Lifting Full Year Outlook; Maintain Buy w/$340 PT

Cerebras SystemsInc(CBRS.O) Lifting Full Year Outlook; Maintain Buy w/$340 PT Cerebras delivered betterthanexpected results/guideon its first earnings asapubliccompany.OurCY26estimatesmovehigheronimprovedfullyearrevenue outlook and better gross margins.Our CY27/28 estimates areroughlyunchanged as we await more details on the Aws deal with salesexpectedin2027.Cerebras'coreedgeisultra-fastinference~13xfasterperthe company which positions it for fast tokens-a key driver of Aleconomics,productivity,and even safety.We maintain our investmentthesis that Cerebras has the first mover advantage in the fast inferencemarketof$130BTAMby2030andourpricetargetof $340or15xEV/Sbasedon~$7B2028corerevenuesremainunchanged. Price (23Jun2616:00) US$226.72TargetpriceUS$340.00Expectedsharepricereturn50.0%Expected dividend yield0.0%Expectedtotalreturn50.0%MarketCapUS$49.790M yet readyto sizetheopportunity.Management explainedthatCerebras just movedfrom a letter of intent to definitive document with Aws in record time,butAWs.WebelieveAWshaslikelyaskedforlargemanufacturingcommitmentsandthecompanyis waiting to receive a purchaseorderin thenext 2-3months. FastInferenceDemand-Management described thefast inference demandexpanding outward from coding (described as the"nucleus"of the market) intodesign-citing Figma as an existing winwith moredesign-space customers in thepipelineand further into agentic Al workflows,which arewell suited totaskinference because they involve many sequential, latency-sensitive steps.Theagenticusecaseswillspanmanyverticals,includingemergingapplicationsinareaslikeshopping. GrossMarginUpsideCerebrasattributedbetter-than-expectedmarginsto:stronger pricing,and more conservative assumptions around the timing andeconomics of systemrent-backsthanthecompanyheld justafewmonths earlier. Atif Malik AC+1-415-951-1892atif.malik@citi.com Papa Sylla+1-212-816-9476papa.sylla@citi.com Mar-QEarningsReviewMarch-QResults:CBRSreporteditsfirstearningsasapubliccompany,andthe resultsshowed strongtop-line.GAAPrevenuecameinat $193.4M,up94%yoyand 13%qoq,whilecore revenue was $191.3M,up 92% yoy.Within that,corehardwarerevenuewas$111.6M,up60%yoyandrepresented58%ofsaleswhilecorecloudand inference services revenue was$79.8M,representing a rapid167%Coregrossmarginforthequarterwas47%,andthecompanypostedacorenetlossof$2.5MvsCiti's-37Mexpectations. $865M,implyingabout69%growthatthemidpoint.Whilethecompanyalsoguided core gross margin down sharply to a range of 38% to 41% and coreoperating margin to a range of negative 28% to negative 32%,both came above Citiexpectationsof~30%GMand-44%OM.Themarginstep-downreflects thecostsofstandingupcapacityforitslargenewinfrastructurecommitmentsasthecompanyinitially leverages one ofits clients'Cerebras servers tofulfill the near-termdemand.Nearterm,CerebrasguidedtheJun-Qcoresalesof~194M(7%above Citi's $181M). Onthedemandside,thecompanydiscloseda contractedbacklog,orremainingperformanceobligations,ofs24.6B.Cerebrasalsoannouncedanewmulti-yearpartnershipwithAmazonWeb ServicestodeployCerebras inferencehardwareinsideAwSdatacenters,whichgivesthecompanyasecond largedistributionchannelbeyonditsowncloud. Withthatinmind,we continuetomodeltheAlinference chipmarkettogrowfromlanguagemodelsand AgenticAlas enterprisesfocus on real-time GenAldeployment and hyperscalers expand infrastructure to support compute-heavy,data-driven decision-making.Webelieve Cerebras competes inthe subsetofinferencestackwherelatencyisafirst-orderconstraint(fast-inference),asegmentthatisitselfgrowingfasterthaninferencebroadlyasreasoningmodelsproliferate.We expect Inference workloadsto aggregatemorethan60% ofall Al computingfromtrainingtoinferenceas deployedAlapplications scale(seeourrecentinitiationreportformoredetails) driven actions (e.g.,automated guardrail checks)requireadditional computebehindthescenes,andthatspeedisthedifferentiatorregardlessofwherethatcompute runs.Theyuseda VPN-setupanalogy fromthe 1990s to illustrate thatneed tobefast and invisible-whichhelps Cerebras's opportunitygiven itsinferencespeedadvantage. extendthe valueoftheir own chips. explainedthataccelerators onlygeneratetheworktobedone,whileCPUsexecutethe actual actions (e.g.,web searches,tool calls)and handle orchestration.ThisiswhyGPU-to-CPU ratioshaveshiftedfromroughly8:2toward4:8 and evendiscussions of1:1,as agenticworkflowsgeneratecascadingworkforCPUstoexecute.Managementframedthis as evidencethatagentic Albenefitsthebroadercomputestack-includingCPUsandDRAM,notjustaccelerators. CerebrasSystemsInc intelligence (Al)training and inference.Its products include inference,Alsupercomputers,Almodel services,cloud,systems,andprocessors.ThecompanyisheadquarteredinSunnyvale,CA. Investmentstrategy thefastinferencemarketof$130BTAMwithinthetotal$650BinferenceTAMby2030.We expect Cerebras to leverage its unique wafer scale technology,2-3xtokenpersecondspeedadvantage,and$20B+partnershipwithOpenAlto achieve 40-50% of the share ofthe fast inferencemarket. Valuation We