Samsung Electronics: How Durable Is the Earnings Surge?

Samsung Electronics has moved from an earnings recovery into a period of record profitability, driven overwhelmingly by memory. The more important question is no longer whether earnings can rise over the next few quarters, but how long AI infrastructure spending, memory scarcity and Samsung’s improving position in advanced memory can sustain the cycle.

Key Takeaways

• Samsung’s current earnings strength is real but highly concentrated. Device Solutions generated KRW 89.2 trillion of operating profit in the second quarter of 2026, compared with KRW 89.5 trillion for the group as a whole.
• The near-term AI demand case remains stronger than the recent market debate suggests. Alphabet and Microsoft expect capital expenditure to rise further, while Meta and Amazon are maintaining exceptionally large infrastructure programs.
• AI CapEx cannot be extrapolated indefinitely. Free cash flow and depreciation are already becoming more important constraints for hyperscalers, making monetization and utilization increasingly important.
• Samsung’s smartphone position is stronger than its current mobile profitability. The company regained global shipment leadership in the second quarter even as its mobile business moved into an operating loss under component-cost pressure.
• Chinese competition is a material long-term risk, but the threat is uneven. Chinese suppliers are gaining ground fastest in conventional memory and cost-sensitive devices, while advanced HBM, premium smartphones and leading-edge manufacturing remain harder markets to penetrate quickly.

The Record Quarter Was Also a Warning About Earnings Concentration

Samsung Electronics reported consolidated revenue of KRW 171.5 trillion and operating profit of KRW 89.5 trillion for the second quarter of 2026. Both were quarterly records. Revenue increased 28% sequentially, while operating profit rose 56%.

The composition matters more than the headline. Device Solutions, which includes memory, System LSI and foundry, generated KRW 127.5 trillion of revenue and KRW 89.2 trillion of operating profit. By contrast, the MX and Networks businesses recorded KRW 33.2 trillion of revenue and an operating loss of KRW 0.7 trillion.

Source: Samsung Electronics, Second Quarter 2026 Results and Samsung Electronics’ 2Q26 earnings presentation.

This creates an important distinction between earnings growth and earnings breadth. Samsung’s earnings outlook can continue improving even while several consumer-facing operations remain weak, provided memory profits expand faster than losses elsewhere. That is currently happening.

It also means that a downturn in memory would have a much larger effect on consolidated earnings than the current headline numbers might suggest. Diversification exists at the revenue level, but considerably less so at the profit level.

Why Semiconductor Earnings Can Still Move Higher

Samsung’s own second-half outlook remains constructive. The company expects server demand to remain robust as AI infrastructure investment continues and agentic AI broadens the amount of computation required across cloud platforms. It expects demand growth in server DRAM, enterprise SSDs and HBM to accelerate sufficiently to keep the memory market undersupplied even as demand for mobile and PC products moderates.

The supply side is equally important. Samsung states that memory supply constraints are likely to persist despite efforts to increase production. That changes the earnings mechanism. The next phase of profit growth does not require consumer electronics volumes to boom. It requires high-value server products to continue absorbing available advanced memory capacity faster than new supply reaches the market.

HBM adds another layer. Samsung scaled HBM4 sales during the second quarter and shipped HBM4E samples to major customers. HBM is important not simply because it is another memory product, but because it shifts the revenue mix toward products whose economics depend on performance, qualification and integration rather than only commodity bit supply.

Foundry could eventually broaden the semiconductor earnings base. Samsung reported stronger advanced-node demand and HBM base-die activity during the second quarter. For the second half, it plans to ramp second-generation 2nm mobile production and expand 4nm LPU and HBM base-die sales.

The limitation is timing. Memory is already generating extraordinary profits. Foundry is still in the process of converting design wins and utilization improvements into durable earnings. Treating the two businesses as equally proven profit engines would therefore be premature.

The Critical External Variable Is AI Infrastructure Spending

The most important external assumption behind Samsung’s semiconductor outlook is that global AI infrastructure investment remains high. There is substantial evidence supporting that assumption through 2026 and into 2027.

Alphabet raised its full-year 2026 capital-expenditure outlook to $195 billion to $205 billion after spending $44.9 billion in the second quarter. It said the increase reflected faster delivery of capacity required to meet demand and continues to expect capital expenditure to increase significantly in 2027.

Microsoft ended its fiscal year with quarterly capital expenditure of $41 billion. Following an accounting-related change in the classification of certain leases, its calendar-2026 CapEx expectation is approximately $175 billion, while the underlying investment plan is unchanged. More importantly for forward demand, Microsoft expects fiscal-2027 capital expenditure to grow year on year and expects more than $50 billion of spending in the first quarter.

Meta expects 2026 capital expenditure, including principal payments on finance leases, of $130 billion to $145 billion. Amazon entered 2026 expecting roughly $200 billion of company-wide capital expenditure, with AI among the major drivers, and its Q2 results showed continued rapid growth in both AWS and its AI businesses.

Note: CapEx definitions differ across companies and should not be added mechanically. Amazon’s figure is company-wide rather than a pure AI-infrastructure number. Sources: Alphabet Q2 2026 earnings call, Microsoft FY2026 Q4 earnings call, Meta Q2 2026 results, and Amazon Q2 2026 results.

This evidence makes an abrupt near-term collapse in hyperscaler infrastructure spending difficult to establish as a base case. Both Alphabet and Microsoft still describe demand as exceeding available capacity. Cloud backlogs and revenue are also expanding.

But “AI CapEx will continue forever” is a much stronger claim than the evidence supports.

AI Spending Does Not Need to Grow Forever for Memory Demand to Stay Strong

The sustainability debate is often framed too narrowly. Samsung does not necessarily need hyperscaler capital expenditure to accelerate every year. It needs the installed AI infrastructure base, new system deployments and replacement cycles to generate enough demand for memory-intensive architectures to keep high-value memory supply tight.

This matters because each generation of AI infrastructure is not simply a repeat of the last. Higher-performance accelerators require more memory bandwidth and increasingly complex memory systems. Inference growth can also broaden memory requirements beyond HBM into conventional server DRAM and enterprise storage.

A plateau in hyperscaler CapEx at a historically elevated level could therefore remain supportive for memory demand. The more damaging scenario would be a combination of slower infrastructure spending and faster memory capacity expansion.

The distinction is critical. CapEx growth is a leading indicator, but utilization and memory intensity determine how much of that spending translates into Samsung’s addressable demand.

The Constraint Is Gradually Moving From Capital Availability to Return on Capital

The risk case should nevertheless be taken seriously because the financial cost of the AI buildout is becoming visible.

Alphabet recorded negative free cash flow in the second quarter as infrastructure investment accelerated. Amazon’s trailing-12-month free cash flow moved into negative territory, with the company explicitly attributing much of the increase in property and equipment spending to AI. Meta produced only modest free cash flow in the second quarter after more than $31 billion of capital expenditure.

Microsoft remains strongly cash generative, but it has also reported pressure on cloud gross margins from AI infrastructure investment and growing usage.

This does not imply an imminent CapEx reversal. It does imply a change in the question being asked by shareholders and management teams. The first phase of the AI infrastructure cycle was dominated by capacity availability. The next phase will increasingly be judged by capacity utilization and monetization.

That creates a more useful thesis breaker for Samsung than simply watching quarterly CapEx. If hyperscalers continue raising infrastructure budgets while AI and cloud revenue, backlog and utilization also rise, the spending cycle remains economically defensible. If spending rises while monetization weakens materially, capital discipline would eventually become a stronger constraint.

Smartphones Present the Opposite Earnings Problem

Samsung’s smartphone business shows how market leadership and earnings can diverge.

Counterpoint Research estimates that global smartphone shipments fell 11% year on year in the second quarter of 2026 as memory shortages and component inflation hit lower-priced devices. Samsung nevertheless regained the largest global shipment share at approximately 24%. Omdia separately estimates a 22% share and also ranks Samsung first.

The difference between the two estimates reflects methodology rather than a different competitive conclusion. Both show Samsung gaining relative ground while the wider smartphone market contracted.

Chinese Android vendors were more exposed to the entry and mid-range segments where component inflation is difficult to pass through. Counterpoint estimated Xiaomi at 12%, OPPO at 11% and vivo at 8% of global Q2 shipments, with all three experiencing greater pressure from the memory-cost shock.

Samsung’s scale, procurement capability and larger premium portfolio therefore become competitive advantages when components are scarce. The company can absorb cost inflation across a broader product portfolio and preserve flagship availability more easily than smaller competitors concentrated in price-sensitive segments.

Yet Samsung’s own MX results demonstrate the cost of doing so. The MX and Networks unit recorded an operating loss despite year-on-year revenue growth and resilient Galaxy S26 and A-series sales. Component inflation overwhelmed the benefit of market-share gains during the quarter.

This means smartphone share gains should not automatically be translated into consolidated earnings growth. The more important metric is whether Samsung can restore mobile margins while retaining the share captured during the component shortage.

The China Risk Is Real, but It Is Not One Market

The phrase “Chinese semiconductor competition” obscures important differences between products.

In conventional DRAM, Chinese competition is already material. Counterpoint estimates Samsung regained the largest global DRAM revenue share in the second quarter of 2026 at 39%, while CXMT reached roughly 7%. CXMT remains much smaller, but its revenue grew rapidly as domestic demand and capacity expanded.

The size of the gap argues against assuming an imminent displacement of Samsung’s DRAM leadership. The growth rate of CXMT argues equally strongly against dismissing the competitive threat.

The first pressure is likely to appear in standardized memory products where qualification barriers and product differentiation are lower. Additional Chinese capacity can increase competition even without taking the number-one position because marginal supply influences industry pricing.

HBM is a different competitive environment. Product qualification, thermal characteristics, stacking, advanced packaging, yield and synchronization with accelerator roadmaps create greater switching costs. Samsung’s progress in HBM4 therefore matters because leadership in conventional DRAM alone would offer less protection if the industry’s profit pool continues migrating toward AI memory.

Samsung should also not be described as the undisputed leader across every advanced-memory category. Competition with SK hynix and Micron remains intense. The more defensible conclusion is narrower: Samsung currently has sufficient manufacturing scale, technology breadth and customer access that Chinese competitors are unlikely to displace it quickly across the entire high-value memory portfolio.

Editor S’s Interpretation

1. The Earnings Cycle Is Stronger Than the Earnings Diversification

Evidence: Device Solutions generated nearly all of Samsung’s consolidated operating profit in Q2 2026, while the mobile business recorded a loss.

Causal mechanism: AI server demand and limited memory supply have created exceptional pricing and mix conditions. Those profits are large enough to overwhelm weaker consumer-facing businesses.

Interpretation: Samsung can continue posting higher consolidated earnings without a synchronized recovery across smartphones, televisions and appliances. The current cycle is therefore more durable than a conventional consumer-electronics recovery, but more concentrated than consolidated revenue suggests.

Time horizon: Most relevant through H2 2026 and 2027.

Confirming indicator: Continued growth in server memory revenue and HBM shipments while DS operating profit remains dominant.

Limitation: Concentration amplifies downside if memory prices or AI infrastructure demand reverse faster than expected.

2. The AI CapEx Risk Is Becoming a Return-on-Capital Question

Evidence: Major hyperscalers are still increasing infrastructure investment, but free cash flow and depreciation pressure are becoming more visible.

Causal mechanism: Scarce compute capacity has justified aggressive investment. As more capacity comes online, future budgets will increasingly depend on whether existing infrastructure produces sufficient AI, cloud and advertising revenue.

Interpretation: A sharp near-term collapse in AI investment is not the most strongly supported scenario. The more credible medium-term risk is a gradual moderation if monetization fails to keep pace with depreciation and financing requirements.

Time horizon: 2027 onward.

Confirming indicator: Cloud backlog, AI revenue and infrastructure utilization continuing to rise alongside CapEx.

Limitation: AI infrastructure investment is unusually concentrated among a relatively small number of companies, making supplier demand sensitive to changes in their capital-allocation policies.

3. Component Inflation Is Temporarily Strengthening Samsung’s Competitive Position in Smartphones

Evidence: Samsung gained smartphone shipment share during a quarter in which the global market contracted sharply and Chinese competitors faced greater pressure in cost-sensitive segments.

Causal mechanism: Scale, procurement capacity and premium-device exposure make rising memory costs easier for Samsung to absorb than for vendors heavily dependent on low-margin devices.

Interpretation: The same memory shortage that damages Samsung MX profitability can improve Samsung’s relative market position. If component costs later normalize, Samsung could emerge with a stronger installed base and competitive position than its current mobile earnings imply.

Time horizon: H2 2026 through 2027.

Confirming indicator: Samsung retaining global smartphone share while MX operating margins recover.

Limitation: Sustained memory inflation can eventually destroy consumer demand and reduce the value of further share gains.

Three Scenarios for the Next Stage

What to Watch

• Hyperscaler 2027 CapEx revisions: More important than short-term stock-market sentiment because they influence the duration of server-memory demand.
• AI infrastructure monetization: Cloud revenue, backlog, AI revenue and utilization should be compared with CapEx and depreciation.
• Samsung HBM4 and HBM4E execution: Shipment growth, customer qualification and mix will determine whether Samsung captures more of the highest-value memory profit pool.
• Conventional DRAM pricing: A slower rate of price increase is not necessarily bearish; an outright supply-driven decline would carry more significance.
• Foundry utilization and 2nm production: Repeatable high-volume execution would reduce Samsung’s dependence on memory earnings.
• MX profitability: Market-share gains become strategically more valuable if mobile margins recover.
• CXMT’s share and capacity expansion: The competitive risk should be measured by incremental supply and technology progression rather than headlines about Chinese self-sufficiency.

Risks and Thesis Breakers

The first thesis breaker would be a material reduction in AI infrastructure budgets across several hyperscalers accompanied by weaker cloud and AI demand. A single company’s quarterly CapEx reduction would be less informative than a coordinated shift in investment and utilization.

The second would be memory supply expanding faster than underlying demand. This could occur if incumbent producers and Chinese manufacturers add conventional memory capacity while AI infrastructure growth moderates.

The third is execution. HBM leadership requires repeated qualification and high-yield production across new generations. Samsung’s improving HBM4 position is strategically important, but future share should not be treated as guaranteed.

The fourth risk is consumer demand destruction. Memory scarcity currently gives Samsung a relative advantage in smartphones, but prolonged component inflation can ultimately shrink the addressable market even for leading vendors.

Finally, Samsung’s unusually high dependence on semiconductor profits means consolidated earnings can remain more cyclical than the breadth of its business portfolio initially suggests.

Sources and Methodology

This analysis was developed by comparing Samsung Electronics’ public financial disclosures with public information from major AI infrastructure buyers and independent semiconductor and smartphone market research. Analyst materials were used only to identify competing interpretations and potential risk factors; proprietary charts, target prices and report structures were not reproduced.

• Samsung Electronics, Second Quarter 2026 Results, July 30, 2026.
• Samsung Electronics, 2Q 2026 Earnings Call presentation, July 30, 2026.
• Alphabet, Q2 2026 Earnings Call, July 22, 2026.
• Microsoft, Fiscal Year 2026 Fourth Quarter Earnings Call, July 29, 2026.
• Meta Platforms, Second Quarter 2026 Results, July 29, 2026.
• Amazon, Second Quarter 2026 Results, July 30, 2026.
• Counterpoint Research, Q2 2026 global smartphone shipment analysis, July 13, 2026.
• Counterpoint Research, Q2 2026 global DRAM market analysis, August 3, 2026.

About the Author

Editor S writes independent analysis for Sector Foundry, focusing on companies, industries, technologies, and global value chains.

Disclaimer