NHN Q2 2026 slides: AI GPU drives record profit, margins expand

NHN reported Q2 2026 results at record levels, with operating income up 164.1% YoY to KRW 57.9B and operating margin expanding to 7.6% (+4.0pp YoY). Shares jumped 15.57% to $48,250 on the announcement date (later easing to ~$47,550) as technology led the turnaround: technology revenue rose 52.9% YoY, with NHN Cloud up 85% YoY, boosted by AI GPU-related B200 chip revenue. Management guided for continued profit momentum in 2H26 and outlined plans to secure at least 30MW of additional AI data center capacity by end-2027, supporting the market’s optimism despite a sequential net income decline.

Analysis

The investable read-through is less about NHN itself and more about what kind of AI spending is proving monetizable: not just model training, but contracted GPU hosting, power delivery, and orchestration at the edge of cloud ecosystems. That is structurally bullish for NVIDIA (NVDA) and, to a lesser extent, for regulated power/load beneficiaries such as KEP if incremental megawatts translate into allowed tariff recovery. AMZN gets only a modest second-order benefit via AWS resale/channel activity; it is a demand signal, not an earnings step-up.

The key risk is that the quarter may be top-heavy on scarce GPU availability and accounting mix, which can fade fast if supply normalizes or if utilization lags the new capacity build. Over the next 1-3 quarters, the market should focus on whether AI infrastructure revenue converts into durable backlog and whether electricity/depreciation drags offset the gross profit gains. If management slips on additional MW deployment or the next two quarters show decelerating cloud growth, the current rerating can unwind quickly.

The consensus likely underweights capex intensity: AI hosting looks high-growth, but it also pulls forward depreciation, power expense, and working capital. That means the right framing is not "new growth engine" but "higher-quality revenue only if utilization stays above threshold." A 6-18 month bull case requires signed long-duration contracts and evidence that incremental AI capacity earns returns above cost of capital; otherwise this is just a cyclical spike with a lower-quality earnings base.

Key Decisions for Investors

• Long NVDA on any post-rally consolidation: use this as a read-through for incremental B200/B300 demand from non-hyperscale customers; risk/reward improves if Nvidia capex commentary stays firm over the next 1-2 quarters.
• Avoid chasing NHN after the gap-up; wait for either a pullback or proof that AI GPU revenue is contracted, not spot-driven. Falsify the bullish case if next-quarter cloud growth drops below ~30% y/y or margins slip back under 6%.
• Watch KEP as a 6-18 month beneficiary of Korean data-center load growth, but only add on evidence of tariff pass-through or signed power contracts. Regulatory intervention is the main downside risk and can cap the rerating quickly.
• Small tactical long AMZN only if you want a weak but positive cloud-demand read-through; treat this as a low-conviction basket add, not a standalone catalyst, and exit if AWS capex growth or cloud margins soften.