Samsung Electronics is increasing its investment in semiconductor manufacturing as demand for artificial intelligence infrastructure pushes its high-bandwidth memory (HBM) chips to capacity. The South Korean technology giant said its HBM supply for 2027 is already sold out, while its capital expenditure for 2026 is expected to reach a record 84.4 trillion won, approximately $63 billion. The spending increase reflects the need to expand production capacity and strengthen Samsung’s position in the fast-growing AI memory market.
The announcement came as Samsung reported record quarterly financial results for July–September 2026, with operating profit reaching 107.4 trillion won. However, the company’s shares fell after its outlook appeared less optimistic than rival SK Hynix’s and it delayed announcing its next shareholder-return policy until January 2027. The developments highlight the opportunities and challenges facing semiconductor manufacturers as AI infrastructure investment accelerates and competition for advanced memory chips intensifies.
Key Takeaways
- 2027 supply sold out: Samsung says its high-bandwidth memory supply for next year is fully booked.
- Record investment: The company expects 2026 capital expenditure to reach 84.4 trillion won, up 31.7 trillion won from 2025.
- Record quarterly profit: Samsung reported third-quarter operating profit of 107.4 trillion won, up from 12.2 trillion won a year earlier.
- HBM4 momentum: Revenue from HBM4 more than tripled quarter-on-quarter in the third quarter.
- Intense competition: Samsung is trying to strengthen its position against SK Hynix in supplying advanced memory for AI accelerators.
- Investor concerns: Shares fell after the earnings announcement as investors assessed the company’s outlook and delayed shareholder-return decision.
Samsung Raises Investment as AI Chip Demand Surges
Samsung’s spending plans reflect a fundamental change in the semiconductor industry. AI data centers require large quantities of advanced memory to support the processors used to train and run increasingly capable models. As technology companies expand their computing infrastructure, demand for high-performance memory has grown rapidly.
Samsung expects its total 2026 capital expenditure to reach 84.4 trillion won, an increase of 31.7 trillion won compared with the previous year. Of that amount, 76.8 trillion won is expected to go toward its Device Solutions division, which includes its semiconductor businesses.
The investment is intended to support new manufacturing facilities, production equipment and advanced semiconductor technologies. Expanding capacity is not an immediate process, however. Building a fabrication plant and installing equipment can take years, meaning increased spending does not automatically translate into a rapid rise in chip supply.
Samsung’s challenge is to invest enough to meet future demand without creating excess capacity if AI infrastructure spending eventually slows.
Why Samsung’s AI Memory Chips Are Sold Out for 2027
High-bandwidth memory, or HBM, is a specialized form of DRAM designed to deliver large amounts of data to processors quickly. It is particularly important for AI accelerators, where memory bandwidth can influence how efficiently large models are trained and operated.
Samsung said its HBM supply for 2027 has already been sold out, with major customers continuing to request additional volumes. The company has also signed long-term memory supply agreements with 10 major customers and is negotiating with another 10.
The situation reflects the growing importance of securing memory supplies in advance. AI infrastructure operators need predictable access to components when building data centers, while manufacturers want clearer demand visibility before committing to expensive production expansions.
However, a sold-out supply position does not mean Samsung can immediately satisfy every request for more chips. Production capacity, manufacturing yields, product qualification and the availability of equipment all affect how much advanced memory can be delivered.
HBM4 Gives Samsung an Opportunity to Close the Gap
Samsung is also working to improve its competitive position in HBM, a market where SK Hynix has established a strong position with major AI customers.
The company’s third-quarter results indicated progress in its next-generation products. Samsung said HBM4 revenue increased more than threefold compared with the previous quarter, while HBM4 accounted for around 60% of its total HBM revenue.
HBM4 is designed to support the demanding memory requirements of next-generation AI accelerators. Stronger sales could help Samsung win more business as customers diversify suppliers and seek additional capacity.
Samsung also said it had shipped samples of HBM4E, its next-generation product, in May 2026 and received positive initial feedback from customers. The company plans to begin mass production of HBM4E in the first half of 2027.
The next challenge is converting product progress into sustained market share and profitable production. Advanced memory chips require close coordination with processor manufacturers, and customer qualification can influence when a supplier’s products enter large-scale deployment.
Samsung’s Third-Quarter Results at a Glance
Samsung’s financial results demonstrate how strongly the AI-driven memory cycle has affected its business.
| Financial metric | Q3 2026 | Q3 2025 | Change |
|---|---|---|---|
| Operating profit | 107.4 trillion won | 12.2 trillion won | About 8.8 times higher |
| Revenue | 195.3 trillion won | Not specified here | Record quarterly revenue |
| Semiconductor operating profit | 106.5 trillion won | — | Approximately 15-fold increase |
| Expected full-year 2026 capital expenditure | 84.4 trillion won | 52.7 trillion won implied by reported increase | About 60% higher |
Sources: Reuters and Associated Press. Capital expenditure figures refer to full-year estimates, not quarterly spending.
The semiconductor division generated nearly all of Samsung’s third-quarter operating profit, illustrating the growing contribution of memory chips to overall earnings. At the same time, the company’s wider business faces pressure from rising component costs, including higher memory prices that can squeeze margins in consumer electronics.
Samsung’s results therefore reflect both the strength of AI-related demand and the risks of relying heavily on a cyclical industry.
Why Samsung Shares Fell Despite Record Profits
Samsung’s record results did not prevent a negative market reaction. Its shares fell as much as 6.2% following the announcement, according to Reuters.
One factor was the company’s outlook relative to SK Hynix. Samsung expects tight memory supply conditions to persist through 2028, while SK Hynix has indicated that shortages could last longer. Investors also compared Samsung’s capital-spending plans and shareholder-return decisions with its rival’s announcements.
Samsung postponed its next shareholder-return policy until January 2027, disappointing investors who had expected more immediate clarity on dividends or share buybacks.
The reaction illustrates a distinction between strong current earnings and expectations for future returns. Investors assess not only how much profit a company generates today, but also whether it can sustain margins, expand its competitive position and return cash to shareholders.
Samsung still needs to demonstrate that its investments in advanced memory and manufacturing can translate into lasting competitive gains.
What the AI Chip Shortage Means for the Semiconductor Industry
The supply shortage is affecting more than the companies producing AI accelerators. Advanced memory competes for manufacturing resources and investment with other semiconductor products, while new capacity requires significant time and capital to bring online.
Samsung expects AI-related memory demand to grow faster than supply in 2027. It also plans to expand long-term supply agreements, aiming for contracts covering 60% to 70% of production capacity by 2030.
These agreements could give manufacturers better visibility into future demand and help customers secure components for planned data-center deployments. However, they cannot eliminate all uncertainty around future AI spending, technology changes or the economics of building additional capacity.
For customers, limited supply can mean higher component prices and longer planning cycles. For chipmakers, strong demand can support earnings, but the industry must avoid investing on the assumption that current shortages will last indefinitely.
The Bigger Picture
Samsung’s investment plans show how the AI boom is reshaping semiconductor manufacturing. Memory has become a critical part of AI infrastructure, and manufacturers are competing to supply the components required by increasingly powerful processors. Securing long-term contracts, advancing HBM technology and expanding production capacity are now central to Samsung’s strategy.
Yet the company’s next phase will involve more than spending money. It must convert investment into dependable supply, maintain product quality, improve its competitive position and demonstrate that its capital allocation creates value for shareholders. The balance between expansion and financial discipline will help determine whether today’s AI-driven earnings strength can be sustained.
FAQs
1. Why is Samsung increasing its semiconductor spending?
Samsung is increasing investment to meet growing demand for AI memory and other advanced chips. Its 2026 capital expenditure is expected to reach 84.4 trillion won, including 76.8 trillion won for its semiconductor-focused Device Solutions division.
2. What does it mean that Samsung’s AI chips are sold out for 2027?
Samsung says its HBM supply for 2027 has been fully booked. This indicates strong demand for its planned production, but it does not mean every type of Samsung chip is sold out or that all additional customer demand can be fulfilled.
3. What is HBM4, and why is it important?
HBM4 is a generation of high-bandwidth memory designed to support AI accelerators and other demanding computing applications. Its ability to move data quickly makes it an important component in modern AI infrastructure.
4. How did Samsung perform financially in the third quarter of 2026?
Samsung reported operating profit of 107.4 trillion won and revenue of 195.3 trillion won for July–September 2026. Semiconductor operating profit reached 106.5 trillion won, driven by strong AI-related memory demand.
5. Why did Samsung’s shares fall after its results?
Investors assessed the company’s outlook against rival SK Hynix, its plans for shareholder returns and questions about the durability of AI-related demand. Samsung postponed its next shareholder-return policy announcement until January 2027.
Looking Ahead
Samsung’s near-term priorities will include expanding advanced-memory production, improving its HBM4 position and preparing HBM4E for mass production. The company’s ability to meet customer requirements and convert long-term supply agreements into profitable sales will be central to its efforts to strengthen its position in the AI semiconductor market.
The broader outlook will depend on how quickly AI infrastructure investment grows, how much new chipmaking capacity becomes available and whether customers continue committing to expensive AI systems. Samsung’s record results reflect strong demand today, but its long-term performance will depend on execution, competition and disciplined investment.
Get the day’s top stories in your inbox
One concise email. No spam, unsubscribe anytime.
