Nvidia deal could change the chip world

San Francisco Chronicle Late Edition - - BUSINESS REPORT - By Don Clark

Nvidia plans to ac­quire Bri­tish chip de­signer Arm from SoftBank in a deal worth about $40 bil­lion, a move that could re­shape the bat­tle over tech­nol­ogy that pow­ers smart­phones and data cen­ters.

The Santa Clara com­pany, best known for sup­ply­ing chips that ren­der im­ages in video games, said Sun­day it would pay SoftBank a com­bi­na­tion of cash and shares in the trans­ac­tion. Nvidia’s mar­ket cap­i­tal­iza­tion has sky­rock­eted to more than $300 bil­lion lately, partly ow­ing to re­cent suc­cess in ar­ti­fi­cial in­tel­li­gence ap­pli­ca­tions and pan­demic­fu­eled growth in chips used for PC gam­ing.

If com­pleted, the trans­ac­tion would in­stantly trans­form Nvidia into one of the most in­flu­en­tial play­ers in smart­phone tech­nol­ogy, a mar­ket that had pre­vi­ously eluded it. Arm, which li­censes de­signs that other com­pa­nies turn into chips, has long de­fined the com­put­ing tech­nol­ogy found in most mo­bile de­vices. And Arm de­signs are start­ing to play a bigger role in cloud data cen­ters.

But the deal is likely to prompt close scru­tiny by an­titrust au­thor­i­ties around the world. In­flu

en­tial Arm cus­tomers po­ten­tially af­fected by the trans­ac­tion in­clude Ap­ple, Sam­sung Elec­tron­ics, Ama­zon.com, Qual­comm and Huawei.

Arm, which was ac­quired by SoftBank in 2016, is widely per­ceived as an in­de­pen­dent en­tity that gives equal treat­ment to all li­censees. In­dus­try ex­ec­u­tives and an­a­lysts have pointed to po­ten­tial con­flicts if a com­pany re­ceived tech­ni­cal as­sis­tance from an Nvidia­owned Arm that could give its own chip busi­ness un­fair ad­van­tages over other li­censees.

Jensen Huang, Nvidia’s chief ex­ec­u­tive, pledged to keep op­er­at­ing Arm as it has been.

“Arm’s busi­ness model is bril­liant,” he wrote in a let­ter to Nvidia em­ploy­ees Sun­day. “We will main­tain its open­li­cens­ing model and cus­tomer neu­tral­ity, serv­ing cus­tomers in any in­dus­try, across the world.”

Nvidia also said it would keep op­er­at­ing Arm from Cam­bridge, Eng­land, and honor com­mit­ments SoftBank made to keep in­vest­ing in Bri­tain. Huang said he and Si­mon Se­gars, Arm’s chief ex­ec­u­tive, had al­ready held ini­tial talks with Bri­tish of­fi­cials, who were “de­lighted” with the com­pany’s ex­pan­sion plans.

Un­der the terms of the trans­ac­tion — ap­proved by the boards of Nvidia, SoftBank and Arm — Nvidia will pay SoftBank $21.5 bil­lion in stock and $12 bil­lion in cash, which in­cludes $2 bil­lion payable at sign­ing. SoftBank may also re­ceive up to $5 bil­lion in cash or com­mon stock if Arm meets cer­tain fi­nan­cial tar­gets. Nvidia will is­sue $1.5 bil­lion in eq­uity to Arm em­ploy­ees.

For SoftBank, the deal rep­re­sents a re­spectable exit from a $32 bil­lion ac­qui­si­tion that had not pro­duced the ben­e­fits ex­pected by its chief, Masayoshi Son. Arm has pushed into many kinds of in­ter­net­con­nected de­vices, as ex­pected, but prof­its have been squeezed by spend­ing on hir­ing and other fac­tors.

“SoftBank is ex­cited to in­vest in Arm’s longterm suc­cess as a ma­jor share­holder in Nvidia,” Son said in a state­ment.

For Nvidia, the trans­ac­tion el­e­vates a com­pany that for decades la­bored in the shad­ows of giants like In­tel in set­ting key tech­ni­cal di­rec­tions for Sil­i­con Val­ley.

Huang was early to rec­og­nize that chores such as com­puter graph­ics weren’t han­dled well by the gen­eral­pur­pose pro­ces­sors of the kind pop­u­lar­ized by Santa Clara’s In­tel. Nvidia built a busi­ness on adding spe­cial­ized ac­cel­er­a­tor chips, mainly through add­in cir­cuit boards plugged into PCs.

Huang later bet shrewdly on mod­i­fy­ing its chips and de­vel­op­ing soft­ware to en­able sci­en­tific and later ar­ti­fi­cial in­tel­li­gence ap­pli­ca­tions. In an­other ag­gres­sive move, Huang opted to pay $7 bil­lion for Mel­lanox, an Is­raeli maker of net­work­ing chips, in a deal that closed in April.

He told an­a­lysts in Au­gust that the Mel­lanox deal was es­sen­tial be­cause cloud ser­vices are in­creas­ingly not be­ing run on a sin­gle server. In­stead, por­tions of ap­pli­ca­tions are be­ing dis­trib­uted among var­i­ous chips and sys­tems in a data cen­ter, re­quir­ing more and more com­mu­ni­ca­tions be­tween the ma­chines, he said.

So Nvidia has a strong in­ter­est in con­trol­ling net­work­ing tech­nol­ogy and other key com­po­nents in the data cen­ter — and that strat­egy, an­a­lysts and in­dus­try ex­ec­u­tives said, ex­tends to con­trol­ling gen­er­alpur­pose com­put­ing tech­nol­ogy like Arm’s, which would al­low Nvidia to chal­lenge In­tel tech­nol­ogy at the heart of most servers.

“The com­put­ing unit is an en­tire data cen­ter now,” Huang said on the con­fer­ence call last month. “We rec­og­nized that, as a com­put­ing com­pany, we have to be a data cen­ter­scale com­pany.”

Huang on Sun­day said the deal would cre­ate the “world’s premier com­put­ing com­pany for the age of AI,” push­ing be­yond the data cen­ter to bring that tech­nol­ogy to dis­trib­uted de­vices in a model known as “edge com­put­ing.” He said Nvidia would also be­gin li­cens­ing some of its own tech­nol­ogy to other com­pa­nies for the first time, us­ing Arm’s ser­vices.

Nvidia said it ex­pected the trans­ac­tion to close in about 18 months, as­sum­ing reg­u­la­tory ap­proval in the United States, Bri­tain, the Euro­pean Union and China.

Newspapers in English

Newspapers from USA

© PressReader. All rights reserved.