ChinaBiz Briefing | Chips, Batteries, and AI: China's Tech Giants Push Global Expansion Amid Intensifying Competition
China's technology and manufacturing sectors are navigating a pivotal moment as leading companies demonstrate resilience through innovation and global expansion despite structural headwinds. From Shanghai's chip champion achieving a blockbuster Hong Kong debut to battery giant CATL proving overseas profitability, and AI developers narrowing the gap with US rivals, this week's developments reveal an ecosystem leveraging efficiency and scale to compete on the world stage—even as questions around monetization and margin sustainability intensify.
Shanghai Chip Giant Montage Surges 57% in Hong Kong Debut, Backed by Alibaba
Montage Technology, the world's largest memory interconnect chip supplier with 36.8% global market share, opened trading in Hong Kong at HK168 per share—57% above its HK106.89 IPO price—valuing the company at HK203.67 billion (US$26.1 billion). The offering attracted cornerstone investors including Alibaba's Alisoft China subsidiary and counted Intel and Samsung among existing shareholders.
Why it matters: Montage's successful dual listing (previously on Nasdaq, now on STAR Market and Hong Kong) signals sustained investor appetite for AI infrastructure plays, particularly companies positioned at critical chokepoints in the semiconductor supply chain. The company's 61.5% gross margin and 90%+ revenue concentration in DDR5 memory interconnect chips—essential for data transmission between CPUs and DRAM in AI servers—places it at the center of accelerating cloud and AI infrastructure buildouts. With DDR5 penetration reaching 45-50% of the DRAM market in 2025 and the company's Q1-Q3 2025 revenue already exceeding full-year 2024 figures, Montage exemplifies how Chinese chip firms are capitalizing on AI-driven demand despite US export controls targeting leading-edge logic chips. Its recent minority investments in GPU developer Biren Technology and a flash memory chip startup underscore ambitions to expand beyond memory interfaces into broader AI infrastructure components.
CATL Maintains Cost Leadership Amid Global Expansion, UBS Teardown Reveals
UBS upgraded CATL to a RMB 500 price target following a comprehensive battery teardown analysis revealing the company's Shenxing 2.0 LFP cell achieves US$55/kWh manufacturing cost — a new global benchmark. Critically, CATL's Germany-produced high-nickel NMC cells cost just US$87/kWh, only US$10-15 higher than domestic production and competitive with European rivals Northvolt (US$88/kWh) and LG (US$86/kWh).
Why it matters: The analysis directly challenges investor concerns that overseas expansion will erode CATL's industry-leading margins, demonstrating the company can replicate its cost advantage internationally through automation rather than labor arbitrage. With 100GWh Hungary capacity starting in 2026, 50GWh in Spain via Stellantis JV, and 15GWh in Indonesia all planned for 2026-2027 production, CATL is executing a localization strategy essential for capturing European and Southeast Asian markets amid rising trade barriers. The company's 38.2% global market share—more than double runner-up BYD—combined with its eighth-generation manufacturing platform that reduces headcount requirements by 70% while cutting costs 33%, positions it to maintain dominance even as lithium carbonate prices doubled to RMB 160,000/ton. UBS's 25% earnings CAGR forecast through 2029 reflects confidence that CATL's scale and efficiency moat will withstand both raw material volatility and the April 2026 reduction in China's battery export VAT rebates from 9% to 6%.
Barclays: China's AI Revolution Narrowing Gap With US Despite Structural Headwinds
Barclays' comprehensive analysis concludes Chinese AI has made "astounding" progress over 12-18 months through aggressive open-source strategies, with DeepSeek V3.2 pricing output tokens at just 3% of GPT-5.2 rates (US0.42vsUS0.42vsUS14 per million tokens) while achieving performance comparable to Claude Sonnet 4.5 on key benchmarks. However, monetization challenges persist—DeepSeek's 249 million global MAUs include only 52 million (21%) outside China, versus ChatGPT's 800 million weekly actives with 90% international distribution.
Why it matters: The report illuminates the fundamental trade-off shaping the US-China AI competition: American companies deploy massive capital for proprietary frontier models targeting high-value enterprise and productivity use cases, while Chinese firms leverage open-source collaboration and extreme cost efficiency under compute constraints to dominate mobile-first, consumer-oriented applications. Barclays estimates aggregate Chinese AI capex at less than one-fifth of top US hyperscalers' spending, forcing efficiency innovations like DeepSeek's approach. ByteDance's Doubao reaching 199 million MAUs and Baidu AI search capturing 347 million MAUs (over half its search traffic) demonstrate scale, but overwhelmingly domestic, often-free usage generates high token volumes with lower yield. The divergent paths suggest Chinese AI will remain competitive on core capabilities—particularly as open-source models continue rapid iteration—while US platforms maintain structural monetization advantages through workflow-heavy, reasoning-intensive applications and global willingness to pay for premium features.
Li Auto Bets on Volume Over Margins With AI-Branded "L9" Flagship
Deutsche Bank maintains its Buy rating on Li Auto while slashing the price target to HKD 100, following analysis of the company's Q2 2026 "Li L9" flagship SUV launch. The vehicle—positioned as an "Embodied AI robot" with claimed Level 4 autonomy—will start at RMB 320,000 (US$44,000), matching the current L8 SUV price despite featuring dual proprietary "M100" 5nm autonomous driving chips delivering 2,560 TOPS, 800V fully active suspension, and complete control-by-wire architecture.
Why it matters: Li Auto's aggressive pricing strategy prioritizes fleet scale and data collection over near-term profitability, reflecting the brutal calculus facing Chinese premium EV makers: they need massive real-world driving data to train Vision-Language-Action models for autonomous driving, but face intense competition from NIO, XPeng, and legacy manufacturers pursuing identical strategies. The "L9 Livis" premium variant's hardware—including 360-degree LiDAR with four to five units and full X-by-Wire architecture previously exclusive to ultra-luxury models like NIO's RMB 768,000 ET9—democratizes technologies essential for autonomous operation. However, Deutsche Bank's substantially reduced price target acknowledges the margin compression inherent in this volume play. The launch represents a critical test of whether Chinese EV makers can successfully transition from selling vehicles to deploying AI-powered mobility platforms before commoditizing themselves in the race toward an autonomous future that remains technologically and economically uncertain.
What to watch: CATL's Hungary plant ramp and European pricing dynamics through 2026 will test the viability of Chinese battery makers' overseas expansion model. Li Auto's Q2 "L9" launch and initial sales trajectory will signal whether premium Chinese EV brands can sustain margins while pursuing autonomous driving scale. And the monetization strategies emerging from ByteDance, Baidu, and DeepSeek over the coming quarters will determine whether China's AI efficiency advantages can translate into sustainable business models beyond domestic mobile applications.