Artificial Intelligence

China’s SMIC expects chip supply shortages to persist through 2027

BUZZINGA team4 min read

Chinese international semiconductor manufacturing company SMIC has projected that constraints and pressures in the AI chip supply chain will continue through 2027, with the memory chip market reaching a balance between mid- and late 2028. China's largest contract chipmaker also pointed to equipment lead times exceeding 24 months and a roughly 30% increase in depreciation expenses in 2026.

Estimates from China’s SMIC paint a picture of a constrained supply chain in AI computing through 2027, governed by memory and advanced packaging bottlenecks and manufacturing equipment lead times exceeding two years.

Zhao Haijun’s forecasts on continuing supply pressures and investment bubble concerns

Semiconductor Manufacturing International Corporation, known as SMIC and China’s largest contract chipmaker, stated that the supply chain tied to artificial intelligence applications will likely remain restricted and pressured through 2027, according to Reuters on November 13, 2026, at 02:01 UTC, corresponding to 06:01 AM Gulf Standard Time.

Co-CEO Zhao Haijun’s remarks came as investors question whether massive investments flowing into AI computing constitute a financial speculative bubble. Zhao described the market as still being in an early phase where demand outstrips available resources for many clients, noting that these statements should be treated as management’s forward-looking outlook rather than confirmed future orders.

Interconnected semiconductor markets across processors, memory, and advanced packaging

Semiconductor supply chains involve several distinct sub-markets; a chip manufacturer may possess the production capacity to make a specific category of processors, while clients still face severe shortages in memory chips, advanced packaging services, manufacturing equipment, or related supporting components.

Integrated AI computing systems require specialized accelerators and traditional processors alongside high-speed memory, storage solutions, and networking, meaning that boosting output in a single category does not automatically resolve all bottlenecks across the rest of the ecosystem.

Memory market balance in 2028 and manufacturing equipment delays exceeding 24 months

Zhao Haijun projected that supply and demand in the memory chip market will reach a better balance between mid- and late 2028, an estimate based on current industry conditions that could shift depending on new factories entering production, shifting demand levels, or technical developments.

SMIC also noted that manufacturing equipment delivery wait times exceed 24 months for select orders, complicating capital expenditure plans. Chip fabrication plants require highly complex machinery, reliable facilities, and skilled personnel, and long lead times limit the speed at which usable actual production can be scaled up.

Surging depreciation expenses of 30%, revenue outlook, and export restrictions

On the accounting front, the company projected depreciation expenses to rise by about 30% during 2026. Depreciation reflects the accounting allocation of the cost of facilities and equipment over their estimated useful lives, meaning it can climb as new production capacities enter service even if those assets support future output. Reuters also reported full-year revenue expectations exceeding $1.2 billion, a figure that should not be used in isolation from full financial statements to deduce the total scale of China’s AI chip industry.

SMIC’s assessments also fall within the context of international export restrictions and domestic efforts to develop local semiconductor capabilities, where acquiring certain advanced manufacturing technologies remains highly challenging, and economic impacts depend on the specific products, tools, and processes involved rather than a single statement.

Separating hardware demand strength from data center profitability and tracking metrics

In evaluating bubble claims, a precise commercial distinction must be made between growing hardware demand and individual project profitability. AI demand may continue to grow while specific projects remain unprofitable, and robust equipment orders do not prove that every data center purchasing them generates sufficient financial returns.

Key upcoming tracking metrics center on manufacturer capacity utilization rates, memory chip prices, equipment delivery lead times, actual chip shipments, and verified customer demand levels from independent sources.

FAQs

Question: Until when does SMIC expect AI chip supply chain shortages to continue?
Answer: SMIC expects the semiconductor supply chain linked to AI to remain restricted through 2027, while memory chip supply and demand may balance between mid- and late 2028.

Question: How long are delivery wait times for certain manufacturing equipment noted by SMIC?
Answer: The company explained that delivery lead times for selected manufacturing equipment orders exceed 24 months, complicating capital expenditure planning and slowing rapid production expansion.

Question: What does SMIC’s projected 30% increase in depreciation expenses in 2026 mean?
Answer: Depreciation represents the accounting distribution of factory and equipment costs over their useful lives, rising naturally when new manufacturing lines and facilities are brought into operational service.

Question: Do strong orders for chip-making equipment prove there are no loss risks for data centers?
Answer: No, strong demand for equipment does not automatically prove that every data center purchasing those chips will generate enough operating revenue to cover expenses and achieve profitability.