[webinar] Embracing Digital Transformation in Maintenance & Plant Operations | March 13 at 10AM EST – Register Now

The New Semiconductor Shortage: How AI Is Reshaping Chip Supply, Prices and Manufacturing

Neeta Verma, Environmental Compliance Leader | Enterprise Quality Governance | Technical Author

Posted 9/23/2026

semiconductor shortage

The semiconductor industry is entering another period of supply disruption. But unlike the shortage that disrupted global manufacturing during the pandemic, the problem emerging in 2026 is increasingly structural.

The pandemic shortage was driven largely by factory shutdowns, logistics bottlenecks and sudden changes in consumer demand. Today’s shortage is being driven by something more fundamental: artificial intelligence is changing what semiconductor capacity is used for, who gets priority and how much companies must pay to secure supply.

The effects are already reaching consumers and manufacturers.

On September 10, Reuters reported that Chinese AI-chip companies including Huawei and Cambricon had raised processor prices because of the global shortage of high-bandwidth memory (HBM). Huawei’s upcoming Ascend 950DT accelerator reportedly rose 20% to 50%, while Cambricon’s next-generation 690 chip increased 20% to 30%. [1]

Apple has also been passing higher memory and storage costs to customers. In June, it raised prices on MacBooks, iPads and other products, citing soaring memory and storage costs associated with AI data-center demand. [2]

And the pressure has now reached the iPhone. On September 10, Reuters reported that Apple had raised iPhone prices in India by as much as 41%, with higher memory and storage costs linked to surging AI demand among the factors behind the increases. [3]

HP offers another indication of the scale of the problem. The company said memory and storage had risen from about 15% to 18% of PC bill-of-materials costs to approximately 35%, while memory costs had roughly doubled sequentially. [4]

These developments show that the semiconductor shortage is no longer simply a chip-industry problem. It is becoming a pricing, manufacturing and supply-chain resilience problem across the technology economy.

AI Is Reallocating Semiconductor Capacity

The heart of the current shortage is memory.

AI accelerators require enormous amounts of HBM because AI workloads depend on moving large volumes of data rapidly between processors and memory. At the same time, AI data centers require conventional DRAM, server memory and enterprise NAND storage.

This creates a capacity conflict.

According to KB Securities, global hyperscaler AI infrastructure investment could reach approximately $1.3 trillion in 2027, about 60% higher than the previous year. KB estimates that memory could account for 57% of AI infrastructure spending in 2027, compared with 14% in 2025. TrendForce has estimated an even higher share. [5]

The result is that memory manufacturers have a powerful financial incentive to prioritize high-value AI memory.

That does not mean conventional memory disappears. It means the same manufacturing ecosystem is being asked to satisfy two rapidly growing markets at once.

This is why the current shortage can affect PCs, smartphones and servers even though the immediate driver is AI infrastructure.

HBM4 Makes the Capacity Equation More Difficult

The transition to HBM4 could intensify the pressure.

KB Securities estimates that HBM4 requires roughly three times the wafer capacity of conventional DRAM. [5]

This is one of the most important characteristics of the current shortage.

Increasing HBM production does not simply create additional memory supply. It consumes manufacturing resources that could otherwise be used for conventional DRAM.

For memory manufacturers, the economics are compelling. HBM is essential to advanced AI accelerators and carries significantly higher value than many conventional memory products.

For other electronics manufacturers, however, the consequence can be tighter availability and higher prices for standard memory.

This creates a new type of semiconductor shortage: not necessarily a shortage of total manufacturing capacity, but a shortage of the right capacity for the right product at the right time.

semiconductor shortage higher demand tighter supply rising prices manufacturing pressure

Memory Inventories Are Already Extremely Tight

The pressure is becoming visible in supplier inventories.

KB Securities reported that Samsung Electronics and SK hynix had memory inventories of less than 10 days of supply in the third quarter. It also expects DRAM and NAND demand growth in 2027 to exceed supply growth by more than 10 percentage points. [5]

That is significant because memory is traditionally a highly cyclical business. Manufacturers normally adjust production to balance supply and demand.

AI is disrupting that balancing mechanism.

Demand from hyperscalers is strong enough to absorb increasing quantities of high-value memory, while manufacturers cannot rapidly create equivalent new capacity.

Industry reports have also indicated that Samsung has reportedly targeted or allocated a substantial share of its memory capacity to long-term agreements extending through 2031, highlighting how aggressively major customers are seeking to secure future supply. [6]

The result is a market in which strategic customers with long-term agreements can be in a much stronger position than buyers relying on spot-market availability.

Prices Are Beginning to Reflect the Imbalance

The shortage is also visible in memory pricing.

Market reports have indicated extreme premiums for some HBM products, with certain HBM3E configurations reportedly trading at several times their long-term contract prices. [6]

The precise price of an individual HBM product should not be treated as representative of the entire memory market. However, the wider trend is clear: scarce AI memory has significant pricing power.

This also helps explain why AI-chip manufacturers themselves are beginning to raise prices.

The September 10 Reuters report on Huawei and Cambricon is particularly important because it shows the shortage moving one step further through the supply chain. [1]

AI companies are not simply paying more for memory. Some are now passing those higher component costs into the prices of the processors they sell.

Consumer Electronics Are Feeling the Pressure

Apple’s pricing decisions provide one of the clearest examples.

In June 2026, Apple raised prices for several MacBook and iPad models as memory and storage costs increased. At that time, iPhone prices were not increased in the same round. [2]

That changed later in the year.

Following Apple’s September product launch, Reuters reported that iPhone prices in India increased by as much as 41%. The company had previously cited rising memory and storage chip costs, driven by AI-related data-center demand, as a reason for global pricing pressure. [3]

The two events should therefore be viewed separately, but together they illustrate the same trend: memory inflation is increasingly reaching the consumer.

HP’s experience reinforces the point. Its executives said memory costs had approximately doubled sequentially and that memory and storage had risen to roughly 35% of PC bill-of-materials costs. [4]

When a component category becomes that large a share of manufacturing cost, companies have limited ability to absorb the increase indefinitely.

They must either raise prices, accept lower margins, redesign products or secure supply through long-term agreements.

A Shortage Does Not Have to Involve Advanced Chips

Another important lesson from recent semiconductor disruptions is that technological sophistication does not determine supply-chain importance.

The 2025 Nexperia disruption demonstrated this in the automotive industry.

Nexperia supplies high volumes of relatively simple semiconductors used in vehicle electronic control systems. When supply was disrupted, the European Automobile Manufacturers’ Association warned that vehicle production could be threatened. Alternative suppliers existed, but qualifying those suppliers and building additional production capacity could take several months, while existing inventories could last only a few weeks. [7]

This distinction is critical:

Physical availability is not the same as usable supply.

A substitute semiconductor may exist in the market, but that does not mean an automaker, medical-device manufacturer or industrial-equipment producer can immediately install it.

Engineering validation, reliability testing, regulatory requirements, customer approval and production qualification can turn a theoretically available component into a months-long replacement project.

semiconductor manufacturing

New Fabs Cannot Solve an Immediate Shortage

The obvious response to a structural shortage is to build more capacity.

That is already happening.

Memory manufacturers and other semiconductor companies are investing billions of dollars in new fabs, packaging facilities and advanced manufacturing technologies. Governments are also providing incentives through programs such as the U.S. CHIPS Act.

But semiconductor capacity takes years to develop.

SK hynix’s Indiana project illustrates the timeline. The company is investing more than $4 billion in an advanced facility, with cleanroom operations expected in 2028 and volume production of next-generation HBM4E targeted for the third quarter of 2029. Its CEO has said the current memory shortage could persist through 2030. [8]

This is why today’s investment announcements should not be confused with immediate supply relief.

A fab announced in 2026 cannot necessarily provide meaningful additional output in 2026 or 2027.

The industry must construct the facility, install equipment, qualify processes, improve yields and then qualify production for customers.

Workforce Capacity Matters Too

Physical factories are only part of the capacity equation.

Semiconductor manufacturing requires highly specialized engineers, technicians, scientists and other skilled workers.

A 2023 Semiconductor Industry Association and Oxford Economics study projected that the U.S. semiconductor industry could add approximately 115,000 jobs by 2030, while about 67,000 could remain unfilled at current education and training rates. [9]

These are 2023 projections, not current 2026 workforce statistics, but they illustrate an important structural constraint.

If countries build new fabs faster than they build the workforce needed to operate them, some of the expected capacity expansion will take longer to reach its full potential.

Geopolitics Adds Another Layer of Risk

The semiconductor shortage is also developing within a fragmented geopolitical environment.

Export restrictions, technology controls and national semiconductor strategies are affecting where companies can obtain advanced components and how efficiently they can build AI systems.

China provides a particularly important example.

Reuters reported that Chinese AI-chip companies have been paying more for HBM as U.S. restrictions on advanced HBM exports add to an already constrained market. Some companies have turned to more expensive sourcing channels to obtain the memory they need. [1]

This creates a feedback loop.

AI demand increases HBM requirements. Export restrictions constrain access for some buyers. Restricted buyers compete for alternative supply. Higher procurement costs then feed into AI-chip prices.

The result is that semiconductor supply risk is becoming increasingly intertwined with geopolitical risk.

More Capacity Does Not Necessarily Mean More Resilience

The industry is investing heavily in new semiconductor capacity, but resilience requires more than additional fabs.

Companies also need visibility into:

  • HBM and conventional DRAM capacity allocation
  • Advanced packaging and testing capacity
  • Supplier inventories
  • Long-term memory agreements
  • Geographic concentration
  • Export restrictions
  • Approved alternative suppliers
  • Component qualification timelines
  • Product redesign options
  • Strategic inventory for critical components

This represents a shift from traditional procurement toward capacity intelligence.

Knowing that a supplier has semiconductor capacity is no longer enough.

Companies need to know whether the supplier has the specific technology, package, memory configuration, qualification status and production allocation required for their products.

What Happens Next?

The evidence points toward a semiconductor market that may remain unusually tight even as manufacturers invest aggressively in new capacity.

AI infrastructure spending continues to expand. HBM demand is consuming substantial manufacturing resources. Conventional memory demand remains strong. Consumer electronics companies are already facing higher component costs. And major new memory facilities will take years to reach meaningful production.

The key issue is therefore not simply whether the world has enough semiconductor manufacturing capacity.

It is how that capacity is allocated.

AI is changing the economic value of semiconductor production. High-value HBM and AI-related components are receiving increasing priority, while traditional electronics continue competing for the remaining capacity.

That means the next semiconductor shortage may look very different from the pandemic shortage.

There may be no single factory shutdown, port closure or sudden logistics crisis at its center.

Instead, the defining issue may be allocation, qualification and access.

For manufacturers, the lesson is clear: semiconductor risk is no longer simply a procurement issue.

It is becoming a strategic business issue involving technology, product design, financial planning, geopolitics, supplier qualification and long-term capacity strategy.

The companies best positioned for the next phase of the semiconductor cycle will not necessarily be those that simply hold the most inventory.

They will be those that understand where capacity is going, which components are becoming structurally scarce and how quickly they can respond when the market changes.


References


[1] Reuters, September 10, 2026. China’s AI chipmakers raise prices as high-bandwidth memory shortage bites. https://www.reuters.com/world/asia-pacific/chinas-ai-chipmakers-raise-prices-high-bandwidth-memory-shortage-bites-2026-09-10/

[2] Reuters, June 25, 2026. Apple raises prices of MacBooks, iPads as memory costs skyrocket. https://www.reuters.com/world/asia-pacific/apple-raises-prices-macbooks-ipads-memory-costs-skyrocket-2026-06-25/

[3] Reuters, September 10, 2026. India gets one of world’s steepest iPhone price hikes. https://www.reuters.com/business/retail-consumer/india-gets-one-worlds-steepest-iphone-price-hikes-2026-09-10/

[4] HP Inc., Q1 FY2026 Earnings Call Transcript, February 24, 2026. https://s203.q4cdn.com/918857832/files/doc_financials/2026/q1/HPQ-Q1-2026-Transcript.pdf

[5] KB Securities / Seoul Economic Daily, September 2026. Analysis of AI infrastructure investment and the global memory supply outlook. https://en.sedaily.com/finance/2026/09/07/samsung-sk-hynix-memory-stockpiles-fall-below-10-days

[6] TrendForce, September 1, 2026. HBM3E Spot Prices Said to Be 4–5× LTA Levels, with Samsung Reportedly Locking 70% of Memory Capacity Through 2031. https://www.trendforce.com/news/2026/09/01/news-hbm3e-spot-prices-said-to-be-4-5x-lta-levels-with-samsung-reportedly-locking-70-of-memory-capacity-through-2031/

[7] European Automobile Manufacturers’ Association (ACEA), October 16, 2025. ACEA calls for quick resolution to critical chip supply shortage. https://www.acea.auto/press-release/acea-calls-for-quick-resolution-to-critical-chip-supply-shortage/

[8] Reuters, August 27, 2026. SK hynix to start AI chip output in Indiana in 2029, sees memory shortage through 2030. https://www.reuters.com/world/asia-pacific/sk-hynix-holds-groundbreaking-ceremony-4-billion-indiana-ai-chip-packaging-2026-08-27/

[9] Semiconductor Industry Association / Oxford Economics, July 2023. Chipping Away: Assessing and Addressing the Labor Market Gap Facing the U.S. Semiconductor Industry. https://www.semiconductors.org/chipping-away-assessing-and-addressing-the-labor-market-gap-facing-the-u-s-semiconductor-industry/


avt-img

Neeta Verma

Designation: Environmental Compliance & Quality Governance Leader | Technical Author & Industry Thought Leader

LinkedIn: https://www.linkedin.com/in/neeta-verma-7384a8a/

Email: nv2000x@yahoo.com

Author Bio:

Neeta Verma is a senior environmental compliance and enterprise quality governance leader with over 27 years of experience in the global electronics industry. Her expertise includes product environmental compliance, regulatory strategy, product stewardship, regulatory intelligence, and enterprise quality governance.Her work connects regulatory intelligence, compliance execution, and business strategy - helping organizations anticipate change, reduce risk, and build sustainable, resilient compliance practices across global operations.

An internationally published technical author and industry thought leader, she has contributed to The Manufacturer, Compliance Week, Quality Digest, Quality Magazine, Industry Today, Electronics for You and Maintenance World. Her LinkedIn thought leadership covers environmental compliance, sustainability, product stewardship, regulatory strategy, and emerging regulatory developments.

She holds a Diploma in Electronics & Communication Engineering, a Bachelor of Computer Applications (BCA), and an MBA in Environmental Management.

Picture of Brawley

Brawley

Join the discussion

Click here to join the Maintenance and Reliability Information Exchange, where readers and authors share articles, opinions, and more.

"*" indicates required fields

This field is for validation purposes and should be left unchanged.

Get Weekly Maintenance Tips

delivered straight to your inbox

"*" indicates required fields

This field is for validation purposes and should be left unchanged.