The stranger number is not 107.4 trillion. It is 111.9 trillion.

On October 8, Samsung Electronics projected consolidated third-quarter 2026 revenue of KRW 195 trillion and operating profit of KRW 107.4 trillion, up 782.5% from a year earlier.

Meritz Securities, however, estimated KRW 111.9 trillion in operating profit for memory alone. It also projected losses of roughly KRW 1.8 trillion in Mobile eXperience and networks, and KRW 0.4 trillion in visual display and home appliances.

These are analyst estimates, not Samsung's finalized segment results. But they expose the central tension: two sides of the same company can experience the memory boom under very different economic conditions.

Prices did more work than volume

According to Meritz estimates reported by Seoul Economic Daily, Samsung's DRAM shipments rose 3.4% quarter on quarter, while average selling prices rose 20.1%. For NAND, estimated shipment growth was 4.2% against a 19% rise in average prices.

The widening difference matters. AI servers and data centers are competing for memory supply, strengthening the position of manufacturers that can decide where limited capacity goes.

A record quarter is therefore not simply a story about producing and selling more chips. It is also a story about who has leverage over the price of scarce components.

HBM and mobile memory are not the same product

High-bandwidth memory for AI accelerators and the LPDDR used in smartphones serve different technical needs. One price does not mechanically determine the other.

Yet production investment and capacity allocation across memory categories are not completely independent. TrendForce reported in July that suppliers' preference for AI-related applications was keeping mobile LPDRAM supply tight.

This is an allocation problem, not a claim that all memory products move in lockstep. Demand from one market can change the constraints faced by another.

Samsung is both the seller and the buyer

For Samsung's memory business, a higher selling price can lift earnings. For its Galaxy and consumer-device businesses, memory is a purchased input that must fit inside a viable product price.

TrendForce projected that smartphone makers could try to pass rising mobile DRAM costs to consumers, at the risk of weaker handset demand.

Device makers face a trade-off: pass costs through and risk sales, or absorb them and risk margins. Shipments, product mix, promotion, currency and other component costs also matter, so memory prices alone cannot explain the division's estimated losses.

The question shifts from memory capacity to performance per gigabyte

When memory becomes more expensive, supplier negotiation is only one response. Device teams can also revisit memory configurations by model, application resource usage, operating-system memory management, and hardware-software design choices.

This is not evidence of a confirmed hiring boom in memory optimization. Banseog's inference is narrower: sustained input-cost pressure may make cross-functional expertise in system performance and product architecture more valuable.

The useful engineering question is no longer only how many gigabytes to install. It is how much customer experience a product can sustain with each gigabyte.

One company, different capability priorities

Memory suppliers may place greater emphasis on capacity allocation, customer contracting and profitability by product. Device makers may need tighter coordination across sourcing, cost engineering, software optimization and product planning.

Sharing a corporate parent does not make these business problems identical. A hiring or training plan should begin by asking whether that operation has pricing power or must adapt to prices set elsewhere.

A practical leadership review can test three things: exposure to memory cost by product tier; room to improve user experience without more memory; and whether contract changes or design changes offer the faster response.

What is established and what remains uncertain

Samsung's October 8 statement provides only group-level preliminary results. KRW 111.9 trillion in memory profit and the losses estimated for its device-related businesses come from Meritz Securities, not a company-confirmed segment breakdown.

Sales volume, competitive pricing, currency movements, promotion, other semiconductor segments and launch cycles will all influence the final picture. A forecast segment loss should not be presented as a confirmed Galaxy operating loss.

The correct use of this early data is to test how pricing power and cost burdens diverge, while awaiting the complete segment results.

BANSEOG VIEW | Pricing power changes the value of skills

The striking feature of Samsung's quarter is that a single enterprise can benefit from memory inflation as a seller while confronting it as a buyer.

The capabilities that matter at those two positions are different. One side decides how to allocate scarce output; the other must deliver competitive products despite constraints.

For talent and product leaders, the starting question is not simply whether AI is booming. It is where their business sits in the chain of pricing power—and which decisions that position makes urgent.

BANSEOG VIEW

Banseog View — pricing power and capability allocation

KRW 107.4T is Samsung's consolidated guidance; KRW 111.9T for memory is an analyst estimate, not a segment disclosure.

The same AI-driven memory cycle can create pricing power for a supplier and cost pressure for a device manufacturer.

Capability priorities should follow a business unit's position in that chain, from capacity allocation to system optimization.

SOURCES

Primary sources and references

Group figures are Samsung's preliminary guidance; segment results, pricing and shipment figures are securities analysts' estimates. Forward-looking observations about product design and capability allocation are Banseog analysis, not confirmed changes in hiring or company policy.