AI summary by TwinMind
China's 7nm Chips: SMIC N+2
Asianometry
13m 39s
SMIC's N+2 Node Technical Overview
- Chinese semiconductor national champion SMIC started producing chips with 7nm process node (N+2)
- N+2 node descendant of N+1 process first taped out October 2020
- Uses DUV immersion lithography with multi-patterning (pre-EUV technology)
- N+2 capable of using EUV but US export ban blocks ASML equipment access
- Semi-analysis via Tech Insights: N+2 99% similar to TSMC's N7 node
- Physical measurements of transistors (FinFETs) and placements nearly identical to TSMC N7
Process Node Context and Meaning
- "Seven nanometers" is marketing term, not physical dimension
- Indicates 70% improvement over previous generation (10nm → 7nm → 14nm progression)
- TSMC calls their version "N7" to avoid controversy
- Process node numbers are marketing devices for power/performance/area improvements
Competitive Positioning and Timeline
- TSMC first shipped N7 chips high volume in 2018
- N7 equivalent node puts SMIC 4-6 years behind leading edge
- N+2 one of most advanced processes available by any foundry to customers
- Without EUV, questionable if economically competitive in open market
Manufacturing Challenges and Capabilities
- 193nm immersion + multi-patterning can fab even N5 equivalent chips (per Dr. Burn Lin, TSMC R&D director, January 2022)
- Key issue: yield in meaningful quantities, not just making few wafers
- Samsung struggles with advanced node yields despite having nodes
- Intel famously struggled 10nm → 7nm equivalent transition without EUV
- TSMC ramped N7 without reported trouble; N+2 expected to follow similar yield optimization curve
Market Dynamics and Limitations
- Market changed since 2018 - competition rougher
- Foundry with EUV has economic advantages over DUV-only
- N7 equivalent with only multi-patterning has manufacturing limitations
- Places serious design restrictions on chip designers
- TSMC's N6 (N7 evolution with EUV) offers better performance
Trade Secret and Legal Concerns
- N+2 close similarity to TSMC N7 likely intrudes on trade secrets
- 2002 and 2006: TSMC successfully sued SMIC in US courts for product import bans
- TSMC could attempt again, potentially removing N+2 export viability
- Would limit N+2 to domestic Chinese products and friendly country exports
- SMIC cares about export capability (highlighted in latest earnings call) but likely no longer depends on it
Chinese Market Opportunities
- Chinese chip design industry much more sophisticated since mid-2000s TSMC lawsuit
- Chinese electronics industry greatly matured
- HiSilicon (Huawei's chip design arm) prime candidate for N+2 usage
- Once China's biggest semiconductor company
- Could use N+2 for high-performing mobile processors and server chips
- Previously TSMC flagship customer alongside Apple
- Fabbed Kirin 980, 985, 990 series chips on N7
- HiSilicon + SMIC partnership could bring serious volume for foundry scaling
- High production volumes essential for yield optimization
Leadership and Strategic Direction
- SMIC co-CEO Liang Mong Song (TSMC alumni) joined 2017 from Samsung
- Known for technical brilliance, difficult personality, dogged chase for leading edge
- Brought 200-person R&D team of Taiwanese and Koreans
- Reoriented SMIC from boring second-source manufacturer to leading edge monster
- Company should continue pushing toward leading edge while Liang remains co-CEO
Export Control and Geopolitical Implications
- US government pushing Netherlands (ASML) and Japan (Nikon) to prevent DUV immersion lithography exports to China
- Timing of push and N+2 shipment likely not coincidental
- US long blocked newest EUV machines to China
- Expanding block to immersion lithography represents another step forward
- Immersion machines use water/fluids to improve resolution of 193nm ARF laser technology
- Only two suppliers: ASML and Nikon
- Dry ARF machines can be modified to immersion (less than 5 years but substantial technical effort)
- Western industry transitioned dry → immersion in 2-3 years
- N7 equivalent requires more sophisticated immersion technologies
Technology Development Timeline and Impact
- Japan and Netherlands haven't decided on ban as of writing
- Chinese government responded with claims of unfairness and foul play
- Export ban of 193nm immersion (without dry ARF ban) would add 4-6 years to SMIC's leading edge march
- Ban likely to stimulate substantial investment in Chinese-made lithography
- First: equipment to maintain existing immersion equipment
- Second: facilitate dry-to-immersion transition
- Gap with ASML/Nikon greater but not insurmountable (several years)
Strategic Assessment and Historical Context
- Legitimate 7nm node at high volume/good yield is triumph for China's advanced manufacturing
- Only three other companies achieved this level of scale and sophistication
- Any successful immersion ban wouldn't be permanently damaging blow
- Can only maintain and perhaps slightly expand Sino-American semiconductor gap
- Administration values those few years highly
- Reference to American response to Japanese semiconductor challenge 30+ years ago
- Government intervention can only buy time - industry must build new markets, improve manufacturing, adopt innovative suppliers
