Arveum Capital PartnersCapital Partners
🔬

Semicon Briefing

September 4, 2026 · 03:48 Uhr

1

ams-OSRAM sells sensor division to Infineon for €570 million

finanztrends.de / ad-hoc-news.de

ams-OSRAM has divested its analog and mixed-signal sensor division to Infineon for €570 million; the deal is already completed and is expected to generate around €230 million in annual revenue for Infineon in 2026. The ams-OSRAM stock has gained 130% since the beginning of the year – the transaction accelerates the realignment of both European semiconductor companies toward core competencies.

CRITICALRead article
2

Nvidia acquires Hugging Face for $12.9 billion

@NasdaqTerminal

According to market reports, Nvidia is set to acquire the AI platform Hugging Face for approximately $12.93 billion – a strategic move that would transform Nvidia from a pure chip supplier to a complete AI stack provider. The acquisition would substantially expand Nvidia's influence on the open-source AI ecosystem and put pressure on competitors like Google and Microsoft.

CRITICALRead article
3

TSMC & Samsung postpone High-NA EUV to 1-nm generation

wccftech.com / ad-hoc-news.de

Both leading contract manufacturers confirm they will deploy ASML's 165-ton High-NA EUV machines (0.55 NA) starting only from the 1-nm generation, expected no earlier than 2030 – only Intel is already using the equipment in production. ASML shareholders thus face a delayed volume ramp, while TSMC and Samsung are relying on multi-patterning workarounds to control costs.

4

Tencent closes $2.94 billion DRAM deal with Chinese CXMT

@SemiconductorsX

Tencent has signed a multi-year server DRAM supply contract worth more than 20 billion yuan (~$2.94 billion) with Chinese memory chip manufacturer CXMT – a signal that Chinese hyperscalers are actively substituting Western memory suppliers like Samsung and SK Hynix. The deal significantly strengthens China's strategy to decouple its AI infrastructure from the Western semiconductor ecosystem.

CRITICALRead article
5

Samsung binds 70% memory capacity in LTAs through 2031

@SemiconductorsX

Samsung's memory division has tied up approximately 70% of its production capacity in long-term agreements (LTAs) through 2031 – primarily to major hyperscalers and AI companies. For new customers, this means significant supply shortages and further price pressure: TSMC and Samsung are both signaling 10–15% price increases across all nodes.

6

US AI data centers: Hidden China dependency in transformers & optics

CNBC

US AI data centers remain heavily dependent on Chinese suppliers for transformers, battery systems, and optical components despite export controls – a systemic vulnerability that Washington now seeks to address with tightened restrictions. The finding undermines the reshoring rhetoric of Lutnick's announcements and shows that decoupling from China runs deeper than previously communicated.

Situation Report

The global semiconductor industry is in a phase of simultaneous consolidation, geopolitical reordering, and technological pivoting: While Nvidia's presumed Hugging Face acquisition elevates it from chip supplier to AI stack monopolist, and Infineon's ams-OSRAM acquisition redraws Europe's sensor landscape, the Tencent-CXMT megadeal structurally sharpens China's decoupling from the Western memory market. TSMC and Samsung's decision to introduce High-NA EUV only at 1 nm slows ASML's volume ramp and temporarily gives Intel, as the sole early High-NA adopter, a technological differentiation advantage. From a security perspective, the CNBC analysis of hidden China dependency in American AI data centers for infrastructure components is concerning – it demonstrates that Washington's chip tariff offensive misses critical supply chain points and that the strategic vulnerability of Western AI infrastructure is greater than publicly acknowledged.

Tokens: 2,415(1,462 in · 953 out)

This website uses cookies. Strictly necessary cookies are always active. By clicking "Accept all" you additionally consent to analytics cookies (Google Analytics). Privacy Policy →