Samsung and SK Hynix Face Sky-High Expectations as Q3 Earnings Loom, Testing Global AI Trade

South Korea's two largest chipmakers, Samsung Electronics and SK Hynix, are at the center of a global semiconductor "super cycle," driven by surging AI demand and advances in next-generation memory technology. As both companies approach their third-quarter earnings reports, their performance will serve as a critical litmus test for the durability of the "AI trade" in global markets.

Market Expectations Run High

Market expectations for both giants are exceptionally high. According to Seoul Economic Daily citing data from FnGuide:

• Samsung Electronics is projected to post 199.1 trillion won in revenue and 105.6 trillion won in operating profit for Q3.
• SK Hynix is expected to report 94.1 trillion won in revenue and 74.1 trillion won in operating profit.

The combined operating profit of the two companies is approaching a record 189.9 trillion won, a direct reflection of strong market momentum amid the current chip supply shortage.

Analyst Estimates Show Subtle Downward Adjustments

Despite the headline figures, market consensus has seen modest revisions in recent months. Over the past three months, analysts have:

• Lowered Samsung Electronics' revenue and profit estimates by 2.6% and 4.4%, respectively.
• Lowered SK Hynix's estimates by 5.3% and 5%, respectively.

These adjustments are primarily attributed to the strengthening of the Korean won, which has reduced the value of dollar-denominated sales, rather than a downturn in the memory market itself. However, the shift is sufficient to prompt a reassessment of how currency fluctuations directly impact the short-term earnings of tech giants.

Deeper Signals Beyond the Numbers

Beyond the surface-level figures, the upcoming Q3 reports will convey deeper signals to the market. Investors are closely monitoring:

• The sustainability of memory chip price increases
• The profit-locking effect of long-term supply agreements (LTAs)
• Whether next-generation high-bandwidth memory (HBM) can translate into tangible bottom-line profits

These factors will help determine whether the current boom is a short-lived spike or a structurally resilient shift.

DRAM and NAND Price Growth Slows; LTAs Smooth Volatility

The sharp rise in average selling prices (ASP) for DRAM and NAND flash memory in Q2 was the primary engine driving strong performance for both companies.

In Q3, prices are expected to continue rising, but at a slower pace. According to Mirae Asset Securities:

• Samsung Electronics' Q3 DRAM ASP is expected to climb 16.5%, but Q4 growth is forecast to narrow to 5.4%, suggesting the price rebound may be losing some momentum.

This moderation is partly due to the growing prevalence of long-term supply agreements (LTAs). Currently:

• Samsung Electronics has locked in 60% to 70% of its memory capacity through LTAs.
• SK Hynix completed LTA negotiations with approximately ten core clients in Q2.

While these long-term contracts provide stability and smooth out the historically volatile memory cycle, they also cap upside profit potential during rapid price increases, as contract prices often lag behind spot prices.

HBM4 Production Ramps Up; Battle for AI Infrastructure Core Heats Up

The third quarter marks a turning point in the next-generation high-bandwidth memory (HBM4) sector, which is essential for advanced AI applications.

• Samsung Electronics expects its Q3 HBM4 sales to more than triple compared to Q2, driven by increased shipments to Nvidia's next-generation AI platform, Vera Rubin. HBM4 is projected to account for over 60% of Samsung's total HBM revenue in the second half of 2026.
• SK Hynix faces the challenge of expanding HBM4 production while maintaining its market leadership in HBM3E.

The Q3 earnings will provide early clues about the landscape of this high-stakes race.

Demand for HBM Reshapes Tech Industry Logic

The hunger for high-bandwidth memory is reshaping the underlying logic of the entire tech industry. According to Global Economics citing a report from KB Securities:

As Meta launches its agent-based AI application Muse, AI services are entering a new era requiring continuous, sequential operations and higher memory bandwidth.

Traditional generative AI processes approximately 100 tokens per second, while agent-based AI demands 1,000 tokens per second. This tenfold increase in data processing volume is fundamentally altering data center architecture.

KB Securities analyst Kim noted:

"In AI data centers, the GPU is the heart, the CPU is the brain, electricity is the oxygen, and memory is the circulatory system that constantly supplies and cycles data. Ultimately, the key to AI data center performance and efficiency will shift to memory."

This trend is already visible across the broader industry. AMD recently joined the "trillion-dollar market cap club" alongside Nvidia, Broadcom, and Micron, and its Q2 data center revenue surged 107% year-over-year, indicating that market attention on CPUs and memory is now on par with GPUs.

In response, Samsung Electronics is already eyeing next-generation "zHBM" technology, which involves stacking memory directly on top of GPUs, with plans to deliver samples by the end of 2027.

Foundry Business Sees a Turnaround; "Chip Inflation" Tests End-User Segments

Beyond its core memory business, Samsung Electronics' non-memory divisions are also under the market's microscope.

After consecutive losses since 2023, Samsung's foundry business is widely expected to return to profitability in Q3, driven by expanded 4nm language processing unit (LPU) capacity and improved yields. If achieved, this would significantly boost investor confidence and remove a long-standing drag on Samsung's overall valuation.

However, Samsung's Device Experience (DX) division, responsible for smartphones and televisions, is facing a "chip inflation paradox."

• In Q2, the DX division recorded an operating loss of 80 billion won, as strong sales of high-end devices like Galaxy Foldables failed to offset rising component costs.
• The surge in memory prices, which benefits the semiconductor business, has become a heavy burden for the end-user hardware division.

The core question for Q3 is whether high-end product sales can absorb these cost pressures, or whether the "profit on one side, loss on the other" dilemma will persist.

Navigating the "Peak" Debate; Reshaping Long-Term Valuation Logic

As the Q3 earnings season approaches, the global semiconductor industry's focus is shifting from the absolute scale of profits to their sustainability.

• If Samsung Electronics and SK Hynix can demonstrate that long-term agreements have effectively locked in demand, that HBM4 is generating strong profits, and that the foundry business is back on track, the current super cycle will prove more resilient than previous ones.
• Conversely, if DRAM and NAND price growth slows faster than expected, or if HBM4 profit contributions fall short, the debate over whether the semiconductor industry has already "peaked" will inevitably reignite.

In the coming weeks, these two earnings reports will not only determine the short-term market capitalization of both giants but will also set the tone for global technology infrastructure investment in the years ahead.

Risk Disclaimer: Market conditions carry risks, and investment decisions should be made with caution. This article does not constitute personal investment advice and does not take into account the specific investment objectives, financial situations, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their particular circumstances. Investment decisions made based on this content are at the user's own risk.