SK Hynix Q2 Earnings Preview & Explanations:Q1

SK Hynix Q2 Earnings Preview & Explanations:

Q1 -> Q2 profit walk based on things like: revenue accruals, wafer economics, utilization game theory, etc.

- Revenue: ₩82.5tn

- Operating profit: ₩63.0tn

- Margin: 76%

Just sharing my estimates based on my ongoing research and assumptions.

Q2 operating profit walk:

1. Q1 2026 operating profit reported: ₩37.6T

2. Pricing (DRAM blended ASP of plus low-40s%, NAND plus ~55%): +₩24.5T

3. Volume (DRAM bits +6%, NAND bits +17%): +₩3.4T

4. Cost-down & operating leverage: +₩1.8T

5. M15X early-ramp depreciation & start-up costs: -₩0.5T

6. FX (won appreciation reversing Q1 tailwind): -₩1.2T

7. Profit-share accrual: -₩2.4T

8. One-offs e.g. Nasdaq listing costs: -₩0.2T

= Q2E operating profit of ₩63.0T

My operating profit is slightly below ₩64.1tn consensus, yet above KIS Securities of ₩60.4tn.

Probably because of more conservative pricing numbers which I explain below.

-> Pricing: +₩24.5T

Really, this is where most estimates differ due to SK Hynix's DRAM book being repriced with the spot rate.

Essentially, their book is split in two parts where roughly half is under multi year LTAs and the other half rides against the spot rate, which ran up 50-60% for convential DRAM in Q2.

When you blend them, DRAM ASP comes to the low 40% range.

NAND is a little cleaner w/ eSSD mix-shift + QLC contract momentum putting blended ASP at roughtly +55%. No LTA drag on this side either vs. DRAM.

So just very conservatively assuming that volumes are flat, and then repricing the Q1 base (DRAM ~₩41.5tn, NAND ~₩10.5tn):

Means that pricing along contributes ~₩25tn of revenue at the low end.

Which flows down to OP at ~97% before any incentive lines kick in as price carries no wafer costs, only things like freight and royalty crumbs attach.

Meaning that price is essentially pure margin (as we've become accustomed to).

That's why pricing is probably way more important than volumes for memory companies right now.

On sensitivity analysis also:

Each "point" of blended DRAM ASP is worth ~₩0.42tn of operating profit.

Which is why I have a gap to consensus estimates.

The entire gap between me and consensus (₩2.1tn) is 4-5 points of DRAM ASP assumption.

So consensus at ₩64.1tn is a bet that the spot-exposed half of SK Hynix's book is materially larger than half, or that LTA resets were more material than +15-25%.

E.g. KIS Securities' model ₩60.4tn OP, which is essentially them using way more conservative pricing assumptions where they carry DRAM ASP nearer to +30%.

Which means that they assume that the LTA portion of the book is more dominant.

I believe that the real "truth" about the LTA structure, which SK hynix has never fully disclosed, gets revealed on Wednesday's earnings by the ASP numbers itself.

So this quarter is a bit of an experiment into my modelling that actually measures their contract business.

Regardless, it'll be fun to see how far off I was based on my limited info + static assumptions through Q2.