AI-Chip Capacity Shortage Gives Global Semiconductor Foundries Fresh Pricing Power

AI-Chip Capacity Shortage Gives Global Semiconductor Foundries Fresh Pricing Power

Global semiconductor foundries are gaining fresh pricing power as surging demand for artificial-intelligence processors consumes advanced manufacturing capacity faster than new fabrication plants can come online. Samsung Electronics has raised prices for selected foundry services by as much as 15%, while China’s SMIC has also increased wafer pricing as utilisation approaches full capacity. TSMC, the dominant producer of advanced AI chips, continues to report exceptionally strong demand and capacity constraints. The emerging shortage represents a major change in semiconductor economics, shifting bargaining power from chip designers toward manufacturers controlling scarce advanced production capacity. (Reuters)

AI Boom Tightens Advanced Semiconductor Capacity

Artificial intelligence is creating one of the most powerful semiconductor investment cycles in recent history.

AI Accelerators Consume Leading-Edge Production

Training and operating advanced AI models requires powerful processors manufactured using some of the industry's most sophisticated processes.

Demand is coming from hyperscale technology companies, cloud providers and semiconductor designers developing GPUs, custom AI accelerators and networking processors.

TSMC has been the biggest beneficiary because it manufactures many of the world's most advanced AI chips.

The company reported record revenue momentum during 2026 and expects strong multi-year AI demand as customers continue building enormous computing infrastructure. (Reuters)

Capacity Cannot Expand Quickly

Semiconductor factories require years to build.

They also need extremely expensive lithography equipment, clean rooms, utilities and specialised engineering teams.

Demand for computing can therefore increase much faster than manufacturing supply.

That mismatch creates scarcity.

When advanced production lines are already heavily booked, semiconductor designers have fewer alternatives and foundries gain greater negotiating leverage.

Samsung Raises Foundry Prices by Up to 15%

Samsung's latest pricing decisions provide one of the clearest indications of the shift.

4nm and 5nm Customers Face Higher Prices

Samsung has increased prices for selected 4nm and 5nm contract manufacturing services by as much as 15% for new orders.

Some customers using selected 8nm technologies are also paying more.

Chinese and US customers have faced increases of roughly 10%-15% in certain cases, while increases for some Taiwan-based customers have been smaller. (Reuters)

The changes demonstrate that pricing power is spreading beyond the absolute newest process nodes.

Customers increasingly need access to several manufacturing generations simultaneously.

Pyeongtaek Line Is Fully Utilised

Samsung's SF4 4nm production line at Pyeongtaek is operating at full capacity.

This represents an important change for Samsung's foundry division, which has struggled with underutilisation and losses since 2022.

High utilisation improves profitability because the enormous fixed costs of semiconductor fabrication are distributed across more wafers.

Adding higher wafer prices creates further operating leverage. (Reuters)

TSMC Remains Centre of AI Manufacturing Boom

The industry's most important capacity constraint remains Taiwan Semiconductor Manufacturing Company.

TSMC Dominates Advanced Foundry Manufacturing

TSMC accounts for more than 70% of traditional global foundry revenue by some recent market estimates, compared with approximately 7% for Samsung.

Its technological leadership and manufacturing reliability make it the preferred foundry for many leading semiconductor companies. (Reuters)

AI demand has booked significant amounts of its advanced production capacity.

Customers unable to secure sufficient allocation are increasingly evaluating alternative suppliers.

That creates opportunities for Samsung, Intel and other manufacturers.

TSMC Has Signalled Interest in Higher Prices

TSMC management has acknowledged the extraordinary demand environment and previously indicated that the company would like greater pricing recognition for the value of its manufacturing capabilities.

The company continues investing aggressively in Taiwan and the United States to expand advanced-node capacity.

Its planned US semiconductor investment has risen to approximately $265 billion as it develops multiple fabs and advanced packaging infrastructure in Arizona. (Reuters)

SMIC Also Raises Wafer Prices

The pricing shift extends into China.

Utilisation Reaches Nearly 94%

Semiconductor Manufacturing International Corporation raised wafer prices as strong domestic demand pushed utilisation to approximately 93.7% during the second quarter.

SMIC shipped around 2.9 million 8-inch-equivalent wafers during the quarter, while average wafer pricing increased 5.7%.

Quarterly revenue exceeded $3 billion for the first time. (Reuters)

AI Demand Supports More Than Leading-Edge Chips

Much of the additional demand reaching SMIC is not limited to headline AI processors.

AI servers require numerous supporting semiconductors.

These include power-management chips, networking components and peripheral processors.

This demonstrates why the AI infrastructure boom can tighten manufacturing across several process nodes rather than only the most advanced 3nm or 5nm technologies.

Foundry Revenue Is Expected to Expand Sharply

Industry forecasts point toward unusually strong growth across semiconductor contract manufacturing.

TrendForce Sees Nearly 25% Growth

TrendForce expects global foundry revenue to increase approximately 24.8% in 2026 as AI processors and related chips drive demand.

The research firm has indicated that advanced-node and advanced-packaging capacity remain heavily loaded while AI infrastructure investment continues to accelerate. (TrendForce)

The growth is significant because the semiconductor industry has historically been highly cyclical.

Periods of shortage frequently generate strong pricing before new capacity eventually catches up.

Advanced Packaging Becomes Another Bottleneck

Manufacturing a semiconductor wafer is only one stage in creating an AI accelerator.

AI Chips Require Sophisticated Packaging

Advanced processors increasingly combine several chiplets with high-bandwidth memory inside complex packages.

These architectures require sophisticated packaging processes.

As a result, customers can secure wafer manufacturing capacity yet still face bottlenecks later in production.

Advanced packaging capacity has therefore become strategically valuable alongside leading-edge fabrication.

Industry estimates expect substantial packaging expansion during 2026, but demand remains strong. (Tom's Hardware)

High-Bandwidth Memory Adds More Pressure

AI accelerators need extremely fast access to memory.

HBM Is Essential to AI Servers

High-bandwidth memory provides significantly greater data throughput than conventional memory technologies.

Modern accelerators often package several HBM stacks close to the main processor.

Demand for AI servers has consequently transformed memory markets.

Samsung, SK Hynix and Micron have increased investment, but supply remains tight.

Strong memory demand can indirectly influence foundry capacity because semiconductor manufacturers must allocate capital and factory resources across competing product categories.

AI Spending Gives Chip Designers Strong Demand but Higher Costs

The shortage produces mixed economics for semiconductor companies without their own factories.

Fabless Designers Depend on Foundries

Companies including Nvidia, AMD, Qualcomm and many custom AI-chip developers design processors but rely on external manufacturers.

When foundry prices rise, production costs increase.

Companies can respond by:

absorbing the cost,

raising chip prices,

improving product mix,

or securing longer-term capacity agreements.

AI-chip companies currently possess significant pricing power of their own because computing demand remains strong.

That makes them more capable of absorbing higher manufacturing costs than companies operating in weaker consumer markets.

Hyperscalers Are Increasing Custom-Chip Demand

Cloud companies are no longer relying exclusively on merchant processors.

More Companies Design Proprietary Silicon

Google, Amazon, Microsoft and Meta are developing custom computing hardware to optimise AI workloads.

These processors still need external manufacturing.

The trend therefore creates additional foundry customers at precisely the moment manufacturing capacity is already constrained.

Broadcom and other semiconductor companies are helping hyperscale customers develop custom AI accelerators and networking chips.

This broadens the number of companies competing for advanced wafers.

Samsung Gains Opportunity to Challenge TSMC

The shortage gives Samsung something it has struggled to obtain consistently: customer leverage.

TSMC Customers Need Second Sources

Semiconductor companies increasingly want alternatives to relying entirely on TSMC.

Supply-chain diversification has become strategically important because of geopolitical risk and capacity scarcity.

Samsung is one of the few companies capable of manufacturing advanced processors at meaningful volume.

Recent relationships involving Tesla, Broadcom, Apple and other major technology companies provide Samsung with opportunities to rebuild its foundry pipeline. (Reuters)

Higher Prices Could Help Restore Profitability

Samsung's foundry division has been loss-making since 2022.

Higher utilisation combined with 10%-15% price increases can materially improve the business's economics.

The division could potentially move toward profitability if strong demand continues.

The challenge is maintaining customers after industry capacity eventually expands.

Intel Could Also Benefit From Foundry Tightness

Intel is attempting to establish itself as another major alternative advanced foundry.

Customers Want More Manufacturing Options

A diversified foundry market benefits customers by reducing dependence on one supplier.

Intel possesses advanced manufacturing assets in the United States and is investing heavily in next-generation process technologies.

The current shortage creates an opportunity to attract customers who might previously have defaulted automatically to TSMC.

However, Intel still needs to demonstrate consistent yields, execution and production economics.

Foundry customers place enormous importance on reliability because manufacturing problems can delay entire product launches.

China Faces Additional Capacity Constraints

Chinese semiconductor designers operate under another layer of difficulty.

Export Restrictions Limit Domestic Expansion

US restrictions on advanced semiconductor equipment have made it more difficult for Chinese manufacturers to build leading-edge production capability.

That increases demand for whatever domestic and overseas foundry capacity remains available.

Samsung's price increases have been particularly steep for some Chinese customers because these companies possess fewer manufacturing alternatives. (Reuters)

SMIC is simultaneously running factories at high utilisation as domestic customers seek local production. (Reuters)

Mature Nodes Are Experiencing Uneven Conditions

The semiconductor shortage is not universal.

AI Creates Highly Polarised Market

TrendForce expects advanced-node capacity to remain strong, while some 12-inch mature processes continue facing softer demand from consumer electronics.

At the same time, certain 8-inch technologies are tightening because manufacturers have reduced capacity while AI-related power-chip demand increases. (TrendForce)

This means the semiconductor industry cannot simply be described as experiencing one broad shortage.

Capacity is tight where AI infrastructure requires specific technologies.

Other segments can remain comparatively weak.

Consumer Electronics Could Face Higher Component Costs

Foundry pricing eventually travels through the technology supply chain.

Smartphones and PCs Compete With AI Infrastructure

Smartphones, computers and other consumer devices need many of the same semiconductor production resources.

AI customers can often tolerate higher chip prices because expensive data-centre processors generate substantial economic value.

Consumer-electronics manufacturers operate with thinner margins.

If foundry and memory costs remain elevated, device manufacturers may need to raise retail prices, reduce specifications or accept lower profitability.

Samsung has already warned that semiconductor shortages could remain significant through 2028. (Reuters)

Automotive Manufacturers Also Face Exposure

Vehicles contain increasing quantities of semiconductors.

Modern Cars Depend on Multiple Chip Categories

Advanced vehicles need processors for infotainment, driver assistance, power management, connectivity and battery systems.

Not every automotive chip uses a leading-edge process.

However, AI infrastructure demand can indirectly affect automotive supply by shifting capital and manufacturing capacity toward higher-value applications.

Selected 8nm price increases at Samsung illustrate how the pressure can spread beyond cutting-edge AI processors.

Higher Prices Encourage Massive Foundry Investment

Scarcity sends a powerful economic signal.

TSMC, Samsung and Intel Are Expanding Capacity

Foundries are investing tens of billions of dollars in new manufacturing plants.

Governments are also offering subsidies and incentives because semiconductor production has become a strategic national priority.

The United States, Japan, Europe and other regions want geographically diversified chip manufacturing.

However, new plants take years to build.

Current price increases can therefore persist before additional supply materially changes the market.

Long-Term Contracts Become More Attractive

Customers need confidence that manufacturing capacity will be available when new products launch.

Capacity Reservations Gain Value

Semiconductor companies may increasingly negotiate multi-year arrangements with foundries.

Long-term commitments give manufacturers revenue visibility.

Customers receive greater supply certainty.

Samsung has also been seeking longer-term chip agreements as demand strengthens. (Reuters)

The semiconductor industry could therefore move toward deeper strategic relationships between chip designers and manufacturers.

Higher Foundry Prices Could Support Next-Generation Technology

Leading-edge development is becoming more expensive.

Sub-2nm Manufacturing Requires Massive Investment

Each process generation requires new equipment, materials and engineering.

Extreme ultraviolet lithography systems alone represent enormous capital expenditure.

Higher wafer prices can help foundries finance these investments.

Customers ultimately pay for access to increasingly advanced technology.

The economics become especially favourable when leading-edge manufacturing capacity is scarce.

Supply Could Eventually Catch Up

The current pricing environment does not guarantee permanent scarcity.

Semiconductor Industry Remains Cyclical

High prices encourage investment.

Large manufacturers are expanding aggressively.

Chinese companies are adding capacity.

Governments are subsidising new fabs.

If AI demand eventually slows while new factories become operational, today's shortage could become future overcapacity.

This pattern has occurred repeatedly across semiconductor cycles.

The key uncertainty is whether AI infrastructure demand can remain strong enough to absorb the enormous new supply now being planned.

India Could Benefit From Global Capacity Diversification

The shortage has strategic implications for India's semiconductor ambitions.

Domestic Manufacturing Becomes More Valuable

India is developing semiconductor fabrication, assembly and packaging capabilities as part of its wider electronics-manufacturing strategy.

The country cannot immediately replace advanced foundries such as TSMC or Samsung.

However, global capacity tightness reinforces the importance of building additional production locations.

A more diversified semiconductor supply chain can reduce dependence on a handful of global manufacturing centres.

Indian Electronics Companies Face Higher Costs

In the near term, Indian manufacturers of smartphones, servers, automobiles and other electronic products remain exposed to international chip prices.

Higher foundry costs can eventually increase processor and component prices.

Companies operating in price-sensitive Indian markets may find it difficult to pass these costs entirely to consumers.

Pricing Power Could Reshape Foundry Profitability

For years, the foundry market has been defined by enormous capital requirements and intense technological competition.

AI is changing the balance.

Scarcity Rewards Manufacturers With Available Capacity

TSMC benefits from leading-edge dominance.

Samsung benefits from customers seeking alternatives.

SMIC benefits from high domestic utilisation.

Intel has an opportunity to establish itself as a credible second source.

The common advantage is ownership of scarce semiconductor manufacturing infrastructure.

The ability to physically produce advanced chips is becoming increasingly valuable as demand for AI computing expands.

Conclusion

The AI-chip capacity shortage is giving global semiconductor foundries a level of pricing power that many manufacturers have not enjoyed for years.

Samsung has raised selected foundry prices by as much as 15% as its 4nm production line reaches full utilisation, while SMIC has increased wafer pricing with factory utilisation approaching 94%. TSMC remains at the centre of the shortage as booming AI infrastructure investment keeps advanced production capacity heavily booked. (Reuters)

The imbalance is spreading across the semiconductor ecosystem. Advanced packaging, high-bandwidth memory, supporting power chips and selected mature nodes are also experiencing tight supply.

For chip designers, higher foundry prices increase production costs. For manufacturers controlling scarce fabrication capacity, the AI boom is creating stronger utilisation, better pricing and potentially substantially higher profitability.

The longer-term question is whether today's shortage becomes structural. New fabrication plants will eventually add capacity, but construction takes years. Until supply catches up with AI-driven demand, semiconductor foundries are likely to retain considerably greater negotiating power over the companies competing for access to the world's most valuable manufacturing lines.