Effectively sold out through 2029—yet only 50%+ on signed LTAs?

At its October 28, 2026 earnings announcement, SK hynix said its DRAM and high-bandwidth memory (HBM) volume was effectively sold out through 2029. The outlook reflects customers trying to secure memory supply years ahead as AI infrastructure expands.

In the same announcement, SK hynix disclosed that signed long-term supply agreements (LTAs) account for more than 50% of its total production capacity for 2027 and 2028. Both statements describe the company's position, but they measure different things. The first is a demand outlook; the second refers to contracts already in place.

The difference is consequential for anyone trying to infer how much revenue is secured and how the company can plan its next generation of products.

The sellout outlook includes demand that is not yet contracted

SK hynix combines volumes already secured under LTAs with quantities being discussed in detail with customers to arrive at its effective sellout assessment. It considers demand under negotiation highly likely to materialize because customers have concrete purchasing plans.

That assessment is not equivalent to saying every shipment or selling price through 2029 is already covered by an executed agreement. Actual orders, shipments and pricing remain dependent on future events.

Nor can an observer simply subtract 50 from 100 to calculate uncommitted HBM volume. The sellout statement covers DRAM and HBM through 2029, whereas the more-than-50% LTA figure is stated against total production capacity for 2027–28. Their time periods and denominators are not identical.

Fourteen major customers, five-year terms—but not fixed prices forever

SK hynix has signed LTAs with 14 major customers, including leading cloud service providers. These agreements typically run for five years; SK hynix says extensions are being discussed with some customers.

Long-term supply does not mean a single fixed price for the entire term. The company says contract pricing reflects market conditions within agreed bands. The earnings-call transcript also refers to safeguards such as deposits intended to help turn customers' purchasing plans into actual orders and deliveries.

Such terms can reduce uncertainty on both sides without removing it completely. The customer gets more supply visibility; the manufacturer gets a firmer basis for planning production and investment. Neither side has eliminated all price and execution risk.

If demand is strong, why preserve capacity outside contracts?

SK hynix says it intends to maintain a similar proportion of long-term agreements while using capacity outside those contracts to address market conditions and demand for next-generation products.

The decision matters during a rapid HBM transition. The company expects HBM4 to overtake HBM3E in both shipment volume and revenue in the fourth quarter of 2026. Capacity tied too rigidly to existing product commitments could reduce the options available when a newer product becomes more valuable.

Uncontracted capacity should not automatically be equated with unwanted inventory. It can give the company room to change its product mix as technology and demand evolve. The actual allocation of such capacity by product, however, has not been made public.

A better demand picture supports investment; it does not guarantee returns

SK hynix expects capital spending to rise significantly in 2027 as Yongin Phase 1 ramps up and domestic and overseas infrastructure investment proceeds. A semiconductor fab requires a long sequence of site preparation, power and water supply, cleanroom buildout, equipment qualification and yield stabilization.

By the time a customer needs the chips, the investment decision may have been made years earlier. Long-term agreements can make this decision less speculative, but they cannot eliminate the risk of a slow manufacturing ramp, weaker demand or changes in product economics.

The company says it will adjust investment scale and timing according to customer needs, product competitiveness, profitability and capital efficiency. The operational challenge is to synchronize contracts, technology readiness and actual manufacturing capacity.

For semiconductor professionals, the real issue is which problem a job solves

The announcement does not establish that SK hynix will increase hiring by any particular amount. It does point to capability questions that become more important when multi-year supply commitments overlap with generational transitions.

HBM design, verification, advanced packaging, quality and yield work connects demanding customer requirements to manufacturable products. Production planning, process and equipment operations, and supply-chain functions then translate commitments into reliable deliveries.

New-product introduction creates a third challenge: keeping existing lines stable while HBM4 and subsequent generations scale. Rather than assuming every HBM-related job will grow, applicants should examine whether a specific role handles customer validation, yield, production reliability, scheduling or process transition. Live job requirements must be checked separately.

A visibility milestone, not guaranteed revenue through 2029

Negotiated demand can still change. Even executed LTAs may incorporate market-based prices, and future sales depend on customers' investment timing, manufacturing yields, production mix and successful delivery.

The disclosure does not provide enough information to calculate customer-specific contract values, future revenue by product or the exact share of HBM capacity that remains uncommitted.

What is established is SK hynix's own assessment that medium- and long-term demand visibility has improved. How that visibility translates into shipments, next-generation performance and profitability remains to be tested.

The balance between contractual certainty and manufacturing optionality

In a tight supply market, capacity itself is valuable. This announcement highlights a related but distinct capability: converting anticipated demand into credible commitments, and then aligning those commitments with manufacturing and investment schedules.

Locking up every unit of capacity could reduce flexibility to respond to the next product generation. Leaving too much supply uncommitted could weaken the basis for capital spending. The appropriate balance depends on products, customer terms and the pace of technological change.

The eye-catching number is 2029. The more lasting strategic question is how a chipmaker makes supply promises it can honor while retaining room to adapt when technology changes.

BANSEOG VIEW

BANSEOG VIEW | Demand visibility is valuable; so is the ability to change course

The 2029 sellout outlook includes executed LTAs and concrete demand still under negotiation. The 50%+ figure refers to signed agreements measured against 2027–28 total capacity. The two cannot be subtracted or treated as a single guaranteed revenue figure.

SK hynix's trade-off is how much demand to secure today while keeping enough flexibility for HBM4 and future generations. The outcome will be visible in actual deliveries, product transitions and returns, not the headline alone.

For professionals, this points to the importance of customer qualification, yield, manufacturing reliability and supply execution—not to a verified increase in job openings.

SOURCES

Primary sources and references

Prepared Oct 29, 2026 based on SK hynix's Oct 28 results release and call. 'Effectively sold out through 2029' combines executed LTAs and concrete DRAM/HBM demand still under negotiation. More than 50% refers to signed LTAs against 2027–28 total capacity, with a different time horizon and denominator. Fourteen customers, typical five-year terms, price bands, deposit safeguards and Q4 HBM4 forecast are company statements. Future hiring, customer-level revenue and exact product-level allocations have not been disclosed. Career implications are Banseog's analysis.