On July 30, 2026, the global semiconductor industry ushered in a new wave of growth opportunities. According to the latest "Global Chip Market Quarterly Report," global semiconductor sales in the second quarter of 2026 reached $158 billion, up 14.8% year-over-year and 3.2% quarter-over-quarter. Among them, AI inference chips, automotive power semiconductors, and memory chips became the main growth engines. This series of data once again confirms a core view: chip investment is at the right time.
AI Inference Chips: A Blue Ocean Market with Explosive Demand
As generative AI technology moves from the training phase to large-scale inference deployment, the demand for AI inference chips is growing exponentially. Tech giants such as OpenAI, Google, and Microsoft continue to expand their cloud inference infrastructure, and Meta's Llama series of open-source models have driven a wave of enterprise-level private deployment. According to IDC estimates, the global AI inference chip market will exceed $35 billion in 2026, with a compound annual growth rate as high as 45%.
On the supply side, NVIDIA continues to dominate the high-end market, but its H200 and B200 series chips are in short supply, with lead times exceeding 6 months. Meanwhile, AMD's MI400 series and Intel's Gaudi 3 are also gaining share in inference scenarios. More notably, a group of startups focused on inference optimization, such as Tenstorrent, Groq, and China's Cambricon, are entering niche markets with unique architectural designs. Breakthroughs in matrix multiplication and sparse computing have improved the cost-effectiveness of inference chips by 3-5 times compared to training chips, clearing the way for the large-scale adoption of downstream AI applications.
From an investment logic perspective, the explosion of AI inference chips not only benefits GPU design companies but also drives demand for advanced packaging, HBM memory, and high-speed interconnect chips. TSMC's CoWoS capacity has climbed to 45,000 wafers per month but still cannot meet orders; SK Hynix and Samsung's HBM4 memory began mass production in 2026, with prices up 20% over the previous generation. The boom in the entire industry chain is being fully ignited by AI inference demand.
Automotive Semiconductors: Structural Opportunities Under Supply-Demand Rebalancing
The chip shortage problem for automotive applications over the past two years has largely eased, but a new supply-demand pattern is taking shape. According to Omdia data, the global automotive semiconductor market is expected to reach $72 billion in 2026, up 11% year-over-year. Among them, power semiconductors (IGBT, SiC, GaN) remain in short supply due to the increasing penetration of electric vehicles; automotive MCUs have experienced temporary oversupply due to inventory adjustments, but demand for high-end domain controller chips is still on the rise.
Regarding silicon carbide (SiC), leading companies such as STMicroelectronics, Infineon, and Wolfspeed have successively started production at their 8-inch fabs, which will release substantial capacity in 2027. However, in the short term, the improvement in SiC substrate yield is slow, keeping chip prices high—around $1,500 per wafer, 4-5 times that of traditional silicon chips. This allows IDM manufacturers with vertical integration capabilities to enjoy higher profits. Meanwhile, gallium nitride (GaN) is accelerating penetration in on-board chargers and 48V mild hybrid systems, and the GaN power chip market is expected to exceed $1.5 billion in 2026.
Investors should pay attention to the structural changes in automotive semiconductors: on one hand, the migration toward third-generation semiconductors with higher voltage and higher efficiency; on the other hand, the demand for sensors and computing chips driven by the improvement of autonomous driving levels. The per-vehicle value of LiDAR, millimeter-wave radar, and cockpit SoCs is rising, from $350 in traditional ICE vehicles to over $800 in EVs. This trend of rising volume and price provides long-term growth momentum for chip companies.
Domestic Substitution: China's Chip Industry Accelerates Breakthrough
Against the backdrop of the U.S. continuously expanding semiconductor export controls on China, the domestic substitution of China's chip industry has moved from the startup phase into an acceleration phase. In the first half of 2026, China's chip imports amounted to $195 billion, down 8% year-over-year, while domestic chip output grew by 22%, indicating a significant substitution effect. Especially in the mature process field, SMIC and Hua Hong Semiconductor maintained capacity utilization above 95% and continued to expand new production lines.
More notably, domestic chips have achieved key breakthroughs in AI inference, automotive MCUs, and power semiconductors. Huawei's Ascend 910B has approached 80% of the inference performance of NVIDIA's A100 and has secured orders from several major domestic cloud operators. SiC products from Starsemi and Times Electric have started entering the supply chains of automakers such as BYD and Geely. In addition, the domestic EDA tool chain is also maturing, with software from Empyrean and Guowei Group now supporting 7nm design flows.
On the policy front, the third phase of the China National Integrated Circuit Industry Investment Fund completed fundraising in 2025, with a scale of 300 billion RMB, focusing on advanced packaging, equipment, and materials. Recent news indicates that the fourth phase will target AI chips and third-generation semiconductors. These policy dividends, combined with downstream market demand, provide investors with abundant target options.
Inventory Adjustment Near End, Equipment Shipments Hit New Highs
The semiconductor inventory adjustment cycle has lasted about one and a half years, and the upstream inventory level has dropped from its peak to a healthy level. According to SIA, global semiconductor inventory turnover days in Q2 2026 were 68 days, close to the historical average of 65 days. Particularly in the memory chip sector, inventories of DRAM and NAND Flash have fallen to below 4 weeks, driving prices up for two consecutive months since May, by 8% and 12% respectively.
Alongside inventory reduction, semiconductor equipment shipments continue to hit new highs. According to data from the Semiconductor Equipment Association of Japan (SEAJ), Japan's semiconductor equipment shipments reached 385 billion yen in June 2026, up 28% year-over-year. North American semiconductor equipment shipments also grew, driven mainly by AI chips and advanced packaging expansion. ASML's High-NA EUV lithography machine orders are booked through 2028, and equipment makers such as Applied Materials and Tokyo Electron have order visibility of more than 12 months.
From an investment perspective, equipment stocks usually start earlier than chip stocks and have strong cyclical sustainability. With global fab capital expenditure expected to reach $120 billion in 2026 (up 12% YoY), equipment manufacturers will directly benefit. On the materials side, such as silicon wafers, CMP slurries, and specialty gases, they will also grow simultaneously. Shin-Etsu Chemical and SUMCO have announced a 15% price increase for 300mm silicon wafers in 2027, reflecting tightening supply and demand.
Investment Strategy Outlook: Position on Three Major Themes
Based on the above analysis, from the second half of 2026 to 2027, chip investment can be focused on three major themes:
- AI inference ecosystem: Focus on GPU design companies, HBM memory, advanced packaging, and high-speed interconnect chips. Key targets include NVIDIA, AMD, SK Hynix, and TSMC.
- Automotive semiconductor upgrade: Focus on SiC/GaN power devices, smart driving SoCs, and in-vehicle sensors. Infineon, STMicroelectronics, Wolfspeed, and domestic Starsemi are worth noting.
- Domestic substitution breakthrough: Lock on mature process foundry, domestic AI chips, EDA tools, and equipment materials. SMIC, Huawei Ascend, and Empyrean are core observation targets.
In addition, the trend of semiconductor supply chain relocation to Southeast Asia is also worth watching. Thailand and Malaysia are becoming new clusters for packaging, testing, and passive components, with many international manufacturers expanding production locally, creating new investment opportunities. Investors can diversify risk through ETFs or participation in related industry chain funds, while closely monitoring policy trends and inventory data from various countries to dynamically adjust positions.
In summary, the chip industry is at the intersection of the AI revolution and the electrification transformation, with high long-term growth certainty. Although there are short-term risks of geopolitical volatility and overcapacity concerns, structural opportunities far outweigh the risks. For investors seeking high growth, the current adjustment is precisely a good opportunity to position.
