Earnings Outran the Stock: A 58% Operating Margin, a Record ₩47 Trillion Year, and a ₩14 Trillion Capital Return. Maintaining Outperform Into a Late-Cycle Peak

Results vs. Consensus

Q4 2025 Scorecard

FY2025 and Q4 Comparatives (K-IFRS)

Quality of Beat

Revenue: The +4.1% consensus beat and +34% sequential jump were broad and pricing-led. DRAM ASP rose ~20% QoQ on a conventional-DRAM price surge, NAND ASP rose nearly 30% QoQ, and high-density DDR5 module shipments grew ~50% QoQ. This is not a volume story (DRAM bits grew only low-single-digit, constrained by supply) but a price-and-mix story, the highest-quality way for memory revenue to grow.

Margins: The 58% operating margin is the headline and a genuine outlier. It reflects broad pricing strength across DRAM and NAND layered on a high-value mix (HBM, server DRAM, eSSD). It is also a flashing late-cycle indicator: memory operating margins in the high-50s sit near the top of what the industry has ever produced, which bounds the room for further margin expansion and makes the next leg dependent on volume.

EPS / net income: The +19% EPS beat is clean and arguably understated. The net line is suppressed by the ₩8.4tn exchangeable-bond derivative loss described above, a non-cash, share-price-driven item, so reported net income of ₩15.246tn (+21% QoQ) materially understates the operating step-up. Pre-tax income was ₩17.7tn. We would not read the modest net-income growth as any kind of deceleration.

Segment Performance

Product-Line Detail — Q4 2025

DRAM — Pricing-Led, Supply-Capped

DRAM bit shipments grew only low-single-digit because production is capacity-constrained, but ASP rose roughly 20% QoQ on a sharp conventional-DRAM price increase, and high-density DDR5 module shipments grew about 50% QoQ. The conventional DRAM tightness that emerged in Q3 became acute in Q4: with HBM and server demand consuming capacity, the conventional pool is short, and management says DDR5 is selling out as fast as it is produced. SK hynix entered full-scale 1c-nm DDR5 mass production and developed a 256GB DDR5 RDIMM on 1b-nm 32Gb die.

Assessment: A textbook supply-constrained pricing quarter. The fact that revenue grew 34% on low-single-digit bit growth tells the whole story: this is price, not volume. That is excellent for current margins and is also the configuration that, historically, eventually attracts the capacity that ends cycles, the out-year watch item.

NAND — Fully Recovered, eSSD Central

NAND shipments grew ~10% QoQ (above guidance) on mobile and eSSD, with ASP up nearly 30% QoQ as pricing accelerated. NAND set a record annual revenue, completing a full recovery from the soft first half. Management reframed NAND as moving from a peripheral storage device to a central element of the AI compute pipeline (KV-cache offload, RAG, high-IOPS enterprise SSD), and is developing next-generation 245TB ultra-high-density products.

Assessment: NAND has gone from the drag we flagged at Q2 to a genuine contributor, with both pricing and a structural AI narrative. We still treat it as the higher-beta leg, but the recovery is complete and the eSSD story is now corroborated by the numbers.

HBM — Doubled, and HBM4 in Volume

FY25 HBM revenue more than doubled YoY, the single biggest contributor to record DRAM revenue and profit. HBM4 secured mass-production readiness first in the industry (September 2025) and is now in large-scale production of customer-requested volumes, using proprietary MR-MUF packaging to target yield comparable to 12-high HBM3E. Management expects to retain overwhelming HBM4 share as it did with HBM3/HBM3E, while candidly acknowledging that because it cannot meet 100% of HBM demand, "some competition is expected to enter the market." Custom HBM (from HBM4E) is the next competitive frontier.

Assessment: HBM remains the franchise spine, and HBM4 is executing in volume on schedule. The notable nuance this quarter is management openly conceding that competitors will get some HBM4 share simply because demand exceeds SK hynix's capacity. That is a share-of-a-growing-pie dynamic rather than displacement, but it is the first explicit acknowledgment that the HBM monopoly loosens at the margin as the market scales, and it is worth monitoring.

Key Topics & Management Commentary

Overall management tone: Confident, structurally bullish, and now backing the words with capital. The defining actions this quarter were a record at a 58% margin, a ~₩14tn shareholder-return program, and the formation of a US "AI Co." investment vehicle, all signaling a company that believes the AI-memory regime is durable enough to both reinvest heavily and return cash. Management was also more willing to name the bounded risks: some HBM4 competition entering, PC/mobile price elasticity, and a potential US semiconductor tariff.

1. A 58% Margin and a Record ₩47 Trillion Year

"Fourth quarter operating profit reached KRW 19.2 trillion… with the operating margin of 58%… full year 2025 revenue reached KRW 97.1 trillion, and operating profit totaled KRW 47.2 trillion." — Song Hyun-jong, President & Head of Corporate Center

Q4 capped a year in which operating profit nearly doubled, at a quarterly margin (58%) and an annual margin (49%) that redefine memory profitability. Management framed this as strategic execution into an AI-focused demand environment, not merely favorable conditions.

Assessment: The scale is historic and the quality is high (price and mix, not volume). The same data point that anchors the bull case, a 58% margin, also anchors the discipline: margins this high are, by definition, late-cycle, and the asymmetry from here is bounded by how long peak pricing persists.

2. The Capital Return Arrives

"In addition to the fixed dividend, we will pay an additional cash dividend of KRW 1,500 per share… the company plans to retire… treasury shares… representing a value of approximately KRW 12.2 trillion… equivalent to 2.1% of total shares outstanding." — Song Hyun-jong, President & Head of Corporate Center

One quarter after telling the market that reinvestment was the best use of cash, management delivered a large return: ~₩2.1tn in FY25 dividends (₩3,000/share) plus a ~₩12.2tn treasury-share retirement worth 2.1% of shares. Management explained that the new 2025 policy always allowed for returns "even before the expiration of the policy period" once financial soundness was secured, which happened faster than expected.

Assessment: This is the exact development our Q3 recap anticipated ("we would expect buybacks/dividends to re-enter the conversation once the build is funded"). A ~₩14tn combined return, a net-cash balance sheet, and a stated willingness to do more is a material support under the stock and a direct relief of the Q3 capital-allocation concern. It meaningfully strengthens the total-return case.

3. 2026 Demand: DRAM >20%, NAND High-Teens, Server-Led

"Demand growth for DRAM and NAND in 2026 is expected to remain at over 20% and high teens %, respectively… demand for server DRAM and enterprise SSDs is expected to grow structurally at a pace well above the overall market growth." — Song Hyun-jong, President & Head of Corporate Center

Management held its accelerated 2026 demand framework (DRAM >20%, NAND high-teens), driven by the shift to inference, distributed architectures, and a structural step-up in server DRAM and eSSD content, even as PC and mobile see slower content growth on price.

Assessment: A server-led, structurally-above-market demand outlook for 2026, against contracted supply, is what underwrites the maintained rating. The honest caveat, addressed in topic 7, is that PC/mobile elasticity is now visible.

4. LTAs Harden Into Firm Multi-Year Commitments

"The LTAs being discussed today are expected to reflect strong mutual commitments between customers and suppliers, not simply indicating the intent to buy… customers now prefer multi-year contracts." — SK hynix management, Q4 2025 earnings call

Management contrasted today's long-term agreements with past ones: where prior LTAs were loose, fluid intent-to-buy arrangements, current LTAs reflect firm mutual commitments because memory now requires cutting-edge technology and far larger investment, compelling suppliers to demand demand-visibility and customers to commit multi-year.

Assessment: This is the structural-cycle thesis maturing from a Q3 assertion into Q4 substance. Firm, mutually-binding multi-year LTAs are precisely the mechanism that could dampen memory's historical volatility and justify a higher through-cycle multiple. Whether they hold through an eventual down-cycle remains the unproven crux, but the direction is unmistakable.

5. Inventory: Selling Out As Produced

"With memory being seen as the bottleneck in data center infrastructure expansion, server customers are expected to keep trying to increase purchase to secure their volume… With memory selling out as soon as it is produced, our inventory is projected to decline even further." — SK hynix management, Q4 2025 earnings call

Both customer and company inventories fell further in Q4. Server customers consume memory into set builds as soon as they secure it, company DRAM inventory declined QoQ, and NAND inventory weeks fell to nearly match DRAM. Management expects tight server-DRAM inventory to persist all year.

Assessment: Inventory selling out as produced is the strongest possible refutation of the pull-forward/inventory-glut bear case, and it has been consistent for three straight quarters. It also means there is essentially no buffer, which makes the system efficient on the way up and potentially abrupt if demand ever air-pockets.

6. 2026 CapEx Up, But CapEx/Sales Held at Mid-30s

"CapEx in 2026 is expected to increase significantly over year-over-year… we do not anticipate any difficulties in maintaining CapEx discipline at the mid-30% range… revenue [is expected] to grow substantially as well." — SK hynix management, Q4 2025 earnings call

2026 CapEx will rise considerably (capacity, tech migration, infrastructure), but management expects to hold CapEx-to-sales in the mid-30% range because revenue is also growing fast. M15X is ramping early; Yongin Phase 1, Cheongju M17, and the Indiana advanced-packaging plant are all progressing.

Assessment: Holding CapEx/sales flat in the mid-30s while absolute CapEx rises sharply is the right discipline metric, it ties investment to revenue rather than to optimism. The out-year risk remains that an industry-wide build, all anchored to today's demand, arrives into a 2027 that may look different. We keep this as the primary structural bear watch.

7. PC and Mobile Elasticity Appears

"Following the recent sharp rise in memory prices, some volume adjustments have appeared mainly among PC and mobile customers… revising their shipment plans or reviewing spec adjustments for price-sensitive, lower-tier products." — SK hynix management, Q4 2025 earnings call

Management acknowledged that surging memory prices are beginning to dampen PC and mobile demand, with some customers trimming shipments or downgrading specs on lower-tier products, while arguing on-device AI sustains high-end replacement and that AI features becoming default spec will structurally lift content over time.

Assessment: The first concrete evidence that memory pricing has reached demand-destruction levels in the consumer tier. Server demand more than offsets it today, but it is a genuine early-warning signal: if pricing stays this high and server demand ever pauses, the consumer offset disappears. A new, real item on the watch list.

8. The "AI Co." and US Investment

"We have decided to establish AI Co. for the purpose of proactively responding to the AI business environment… The AI Co. will be established in the U.S.… the investment commitment… is not large relative to our financial performance… not included in CapEx." — SK hynix management, Q4 2025 earnings call

SK hynix is forming a US-based "AI Co." investment vehicle to move beyond being a component supplier toward a full-stack AI data-center ecosystem partner, exploring companies with key AI capabilities. Management stressed the commitment is modest relative to cash generation and sits outside CapEx and FCF.

Assessment: A small, strategically-rational option on participating further up the AI stack, sized so it does not distort the capital-allocation framework. We view it as low-risk optionality rather than a thesis driver, and note management was careful to ring-fence it from CapEx discipline.

9. The US Tariff Threat

"The U.S. government recently talked of a 100% tariff on semiconductors if the fabs are not built in the U.S.… for now, we will monitor the discussions between the governments and communicate the company's direction at a later date." — SK hynix management, Q4 2025 earnings call

Asked about a mooted 100% US semiconductor tariff for non-US production, management declined to commit to additional US fabs, saying overseas fab decisions involve many factors and it will monitor government discussions.

Assessment: A genuine macro/policy tail risk, though SK hynix's product reaches the US largely embedded in customers' systems rather than as direct chip exports, which mutes the direct impact. The Indiana advanced-packaging plant provides some US footprint. We treat tariffs as a monitored macro risk, not a near-term earnings driver.

10. HBM4 in Volume, Competition Acknowledged

"After securing mass production readiness in September last year, we are currently in mass production of the volume requested by customers… Even as we maximize production, we cannot meet HBM demand 100%, so some competition is expected to enter the market." — SK hynix management, Q4 2025 earnings call

HBM4 is in large-scale production on the agreed customer timeline, using MR-MUF packaging to target HBM3E-comparable yield. Management expects to retain overwhelming share but, for the first time, openly acknowledged that demand exceeding its capacity will let some competition in.

Assessment: The candor is healthy and the framing is right, this is competitors taking a slice of a market growing faster than any one supplier can serve, not eroding SK hynix's lead. But it is the clearest signal yet that the HBM scarcity premium will be shared more as the market scales, which is a gradual headwind to monitor over multiple years.

Guidance & Outlook

SK hynix guides bit shipments and frames pricing qualitatively. For Q1 2026:

The flat-to-down Q1 bit guide is, once again, a supply story rather than a demand story: management says customer demand remains strong but production is capacity-constrained, and a seasonally weak Q1 plus tight supply caps shipment growth. The pricing environment is expected to stay favorable on persistent tightness. Product cadence into 2026: 1c-nm DRAM migration, CXL 2.0 and GDDR7, 321-layer NAND, next-gen 245TB eSSD, and HBM4 scaling. The forward earnings driver is now clearly volume and contracted pricing, not further margin expansion off a 58% base.

Analyst Q&A Highlights

HBM4 Status and Competitive Position

The opening question addressed recent "noise" on SK hynix's HBM4 progress and how it will defend competitiveness. Management reaffirmed mass production on the customer timeline and framed its lead as built on trust and execution, not just specs.

Q: "There has been some noise about SK hynix's progress on HBM4… can the company share your current status of the HBM4 development and the expected timing for mass production? And… plans to maintain HBM performance and mass production competitiveness?"
— Peter Lee, Citigroup

A: "Preparations for HBM4 are underway as planned… we are currently mass producing volumes requested by customers… Using our proprietary packaging technology, advanced MR-MUF, we plan to secure yield comparable to the 12-high HBM3E products. Even as we maximize production, we cannot meet HBM demand 100%, so some competition is expected to enter the market… our market leadership… will continue."
— SK hynix management

Assessment: HBM4 in volume at HBM3E-comparable yield is the key execution read, and it directly rebuts the "noise." The open acknowledgment that competitors take some share of unmet demand is realistic and not, in itself, a thesis risk, but it is the gradual loosening of the monopoly to watch.

How 2026 LTAs Differ From the Past

An analyst asked how the long-term agreements being struck for 2026 differ from prior LTAs. Management described a shift from loose intent-to-buy to firm mutual commitments.

Q: "Can the company share an update about the LTAs that are underway for the year 2026? And how are the LTAs different from past LTAs?"
— Ryu Hyung-kyun, Daishin Securities

A: "LTAs already existed, but they were generally loose contracts on volumes and tended to be quite fluid… The LTAs being discussed today are expected to reflect strong mutual commitments between customers and suppliers… customers now prefer multi-year contracts, but capacity constraints make it difficult to accommodate all customer requests."
— SK hynix management

Assessment: The hardening of LTAs into firm multi-year commitments is the most important structural development for the long-term multiple. If durable, it dampens the volatility that has always discounted memory. We weight it heavily, while reserving judgment until it survives a down-cycle.

Real Demand vs. Pull-In, and Inventory

An analyst probed whether the demand surge is real or inventory-driven pull-in, and asked for inventory levels. Management argued demand is real and inventories are draining.

Q: "Some suggest that it could be driven by pull-in demand to secure inventory, but most see this as growth in real demand due to tight customer inventory… what is the customer's inventory level… and the company's own inventory status?"
— Han Dong-hee, SK Securities

A: "Supply cannot keep pace with demand, resulting in a severe supply-demand imbalance… customer inventory levels have decreased overall… With memory selling out as soon as it is produced, our inventory is projected to decline even further… NAND inventory is also observed to be falling rapidly."
— SK hynix management

Assessment: Three consecutive quarters of draining inventory on both sides is strong evidence the demand is real, not channel-stuffing. It also leaves no buffer, which is bullish in an up-cycle and a source of fragility if demand ever turns.

Continuing Shareholder Returns and an ADR

An analyst asked whether the extra dividends and cancellations will continue and whether an ADR is planned. Management committed to ongoing review and confirmed it is studying corporate-value options.

Q: "Are there any plans to continue with the extra dividends and share cancellations going forward? And… are there plans to issue ADRs?"
— Kim Sun-woo, Meritz Securities

A: "Leveraging the financial room secured last year, we are implementing additional shareholder return… we plan to continue reviewing additional shareholder return measures and timing based on performance and cash flow… we are looking into various options to enhance corporate value. Nothing has been finalized to this date."
— SK hynix management

Assessment: A clear signal that returns will recur, and that an ADR (or similar corporate-value action) is under active study. A US listing would widen the investor base and could compress the Korea discount, a non-cyclical re-rating catalyst we will track.

The AI Storage Market and NAND Strategy

An analyst asked for the AI-storage outlook and SK hynix's NAND response. Management described NAND moving to the center of the AI compute pipeline.

Q: "What is the company's outlook on the AI storage market down the road? And… what is the company's plan for response to this demand in the NAND storage?"
— SK Kim, Daiwa Securities

A: "NAND today is changing completely… becoming a storage solution that directly supports AI computation workflows… whereas the SSD was a peripheral under the CPU centric architecture… it is increasingly becoming a central part in the compute pipeline… we are developing… ultra performance enterprise SSD… high IOPS."
— SK hynix management

Assessment: The structural NAND/eSSD narrative is now consistent across two quarters and backed by record NAND revenue. It elevates NAND's role in the thesis, though we still treat its pricing as higher-beta than DRAM/HBM.

2026 CapEx Magnitude and Discipline

An analyst asked how much 2026 CapEx will rise and whether CapEx/sales stays in the mid-30s. Management confirmed a significant increase while holding the ratio.

Q: "It was mentioned that the 2026 CapEx will increase significantly year-on-year. So by how much exactly? And… can the company maintain CapEx to revenue around the mid-30% level this year?"
— Ryu Young-ho, NH Investment & Securities

A: "CapEx in 2026 is expected to increase significantly… we will maintain CapEx discipline by monitoring market conditions and balancing demand visibility with investment efficiency… We do not anticipate any difficulties in maintaining CapEx discipline at the mid-30% range… the investment in the AI Company… is not included in CapEx."
— SK hynix management

Assessment: Tying CapEx to a mid-30s percentage of a fast-growing revenue base is the right discipline, but a large absolute CapEx increase across the whole industry is the mechanism that builds 2027 supply. This remains the central structural risk to the multi-year thesis.

PC/Mobile Price Sensitivity

An analyst asked whether skyrocketing DRAM prices are prompting PC/mobile customers to cut shipments or downgrade content. Management confirmed some adjustment but argued it will not broaden.

Q: "If the prices continue to rise, then the cost burden on the PC and mobile customers could also grow… are there any requests or demands to adjust the set shipment or to downgrade the content?"
— Kim Rok-ho, Hana Securities

A: "Some volume adjustments have appeared mainly among PC and mobile customers… Some customers are becoming more conservative, revising their shipment plans or reviewing spec adjustments for price-sensitive, lower-tier products… it is not likely that the impact… will spill over into a broader demand contraction."
— SK hynix management

Assessment: The single most important caution in the call. Management is conceding real price elasticity in the consumer tier, offset for now by server strength. It is the early edge of the demand-destruction risk that a high-priced memory cycle eventually confronts, and the clearest reason we frame the rating as late-cycle.

Market Reaction

All price levels are FMP-verified KRX closing prices; the benchmark is the KOSPI (^KS11). SK hynix discloses before the Korean market open, so the reaction is the print-day session.

• Pre-print close (2026-01-27): ₩800,000, an all-time high.
• Reaction (2026-01-28) close: ₩841,000, +5.1% on the day, intraday high ₩854,000 (+6.7%), while the KOSPI rose +1.7%.
• Volume: ~6.7M shares versus a ~3.8M 30-day average (1.8x), heavy, on a clear upside reaction.
• Setup: +262% over the trailing twelve months and +25% in the prior 30 days, with only +22.9% YTD (the run is now being measured off a much higher base).

The reaction is the most important behavioral shift since we initiated. After selling off 2.6% on the Q3 record, the stock rallied 5.1% on this one, outperforming a rising KOSPI. The difference is that Q4 paired the record with two things the market had been waiting for: a re-acceleration of earnings (operating profit +68% QoQ, faster than the share-price run) and a concrete ~₩14tn capital return. When good news starts working again after a sell-the-news quarter, it usually means the prior profit-taking cleared the froth and the marginal buyer re-engaged on improved fundamentals.

For our rating, this relieves the central tension of the Q3 note. The "the stock has run too far" worry is harder to make when earnings are outrunning the stock and the company is returning ₩14tn. We still treat the absolute price level and the 58% margin as late-cycle signals, but the market's positive verdict on a re-accelerating, cash-returning quarter is consistent with maintaining Outperform.

Street Perspective

Debate: Is a 58% Margin a New Normal or a Cycle Peak?

Bull view: The margin reflects structural HBM/AI mix and firm multi-year LTAs that lock in pricing; with 2026 sold out and supply constrained, the high-50s margin is sustainable longer than skeptics think. Conventional DRAM tightness is structural because HBM consumes capacity.

Bear view: Memory operating margins in the high-50s have always been cycle peaks. The next move in margins is down, not up, and the operating leverage that produced 58% on the way up works in reverse violently when pricing rolls. The build that the whole industry is funding will eventually supply the demand.

Our take: Closer to peak than to new-normal, but with a longer plateau than prior cycles. The LTA hardening and HBM capacity constraint genuinely extend the high-margin period, but we do not underwrite 58% as a through-cycle figure. The rating rests on the contracted near-term earnings, not on the margin staying here.

Debate: Does the Capital Return Change the Investment Case?

Bull view: A net-cash balance sheet, a ~₩14tn return, and a commitment to more (plus a possible ADR) transform SK hynix from a leveraged cyclical into a cash-returning compounder, deserving a higher multiple. The return also signals management confidence in the durability of the cash flows.

Bear view: Returning cash at a cycle peak, when the stock has tripled, is poor timing; the cash may be needed when the cycle turns and CapEx must continue. The buyback-equivalent (cancellation) is small at 2.1% against the share-price run.

Our take: Net positive. The return is modest relative to cash generation, the balance sheet is net cash, and the signal value (returns recurring, ADR under study) supports a re-rating of the through-cycle multiple. It does not, by itself, justify chasing the stock, but it improves the total-return profile and relieves our Q3 capital-allocation concern.

Debate: How Worrying Is PC/Mobile Elasticity?

Bull view: Consumer softness is fully offset by server strength, on-device AI sustains high-end replacement, and AI features becoming default spec structurally lifts content. The mix shift toward server is exactly what SK hynix wants.

Bear view: Price-driven demand destruction in PC/mobile is the first crack, and it always starts at the margin. If server demand ever pauses while consumer stays elastic, the offset vanishes and high prices accelerate the downturn.

Our take: A real early-warning signal, not yet a thesis risk. Server demand dominates and is contracted, so the consumer softness is immaterial to 2026 earnings. But it is the leading edge of the elasticity that ends every high-priced memory cycle, and we elevate it to a primary watch item for the back half of 2026.

Thesis Scorecard: 2025_Q3 Signposts Revisited

Our Q3 2025 update set signposts for Q4. They graded uniformly bullish, including the capital return we explicitly predicted.

Scorecard summary: All seven signposts bullish, three strongly so, with zero misses. The capital-return signpost, which we framed at Q3 as a prediction, paid off directly. The one item carrying an asterisk is the out-year CapEx supply risk. This is the second consecutive clean-sweep scorecard, the difference at Q4 being that the new caution (PC/mobile elasticity, peak margin) is about the cycle's maturity, not about execution.

Bottom Line: Maintaining Outperform, Late-Cycle

Rating decision: We maintain Outperform, the third consecutive in our coverage. Q4 confirmed every Q3 signpost and added the capital return we anticipated: a record 58% operating margin, a historic ₩47tn FY25, HBM4 in volume, a net-cash balance sheet, a ~₩14tn shareholder return, and 2026 demand sold out against firm multi-year LTAs. Critically, earnings re-accelerated faster than the share price, and the market rewarded the print with a 5.1% rally, relieving the asymmetry concern we raised at Q3.

The honest evolution in our view is that the rating is now explicitly late-cycle. A 58% operating margin is near the top of what memory has ever produced; the room for further margin expansion is small, so the forward return must come from volume and contracted pricing rather than another leg of margin. We remain Outperform because the near-term earnings are contracted, the balance sheet is returning cash, and the demand outlook is server-led and above-market. But we would keep trimming into strength, and we are watching the first cracks, PC/mobile price elasticity and the slow loosening of the HBM monopoly, that typically precede a downgrade.

What would move us to Hold: the operating margin rolling from its 58% peak; a stall or competitive-share loss in HBM4; PC/mobile elasticity broadening into a server pause; or the multiple running ahead of the now-near-peak earnings.

What would move us to Underperform: evidence the 2026 sold-out framework is cracking (LTA renegotiation/cancellation); clear signs the industry CapEx wave is building 2027 oversupply faster than demand absorbs it; or a Samsung/Micron HBM4 breakthrough that materially contests the scarcity premium.

Signposts for Q1 2026 earnings (late April 2026):