SK Hynix Gets Its First Class A Rating. But the Q2 Earnings Miss Tells the Real Story.

Moody's upgraded SK Hynix to A3 on August 3, 2026 - the company's first Class A rating among the three major agencies since joining SK Group in 2012. S&P had already moved to BBB+ in February. Fitch followed in April. The credit market just delivered a clean sweep.

But on July 28, a week before Moody's acted, SK Hynix shares fell 10% after Q2 earnings missed consensus.

The two events are about the same company, but they are not telling the same story. One is a backward-looking verdict on financial structure. The other is the market asking whether the AI-driven profit engine can actually keep running.

Decomposing the Q2 number shows why both can be right at the same time - and where the real risk sits.

Decomposition: Record Net Profit vs. Missed Operating Profit

SK Hynix reported Q2 2026 revenue of 79.3 trillion won and a net profit of 93.9 trillion won - more than 13-fold year-over-year growth, a figure that exceeds the quarter's revenue. On the surface, the company made more in profit than it brought in as sales.

But the net profit number is not operating earnings. Of the 93.9 trillion won, approximately 63.3 trillion won - roughly 67% - came from one-time gains on the sale of its Kioxia stake, a position the company had held since 2018. Strip that out, and the underlying net profit is roughly 30.6 trillion won. That is still a strong number, but it is not the 13-fold headline.

More telling is the operating profit. SK Hynix posted 60.5 trillion won - a record, up 557% year-over-year. But analysts expected 64 trillion won. Revenue of 79.3 trillion won also missed the 84 trillion won consensus. The stock sold off on the miss.

That is the gap: a credit agency is upgrading on structural balance sheet improvement, while equity investors are downgrading on an earnings miss. The operating margin of 76% is extraordinary by any industry standard, but the miss on top-line revenue and the shortfall on operating profit tell you that not everything in the supply chain went as smoothly as the HBM narrative assumes. HBM4 shipments came in below expectations, pushing some revenue recognition into later periods, according to analysts cited by Reuters.

The Credit Story: Why the Rating Upgrade Is About Balance Sheet, Not Narrative

Moody's projected SK Hynix's adjusted EBITDA - earnings before interest, taxes, depreciation, and amortization, a proxy for cash-generation capacity - at $192.8 billion for 2026, up from approximately $45.7 billion last year. Adjusted net cash surged from about $7 billion at the end of last year to approximately $47 billion by mid-2026. The company's net cash position on the Korean balance sheet was 69.4 trillion won (88 trillion won in cash less 18.6 trillion won in total debt).

That is the number structure Moody's is pricing. The company has transformed from a leveraged memory player into a net-cash entity with EBITDA projected to quadruple. A higher rating lowers the cost of bonds SK Hynix issues to fund its capex program - critical when the company plans to raise capital spending to the high-40 trillion won range in 2026, up from 30.2 trillion won in 2025. It also widens the pool of eligible institutional buyers, including pension funds and sovereign wealth funds with mandates restricted to upper-medium-grade debt.

But the rating also contains its own risk disclosure. Moody's flagged three persistent threats: memory market volatility, competition from Chinese memory manufacturers, and the burden of continuous large-scale capital expenditure.

Supply Chain Position: Dominance, But With Expiring Moats

SK Hynix controls approximately 56% to 62% of the HBM market, by various analyst estimates. It secured over two-thirds of HBM supply orders for Nvidia's next-generation Vera Rubin platform. The company began HBM4 mass shipments in Q2 and has finalized long-term agreements with around 10 major customers, many spanning five years and backed by deposits to ensure execution.

That is a structural chokepoint. Nvidia designs its AI accelerators around whatever memory is ready first, and SK Hynix has consistently been first across HBM generations. The June 2026 technology partnership with Nvidia to align roadmaps for years ahead is the kind of lock-in that matters.

But the moat is not static. Samsung has been making material progress on HBM4 and is expected to lift its HBM market share above 30% in the coming year, up from around 22% in late 2025. Samsung's broader business diluted its 2025 operating profit to 43.6 trillion won, below SK Hynix's 47.2 trillion won - but Samsung's memory segment alone generated 24.9 trillion won, and its pricing power in conventional DRAM has been more aggressive than SK Hynix's. Micron also overtook Samsung in third-place HBM share in the second half of 2025.

SK Hynix's long-term contracts reduce commodity-cycle exposure, which is the company's explicit goal. But they also temper near-term pricing gains. When demand is strong, long-term agreements leave money on the table relative to a spot-pricing strategy. The Q2 miss itself partly reflects this trade-off.

What to Watch

• Shareholder return policy timing and structure. SK Hynix has not yet disclosed details and plans to announce later in 2026. With net cash approaching 70 trillion won and investors pressing for capital deployment, the size, form (buybacks vs. dividends), and schedule of this plan will determine whether equity holders accept the current valuation. This is the single most actionable signal.
• HBM4 ramp pace in Q3 and Q4. Revenue recognition delays in Q2 suggest the HBM4 production curve is steeper than initially priced. Whether those volumes materialize in the second half determines whether the earnings miss was a timing issue or a demand softening.
• Samsung's HBM4 qualification progress. Samsung has said customers praised its HBM4 competitiveness. If Samsung accelerates its Nvidia qualification timeline, SK Hynix's pricing power and margin trajectory will face structural pressure.
• Capital expenditure execution relative to the 40 trillion won ceiling. The company plans to accelerate its M15X facility and open Yongin Phase 1 in early 2027. Whether capex stays within the stated range, or whether AI infrastructure capex inflation pushes it higher, changes the debt trajectory and the sustainability of the net-cash position.
• Moody's language on Chinese competition. The agency flagged this risk but did not quantify it. Any material expansion of Chinese HBM capability - particularly from CXMT, which has been advancing its HBM program - would change the competitive math across the entire supply chain.

The rating upgrade is real. The balance sheet transformation is real. The HBM dominance is real. The earnings miss and the 10% stock drop are also real. The question is not whether SK Hynix has improved. It is whether the improvement justifies the amount of AI future the market has already asked the stock to deliver.