I’m tru­ly grate­ful, as Smoltek CEO to be invit­ed to par­tic­i­pate in these dis­cus­sions that reflect the grow­ing rel­e­vance of ultra-thin capac­i­tor tech­nolo­gies in the future AI infra­struc­ture stack. I had the great hon­or of meet­ing the Yageo man­age­ment team dis­cussing the future of ultra-thin capac­i­tor tech­nol­o­gy and not least meet­ing Mr. Pierre Chen, the tru­ly inspir­ing founder and chair­man, The vision and road map for accel­er­at­ing AI as a lead­ing play­er is already being exe­cut­ed with speed by Yageo.
Mag­nus Ander­s­son, CEO at Smotek

The AI revolution is entering a new phase

What began as a race to build larg­er AI mod­els is rapid­ly becom­ing a race to build the infra­struc­ture capa­ble of pow­er­ing them. At the 2026 AI Sum­mit host­ed by YAGEO Group in Taipei, indus­try lead­ers gath­ered around a com­mon real­iza­tion: scal­ing AI is no longer only a soft­ware chal­lenge — it is fun­da­men­tal­ly an infra­struc­ture challenge. 

This shift is cre­at­ing major oppor­tu­ni­ties across the hard­ware ecosys­tem, espe­cial­ly for com­pa­nies devel­op­ing next-gen­er­a­tion solu­tions in pow­er deliv­ery, advanced mate­ri­als, and com­po­nent miniaturization. 

For Smoltek, par­tic­i­pa­tion in these dis­cus­sions reflects the grow­ing rel­e­vance of ultra-thin capac­i­tor tech­nolo­gies in the future AI infra­struc­ture stack.

“Yageo has shown an amaz­ing per­for­mance and jour­ney towards high-tech AI solu­tions dis­played at the Yageo AI Sum­mit in Taipei. The tran­si­tion to the next gen­er­a­tion AI infra­struc­ture requires new lev­els of per­for­mance and pow­er dis­tri­b­u­tion built with advanced new mate­ri­als.“
Mag­nus Andersson

Hypergrowth in AI drives infrastructure spending

The scale of invest­ment now being direct­ed toward AI infra­struc­ture is unprece­dent­ed and accord­ing to Dell’Oro Group, glob­al data cen­ter cap­i­tal expen­di­tures are pro­ject­ed to reach approx­i­mate­ly $1.7 tril­lion annu­al­ly by 2030, dri­ven pri­mar­i­ly by AI work­loads and hyper­scale infra­struc­ture expan­sion. McK­in­sey esti­mates the broad­er AI-dri­ven data cen­ter build­out could require as much as $6.7–7 tril­lion in cumu­la­tive glob­al invest­ment by 2030. (McK­in­sey & Com­pa­ny).

Invest­ments are dri­ven by capac­i­ty demand in AI-relat­ed data cen­ters that are pro­ject­ed to increase 3.5x by 2030 (McK­in­sey & Com­pa­ny) and the four largest hyper­scalers alone are expect­ed to spend near­ly $600 bil­lion in 2026 on AI infra­struc­ture. (PR Newswire). The num­bers illus­trate a struc­tur­al shift where AI is becom­ing one of the largest infra­struc­ture invest­ment cycles in mod­ern tech­nol­o­gy history.

The bottleneck is moving towards power delivery

As AI proces­sors become more pow­er­ful, ener­gy con­sump­tion and pow­er integri­ty are emerg­ing as key con­straints. Mod­ern AI accel­er­a­tors already con­sume hun­dreds of watts per chip, and future gen­er­a­tions are expect­ed to push sig­nif­i­cant­ly high­er. Research on AI super­com­put­ers shows that pow­er require­ments for lead­ing sys­tems have his­tor­i­cal­ly dou­bled approx­i­mate­ly every year. A small proces­sor (xPU) requires 1 kW in pow­er, more than mod­ern vac­u­um clean­er (that requires about 0.6–0.8 kW). 

This cre­ates increas­ing pres­sure on:

• Pow­er dis­tri­b­u­tion efficiency
• Volt­age stability
• Com­po­nent density
• Ther­mal performance
• Advanced pack­ag­ing integration

In prac­ti­cal terms, the indus­try is approach­ing phys­i­cal lim­i­ta­tions in how effi­cient­ly pow­er can be deliv­ered to next-gen­er­a­tion proces­sors. This is where advanced capac­i­tor tech­nolo­gies become strate­gi­cal­ly important. 

Smoltek’s CNF-MIM capac­i­tor tech­nol­o­gy is being devel­oped specif­i­cal­ly to address these emerg­ing require­ments through: Ultra-thin form fac­tors, high capac­i­tance den­si­ty poten­tial, improved inte­gra­tion close to proces­sors and sup­port for advanced semi­con­duc­tor pack­ag­ing archi­tec­tures. As AI sys­tems con­tin­ue to scale, enabling tech­nolo­gies at the com­po­nent lev­el may become increas­ing­ly crit­i­cal to over­all sys­tem performance.

AI is expanding beyond the data center

Anoth­er impor­tant theme at the sum­mit was the rapid growth of Edge AI that dri­ves demand for data pro­cess­ing capac­i­ty and advanced com­po­nents and pow­er sup­ply in appli­ca­tions such as AI PCs, Smart­phones, Auto­mo­tive, Robot­ics, Indus­tri­al automa­tion and IoT systems. 

This expan­sion dra­mat­i­cal­ly increas­es the num­ber of devices requir­ing advanced pow­er archi­tec­tures and minia­tur­ized elec­tron­ic components. 

Indus­try fore­casts esti­mate bil­lions of AI-enabled edge devices will enter the mar­ket over the com­ing years, cre­at­ing sub­stan­tial demand for more com­pact, ener­gy-effi­cient elec­tron­ic solu­tions.
The oppor­tu­ni­ty for Smoltek CNF-MIM capac­i­tor tech­nol­o­gy there­fore extends beyond hyper­scale infra­struc­ture and into a much broad­er glob­al elec­tron­ics market.

Power consumption Is becoming a strategic industry issue

The growth of AI infra­struc­ture is also cre­at­ing entire­ly new ener­gy challenges. 

The indus­try real­i­ty is that future AI com­pet­i­tive­ness will not only depend on com­pute per­for­mance, but also on how effi­cient­ly pow­er can be dis­trib­uted, man­aged, and uti­lized through­out the infra­struc­ture stack. This trend strength­ens the long-term mar­ket rel­e­vance of tech­nolo­gies capa­ble of improv­ing pow­er effi­cien­cy and com­po­nent integration.

Strategic ecosystem participation matters

The AI infra­struc­ture mar­ket is evolv­ing through increas­ing­ly inter­con­nect­ed ecosys­tems involving:

• Semi­con­duc­tor manufacturers
• Pas­sive com­po­nent suppliers
• Advanced pack­ag­ing companies
• Hyper­scalers
• Sys­tem integrators

Par­tic­i­pa­tion in strate­gic indus­try forums there­fore mat­ters. Smoltek’s CEO Mag­nus Ander­s­son attend­ed the sum­mit and met with senior YAGEO lead­er­ship to dis­cuss future AI infra­struc­ture require­ments and next-gen­er­a­tion capac­i­tor tech­nolo­gies.
For Smoltek, these dis­cus­sions rep­re­sent an oppor­tu­ni­ty to posi­tion its tech­nol­o­gy with­in one of the fastest grow­ing and most strate­gi­cal­ly impor­tant indus­tri­al tran­si­tions globally.

A structural opportunity, not a short-term trend

The AI tran­si­tion is still in its ear­ly stages. 

What is increas­ing­ly clear is that long-term val­ue cre­ation will extend far beyond soft­ware and mod­els. The phys­i­cal infra­struc­ture lay­er — includ­ing pow­er deliv­ery, advanced mate­ri­als, and minia­tur­ized com­po­nents — is becom­ing equal­ly impor­tant. As glob­al AI infra­struc­ture invest­ments accel­er­ate toward mul­ti-tril­lion-dol­lar lev­els, demand for enabling tech­nolo­gies that improve effi­cien­cy, den­si­ty, and sys­tem per­for­mance is expect­ed to grow along­side it.
For Smoltek, this rep­re­sents an oppor­tu­ni­ty to par­tic­i­pate in a long-dura­tion struc­tur­al growth mar­ket dri­ven by the glob­al expan­sion of AI infrastructure.

---

Want to learn more about the extreme­ly small and ultra-thin CNF-MIM capac­i­tor? Go here!