you we all we for memory So, can looking the for built-in if road magnetic the so is put have map, performance combine future that, But into for are can don't existing we

Orin Hirschman

On a question. that closing note,

memory discussed just just me? so terms terms in and one used can of then and Gemini-III. geometry what affect what in the et that you that? being will obviously, from us for if Gemini-II cetera, ability it kind that's you in Just tell of Thank what about pack much. follow-up, Gemini-I, for is of And you're more thinking to of nanometer the - and And you

Lee-Lean Shu

build at sales Gemini-II we If future and look because memory - the X-nanometer. way is look GPU a today's Yes. is and get X-nanometer is then the okay, memory is Gemini-I framework our only XX-nanometer, Today, XD then reasonable can do XX-nanometer. in future, we at we And a we memory. and the a you with in if the there, XD high we capacity

put a X-nanometer, we if of the XD in we'll So, be better. with memory order magnitude

Orin Hirschman

the understanding - terms but here with I knowing of Gemini-II the coming going Gemini-III, mean, up the question that follow-up key is in that to is shortly, be this So platform. platform

In ability NLP terms your of futuristic question? that accelerate forget to about non-visual again,

terms in in optimized and GPU, of you today et question? you so superior last here edge NLP it even ChatGPT, Hopper-style That's of accelerating GPU. much. Gemini-II? to even like is, applications on Have with only leapfrog that can But GPU further? so cetera, my that Thank And type is enough it application that question Gemini-II passed you got flash competitive leading your the

Lee-Lean Shu

As memory capacity okay? them. We I big big mentioned things, memory there have or are bandwidth, two of one

have that, are best if there. a cases be the Even out humongous our There fit many, it So the like doesn't any chip many will workload be into solution okay. to ChatGPT, can dataset.

Okay.

it dataset. be smaller So a can

And so, dataset into chip. can our fit the

in will the - the number We be market. one

Orin Hirschman

thank much. great, so you Okay

Operator

of question Instructions] Private is Bohn], Please next [Operator proceed the question. with Investor. The from your line [Luke

Unidentified Analyst

Services clarifications and fielding, a to integration well, be play cases processes hear server could use minor without just near-term around of less end to the hope just development. like broader of for of complex more Amazon more companies, that potential you announcement bit with is of or or plug Web little having and really capable very comprehensive all kind range through with good your exciting offering processes, play? go other embedding the and back, there? say, potential. layout service Great Wondering if going a broader all Just your Hi, to

from do what also for that. their Yes facility, advanced applications, to mentioned, offering data, the of relate rollout you're things own you wondering potential seeing registration, Azure how your a earlier on say, maybe but like rich for how servers like own expand that about so your the especially and Gemini-II Amazon for similar to can or thinking maybe Web Services clouds on metaverse, search, other or cloud? especially vector an And was would

Didier Lasserre

in with So, we've - the started past, we've we've started Search. discussed integration as the Open

And we - already and set our have so It's really ongoing. servers for own that's that. up

facility facility, facility We directly AWS West, some in street in that's have across directly we Israeli the at our also have here and and it's some then some an Sunnyvale connected. our off-site from

in with the that place we've Gemini-I. So had

Gemini-II. Over time, those to would we migrate obviously,

So those run. We run have people of can that be able in set sets your SAR up do some data. to place. which can do yet there, that that in not remotely. the off - demos you to own already are those are data It's It's load

And so we're least not point datasets. at you larger at the your own not yet data, where can enter

we the yet. quite So We're - but not there we're certainly is going. that direction just

Unidentified Analyst

Do roll would a on when you have timeline you be features and capacities? able to out those interactive

Didier Lasserre

year. this for shooting We're

going is some now. to have the this that board help - data up examples on of be science we're going from get to brought Some we contractor you

we're this to something year. it's of rollout trying So second half

Unidentified Analyst

yes. All so appreciate yes right, I Thanks that's [indiscernible] it, much. have all

Didier Lasserre

Luke. Thanks

Operator

of question is a And Cowen. the with line next Jeff the follow-up TD from Bernstein

Jeff Bernstein

Elta SAR application, can and there? wanted just see on give update if the on going what's an to us I you

Didier Lasserre

Yes.

much them, was recall, broad POC. we from you did it It could POC be or much, multitudes into as XXX So for higher, used with and to of of be the vessels. very a space. different meters a used obviously, lot It at vehicles heights could

using GPU. they for And And program been you had could they're call guess at it us already laptop, so, a the looking was I just that. single a initial

program. GPU for using And that the so they're still

looking There's a now. form us program they're for at that

so, going And we're through that process them. with

than of algorithm, the same POC, done one, bit and but it will simple we've be So of a with it still because will a under be won't it working already a it on - a it will are kind be But we will should what SAR, them. another be integration. be different still be it project so

Jeff Bernstein

scrubbed? And just now some waiting have it might on visibility SRAM wondering Rad-Hard that if or about then permanently when get Okay. the launch guys wondering and happen to you any on hopefully space providence is

Didier Lasserre

permanently No, frustration you get it's - one. because scrubbed. not I'm on I this your with We follow-up

scrubbed. they at contractors were So of a talked, scrubbed. we have There was I larger been certainly scrubbed, say programs They're when but they, defense been that a There the it's multiple we're have have as using couple there. the that it. we're been out there not ones - still not looking few programs

were been we're components. now we get Originally, a it. they've It's And to the out some be. exactly getting matter launch critical delayed, them it's - getting And they is just not good do going when feel actually of pushing dates. just a couldn't the for know next launch because just to just they

the It's still So, to the is it. just we're is timing on optimistic it's for going us when actually about elusive answer happen.

Jeff Bernstein

the or on distributor of in Rad-tolerant - so anything European And Rad-Hard that get Europe? piece can a or just until with approach the you doing they is province stuck space do different are kind it

Didier Lasserre

they're going everything. be after going to No, definitely

So the says the just were already space, that the sent really world get and folks have into your is looking work. heritage part. to And heritage that parts they the parts to to that signal been we've it's heritage, a just launched

are the lot mark of it's additional up fact And for and an work internally so that it's it parts these already check qualified a But folks. really to our there. doesn't a change - box

will work they the have So, know we done. of testing we off based that

short-term we this for do heritage So folks be opportunities those additional companies. that us. I mean, the launches, to European to looking finding is the were distributor were U.S.-based the for going

least not to customer, some they were a the European ones and anticipated Rad-Hard to get but one we heritage. shipped at have us initial the We Rad-tolerant

Jeff Bernstein

pharma, customers the Institute great. scientific any analogous Okay, med-tech, Weizmann then And for universities, biotech, in any boards back type cetera? all et or update on come of more right, some is applications

Didier Lasserre

on so for opportunities revenue today yes, candidly lot time are market. universities There are other - we're a we've are The that not for Universities, at think yes, discussed of We that different spending two. two them me genomics. that larger. looking the universities two have let applications different markets do - two for

the Biovia, - Tanimoto. done be be the we not specifically will plug-in, personally, But doing they ourselves. write-up. We've already doing are so and And for algorithms much, a they'll spending the running efforts essentially

revenue for we're for limited as to time are just And spend they more markets make so, more doesn't off great resources just developing in as platforms. it algorithms aren't more there us based addressing. of sense our other volumes The

Jeff Bernstein

sense. sense, Make make Thanks.

Operator

turn are to this presentation hosts. no the I at will There back the time. further questions now

Lee-Lean Shu

for all speaking to fiscal We when XXXX you look forward Thank you. our again we joining with you report Thank us. first quarter results.

Operator

your That ask please and does for today's you conference. disconnect participation your conclude you thank line. We that

GSI Technology (GSIT) 16 May 232023 Q4 Earnings call transcript