Chips Act: can Europe catch up in the global semiconductor race?

Since short­ages caused by the Cov­id-19 health crisis, the high con­cen­tra­tion of pro­duc­tion in Taiwan and the Sino-Amer­ic­an tech­no­logy war over arti­fi­cial intel­li­gence, semi­con­duct­ors have occu­pied a spe­cial place in indus­tri­al policy. Found in smart­phones, data centres, vehicles, med­ic­al equip­ment and defence sys­tems, they account for a grow­ing share of tech­no­lo­gic­al invest­ment and sov­er­eignty strategies¹.

In this land­scape dom­in­ated by a hand­ful of play­ers cap­able of pro­du­cing the most advanced chips, Europe finds itself in a very spe­cif­ic pos­i­tion. The con­tin­ent has lead­ing com­pan­ies in cer­tain seg­ments of the value chain, start­ing with ASML in EUV litho­graphy, but remains absent from the man­u­fac­ture of the most advanced nodes, a sec­tor cur­rently dom­in­ated by TSMC and Sam­sung². This situ­ation has led the European Uni­on to adopt the Chips Act, which aims to strengthen the continent’s indus­tri­al cap­ab­il­it­ies and reduce cer­tain stra­tegic depend­en­cies³. How­ever, the chal­lenge is not lim­ited to build­ing new factor­ies or mas­ter­ing the most mini­atur­ised man­u­fac­tur­ing processes.

Some of the work cur­rently being car­ried out in industry and research focuses on oth­er tech­no­lo­gic­al path­ways, not­ably around chip­lets, het­ero­gen­eous integ­ra­tion and advanced com­pon­ent assembly tech­niques⁴. These approaches make it pos­sible to com­bine, with­in a single sys­tem, tech­no­lo­gic­al build­ing blocks developed using dif­fer­ent pro­cesses, rather than simply con­tinu­ing the race towards miniaturisation.

As Sci­entif­ic Dir­ect­or of the “Integ­ra­tion and Pack­aging” pro­gramme at CEA-Leti, Pas­cal Viv­et has been work­ing on advanced semi­con­duct­or integ­ra­tion tech­no­lo­gies for numer­ous years⁵. His research focuses in par­tic­u­lar on three-dimen­sion­al archi­tec­tures, chip­lets and assembly pro­cesses that enable dif­fer­ent elec­tron­ic func­tions to be com­bined with­in a single sys­tem. Not­able among these pro­jects is the IntAct pro­ject, which is ded­ic­ated to new multi-core archi­tec­tures based on the three-dimen­sion­al stack­ing of chip­lets⁶.

His­tor­ic­ally, Europe pos­sessed expert­ise in the most advanced tech­no­lo­gies, driv­en in par­tic­u­lar by STMi­cro­elec­tron­ics, Infin­eon and NXP. But around fif­teen years ago, European man­u­fac­tur­ers gradu­ally stopped pur­su­ing this tech­no­lo­gic­al race, leav­ing the field open to Amer­ic­an play­ers (such as Intel) and Asi­an play­ers such as TSMC and Sam­sung, who were the only ones cap­able of mobil­ising the massive invest­ments now required.

The cur­rent gap is more a mat­ter of invest­ment than a tech­no­lo­gic­al short­fall. Europe retains world-class expert­ise, par­tic­u­larly in litho­graphy through ASML. At CEA-Leti, work is con­tinu­ing on advanced FD-SOI pilot lines around the 10 nm mark, with the aim of trans­fer­ring these innov­a­tions to industry. Achiev­ing more advanced tech­no­logy nodes there­fore remains pos­sible. Mov­ing from 28 nm or 18 nm, cur­rently used indus­tri­ally in Europe, to 10 nm or 7 nm would already rep­res­ent a con­sid­er­able tech­no­lo­gic­al leap. The real ques­tion is what resources Europe is pre­pared to com­mit to this ambition.

The main vul­ner­ab­il­ity lies in the eco­sys­tem of design com­pan­ies, the fab­less com­pan­ies [edit­or­’s note: com­pan­ies without their own factor­ies], which devel­op new cir­cuit archi­tec­tures and new sys­tems. It is often with­in these organ­isa­tions that the most dis­rupt­ive innov­a­tions emerge. Europe has large, robust indus­tri­al groups, but it still lacks a dense net­work of start-ups cap­able of nur­tur­ing the sector’s future lead­ers. The United States has long benefited from a par­tic­u­larly dynam­ic eco­sys­tem, sup­por­ted by ven­ture cap­it­al cap­able of rap­idly fuel­ling the growth of innov­at­ive com­pan­ies. Two examples worth not­ing are start-ups design­ing AI accel­er­at­ors using advanced tech­no­lo­gies, such as AXELERA (in the Neth­er­lands) and VSORA (in France).

The Chips Act aims to address these chal­lenges by cre­at­ing shared design plat­forms, provid­ing com­mon tools and open tech­no­logy lib­rar­ies⁷. How­ever, the main dif­fi­culty remains the abil­ity to fin­ance the scal­ing up of these com­pan­ies. European start-ups often have the neces­sary skills and ideas, but struggle to scale up quickly enough to com­pete with their Amer­ic­an or Asi­an rivals. The prob­lem is there­fore not so much tech­no­lo­gic­al as fin­an­cial. Europe is still strug­gling to rap­idly trans­form its innov­a­tions into lead­ing indus­tri­al players.

FD-SOI is a real asset for Europe, even if it does not cov­er all its needs. For high-per­form­ance com­put­ing applic­a­tions, data centres or the most advanced arti­fi­cial intel­li­gence infra­struc­ture, cut­ting-edge CMOS remains indis­pens­able. On the oth­er hand, for fields such as quantum com­put­ing, micro­con­trol­lers, ima­ging, sensors, low-power applic­a­tions or cyber­se­cur­ity, FD-SOI offers sig­ni­fic­ant advant­ages and can be a key dif­fer­en­ti­at­ing factor. Provided the neces­sary invest­ment con­tin­ues, this tech­no­logy can there­fore rep­res­ent a stra­tegic pil­lar across a range of major indus­tri­al applications.

The issue of skills is one of the main chal­lenges. At present, Europe does not yet have the human resources required to rap­idly roll out sev­er­al state-of-the-art mega-factor­ies. This is pre­cisely one of the issues addressed under the Chips Act. In France in par­tic­u­lar, sig­ni­fic­ant efforts are under­way in uni­ver­sit­ies, engin­eer­ing schools, research organ­isa­tions and labor­at­or­ies to increase train­ing capa­city and the num­ber of gradu­ates in fields related to micro­elec­tron­ics. But devel­op­ing the skills required for an industry of this scale takes time. Even with adequate indus­tri­al invest­ment, build­ing up human resources is a long-term process.

It is dif­fi­cult to assess, as indus­tri­al tra­ject­or­ies depend so heav­ily on tech­no­lo­gic­al, human and polit­ic­al factors. Bey­ond the tech­no­lo­gic­al chal­lenges them­selves, we must con­sider the time required to ramp up pro­duc­tion, recruit­ment dif­fi­culties and the com­plex­ity of European decision-mak­ing pro­cesses. The polit­ic­al will is there but trans­lat­ing it into oper­a­tion­al action is often a slower pro­cess. The tar­get is ambi­tious and sends an import­ant sig­nal to European industry. As for wheth­er it will be achieved by 2030, it is still too early to say.

No single tech­no­lo­gic­al bot­tle­neck cur­rently appears to be pre­vent­ing Europe from mov­ing for­ward. Access to ASML’s EUV equip­ment, for example, remains assured as it is a European tech­no­logy. The main obstacles seem to lie more in the areas of invest­ment, skills, indus­tri­al organ­isa­tion, design cap­ab­il­it­ies and the devel­op­ment of the entre­pren­eur­i­al ecosystem.

In the cur­rent geo­pol­it­ic­al con­text, char­ac­ter­ised by mount­ing inter­na­tion­al ten­sions and a resur­gence of con­cerns over sov­er­eignty, Europe nev­er­the­less has a sig­ni­fic­ant oppor­tun­ity. Grow­ing demand in the defence, space and crit­ic­al infra­struc­ture sec­tors is cre­at­ing a favour­able envir­on­ment for the devel­op­ment of new indus­tri­al cap­ab­il­it­ies. Even if Europe lags some­what behind in the most advanced pro­cesses, the invest­ments made today could yield sig­ni­fic­ant res­ults with­in the next 5 to 10 years.

Not entirely. The focus on the most advanced tech­no­logy nodes some­times provides an incom­plete pic­ture of cur­rent indus­tri­al chal­lenges. One of the major devel­op­ments in the sec­tor con­cerns het­ero­gen­eous integ­ra­tion, that is, the abil­ity to com­bine sev­er­al spe­cial­ised tech­no­lo­gies with­in a single sys­tem. The aim is no longer simply to man­u­fac­ture a single, ever-more-mini­atur­ised chip, but to assemble dif­fer­ent chip­lets spe­cific­ally ded­ic­ated to com­put­ing, memory, sensors, power man­age­ment, cyber­se­cur­ity or communications.

3D integ­ra­tion and advanced pack­aging tech­no­lo­gies make it pos­sible to bring these dif­fer­ent tech­no­lo­gic­al build­ing blocks togeth­er. In this field, Europe already has sig­ni­fic­ant strengths. CEA-Leti in France, as well as sev­er­al organ­isa­tions in Ger­many and Bel­gi­um, have been devel­op­ing recog­nised expert­ise in these fields for sev­er­al years. This approach can meet the needs of stra­tegic European sec­tors such as the auto­mot­ive, aerospace and defence indus­tries, quantum com­put­ing, AI, com­mu­nic­a­tions and sensors, whilst lim­it­ing the costs and risks asso­ci­ated with the con­stant race for the most advanced processes.

Europe has strong pos­i­tions in sev­er­al areas that com­ple­ment the semi­con­duct­or sec­tor itself. Embed­ded soft­ware, embed­ded sys­tems, cyber­se­cur­ity and tech­no­lo­gies for pro­tect­ing digit­al infra­struc­ture are sec­tors in which European play­ers pos­sess recog­nised expert­ise. These cap­ab­il­it­ies are essen­tial for secur­ing data centres, cloud infra­struc­ture, pub­lic ser­vices and crit­ic­al indus­tri­al sys­tems. They rep­res­ent import­ant levers for tech­no­lo­gic­al sov­er­eignty and enable Europe to main­tain strong pos­i­tions in high value-added segments.

This ques­tion is cur­rently the sub­ject of much debate with­in the European eco­sys­tem. Quantum com­put­ing is already attract­ing sig­ni­fic­ant invest­ment. Indus­tri­al applic­a­tions are still some way off, but fail­ing to invest today would mean fall­ing behind in a way that could prove irre­vers­ible tomorrow.

Both approaches remain on the table. At the same time, Europe would bene­fit from step­ping up its invest­ment in areas where it already has com­pet­it­ive advant­ages: het­ero­gen­eous integ­ra­tion, FD-SOI, the Inter­net of Things, indus­tri­al robot­ics, Industry 4.0, cyber­se­cur­ity and tech­no­lo­gies for the health­care sec­tor. Micro­elec­tron­ics now lies at the heart of all these eco­sys­tems. Strength­en­ing these sec­tors there­fore amounts to strength­en­ing European tech­no­lo­gic­al sov­er­eignty more broadly.

Interview by Aicha Fall