J.P.Morgan reviews coverage of Korean autos, batteries, and nuclear EPC, with HMG robotics and ADAS computing power becoming key valuation upside focuses

Overview

This is a J.P.Morgan industry coverage summary on Korean autos, batteries, and nuclear EPC published in June 2026. The cover page lists the research scope as Korea Autos, Batteries, and Utilities, with Sonny Lee as the lead author. The report covers multiple companies, including Hyundai Motor Company, Kia Corp, Hyundai Autoever, Hyundai Mobis, HL Mando, Samsung SDI, L&F, Ecopro BM, POSCO Future M, POSCO, Doosan Enerbility, Hyundai E&C, KEPCO E&C, and KEPCO. The visible text mainly focuses on valuation methodology, global vehicle demand, HMG's robotics business, the NVIDIA/Google DeepMind ecosystem, ADAS training compute, automotive software/hardware architecture, and nuclear timelines.

Core views

The report's core view is that traditional valuation of the Korean auto sector is being augmented by the option value of ADAS and robotics businesses, and SOTP valuations for HMG-related names incorporate equity value from the robotics business. HMG's competitiveness in robotics and autonomous driving depends not only on vehicle and manufacturing capabilities, but also on technology stacks such as NVIDIA's compute ecosystem, Google DeepMind-like model capabilities, world models, VLA, and real-time motion control. Ratings across the battery and materials chain are differentiated, indicating that opportunities are not uniformly attractive across the industry. The nuclear EPC and utilities sectors are also included in coverage, though the available text contains relatively few details on nuclear projects.

Analysis framework

The report adopts an industry coverage summary framework: it first presents covered companies, ratings, prices, and valuation methods; then expands on global vehicle demand and HMG's technology roadmap; next breaks down robotics task workflows, the NVIDIA ADAS/robotics ecosystem, and training compute requirements; and finally supplements company-level risk disclosures with nuclear timelines and disclosure information.

Methodology notes

• Valuation methodsSOTP

  Sum-of-the-parts valuation

  Sample target prices are based on SOTP, valuing the core auto business at 2027E P/E and adding equity value from HMG's robotics business, where robotics value uses the midpoint of bull/bear analysis and applies a 30% holding-company discount.

• Valuation methodsP/E relative valuation

  2027E P/E multiple

  The text mentions that the auto business is valued at 11x or 9x 2027E P/E, with reference to non-traditional OEMs with ADAS capabilities.

• Technology breakdownRobotics task chain

  From language input to action execution

  Using the example of moving a mouse from the left side of a keyboard to the right side, the report breaks down a robotics task into goal recognition, spatial relationship understanding, world perception, task sequence planning, motion candidate generation, motion planning, MPC real-time fine-tuning, and actuator control.

• Technology ecosystemNVIDIA ADAS/robotics ecosystem

  Omniverse, Cosmos, Alphamayo, GR00T N1, Isaac Sim, Jetson

  The report explains NVIDIA's ecosystem as a software-hardware stack spanning digital-twin simulation, world-model inference, ADAS decision-making, foundational robotics policy models, robotics simulation, and edge AI hardware.

• Compute estimationTraining compute estimation

  tokens, model parameters, training FLOPs multiplier, and driving mileage

  Using 288,000 tokens, 10bn parameters, an approximately 6x training FLOPs multiplier, a 25 mph average speed, and 4.5bn miles of data, the report estimates the total compute and time required to process Tesla's cumulative driving data.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• Hyundai Motor Company (005380.KS)

  HMG's core vehicle manufacturer with ADAS/robotics exposure; listed at W503,000 with an OW rating in the coverage list.

  Strengths

  SOTP valuation can incorporate P/E for the core auto business and HMG robotics equity value, while ADAS capability can improve the valuation narrative versus non-traditional OEMs.

  Weaknesses

  Auto sales remain affected by macro uncertainty, and the value of robotics and ADAS requires delivery of technology milestones.

  Comparison

  The report uses non-traditional OEMs with ADAS capabilities as one of the P/E reference groups.

  Risks

  Declining auto sales, delays in ADAS development, and setbacks to HMG robotics development and manufacturing plans.

• Kia Corp (000270.KS)

  An HMG vehicle name, listed at W138,900 with an OW rating in the coverage list.

  Strengths

  Benefits from HMG's vehicle platform, ADAS capabilities, and potential equity value from the robotics business.

  Weaknesses

  Valuation upside likewise depends on progress in ADAS and robotics, while the traditional vehicle demand cycle remains a constraint.

  Comparison

  Like Hyundai Motor Company, it has HMG-related vehicle exposure, but the target-price and equity-value samples use different P/E and robotics equity assumptions.

  Risks

  Weaker macro demand, delays in ADAS milestones, and slower-than-expected commercialization of the robotics business.

• Hyundai Autoever, Hyundai Mobis, HL Mando

  Names related to automotive software, components, ADAS, and hardware ecosystems, with coverage-list ratings of N, N, and UW respectively.

  Strengths

  The report discusses layered software, hardware, and middleware, and these companies may benefit from in-vehicle software, ECUs, ADAS, and platform upgrades.

  Weaknesses

  Ratings are weaker than for the core OW vehicle names, indicating differences in earnings, valuation, or competitive positioning.

  Comparison

  The report links the division of labor among software, hardware, and middleware to intelligent vehicle architecture rather than evaluating them solely under a traditional auto-parts framework.

  Risks

  High software-hardware integration complexity; delays in ADAS development or insufficient progress in middleware platforms could impair value realization.

• Samsung SDI, L&F, Ecopro BM, POSCO Future M, POSCO

  Korean battery and materials chain names, with ratings ranging from OW to N to UW.

  Strengths

  Samsung SDI and L&F are rated OW in the coverage list, indicating that some battery-chain assets still carry positive investment views.

  Weaknesses

  Ecopro BM is rated N, POSCO Future M is UW, and POSCO is N, showing that opportunities across the battery materials chain are not uniformly attractive.

  Comparison

  Ratings diverge within the same battery industry chain, requiring company-by-company assessment based on fundamentals and valuation.

  Risks

  Changes in new energy vehicle demand, materials prices, capacity utilization, and company earnings forecasts may lead to rating and valuation volatility.

• Doosan Enerbility, Hyundai E&C, KEPCO E&C, KEPCO

  Names related to nuclear EPC, utilities, and power engineering; the first three are OW, while KEPCO is UW.

  Strengths

  Ratings for nuclear EPC-related companies are generally positive in the coverage list, and the report includes a 'Nuclear: Estimated timeline' section.

  Weaknesses

  The available text provides insufficient detail on nuclear project progress, orders, and earnings, and KEPCO is rated UW.

  Comparison

  EPC/equipment-related names and utility companies are rated in different directions, suggesting different investment logic between the nuclear construction chain and power companies themselves.

  Risks

  Nuclear project timelines, regulatory approvals, construction progress, cost control, and changes in utility policy.

Key data

• Report titleKorea Autos, Batteries, and Nuclear EPC Coverage SummaryBoth the file name and body title point to a coverage summary of Korean autos, batteries, and nuclear EPC.
• Research institutionJ.P.MorganThe cover page lists J.P.Morgan Securities (Far East) Limited, Seoul Branch.
• Lead analystSonny LeeThe cover page lists contact information +82-2-758-5716 and sonny.lee@jpmorgan.com.
• Price reference date2026-06-25 closeThe disclosure page states that all prices in this report are as of the close on June 25, 2026 unless otherwise specified.
• Hyundai Motor Company005380.KS / W503,000 / OWListed in the covered companies list.
• Kia Corp000270.KS / W138,900 / OWListed in the covered companies list.
• Hyundai Autoever307950.KS / W540,000 / NListed in the covered companies list.
• Hyundai Mobis012330.KS / W512,000 / NListed in the covered companies list.
• HL Mando204320.KS / W48,800 / UWListed in the covered companies list.
• Battery and materials chain sampleSamsung SDI 006400.KS/W481,500/OW; L&F 066970.KQ/W107,500/OW; Ecopro BM 247540.KQ/W144,600/N; POSCO Future M 003670.KS/W170,900/UWListed in the covered companies list, showing rating differentiation.
• Nuclear EPC and utilities sampleDoosan Enerbility 034020.KS/W88,500/OW; Hyundai E&C 000720.KS/W111,900/OW; KEPCO E&C 052690.KS/W106,400/OW; KEPCO 015760.KS/W38,850/UWListed in the covered companies list; EPC-related names are rated more positively, while KEPCO is UW.
• Target price samplesDec-26 PT W670,000; Dec-26 PT W240,000The text states both are based on SOTP, using 11x and 9x 2027E P/E respectively, and including equity value from HMG's robotics business.
• Sample robotics equity value~W14tn (28% equity); ~W8tn (17% equity)The text says these values reflect the midpoint of HMG robotics business bull/bear analysis and apply a 30% holding-company discount.
• NVIDIA Blackwell GPU estimate50,000 units; about 100 EFLOPS processing capacity frameworkThe report uses this compute power to estimate the time needed to process 4.5bn miles of driving data.
• Tesla cumulative driving data4.5bn milesUsed as a reference for ADAS training data processing volume.
• Training compute assumptionsabout 17.3 PFLOPs/sec; about 2.49 EFLOPs per mileBased on 288,000 tokens, 10bn parameters, an approximately 6x training FLOPs multiplier, and a 25 mph average speed.
• Estimated processing time1,296 days, about 3.6 yearsThe report derives this duration by dividing 11,200 YFLOPs by 100 EFLOPS.

Impact & implications

For investors, this report shifts the comparative focus for Korean autos, batteries, and nuclear EPC from pure sales volumes and financial forecasts to technology platform capabilities. Valuation upside for HMG-related auto names comes from the combination of ADAS, robotics, and large-scale training compute, but realization depends on development milestones and manufacturing execution. Differentiated ratings across the battery materials chain suggest large differences in fundamentals and valuation appeal within the same industry chain. Although many nuclear EPC-related companies in the coverage list are rated OW, the available text does not provide enough project detail, so further verification against the full report and company models is needed.

Risks

• Heightened macro uncertainty may put downward pressure on auto sales volumes.
• Key development milestones for ADAS systems may be delayed.
• HMG's robotics business development and manufacturing plans may encounter setbacks.
• Differentiated ratings among battery and materials chain companies mean changes in demand, pricing, and earnings forecasts could amplify share-price volatility.
• The investment thesis for nuclear EPC-related names depends on project timelines, approvals, construction progress, and cost control.

What to watch

• Whether the outlook for global vehicle demand improves or deteriorates.
• Progress in robotics and ADAS collaboration involving HMG, NVIDIA, and Google DeepMind.
• Whether investment in training compute using 50,000 NVIDIA Blackwell GPUs translates into ADAS data processing and model capabilities.
• The speed of HMG robotics business execution in language understanding, world perception, task planning, motion planning, and action execution.
• Changes in Dec-26 target-price assumptions and 2027E P/E multiples for core HMG names such as Hyundai Motor Company and Kia Corp.
• Whether rating divergence among Korean battery-chain companies widens or narrows.
• Timelines, orders, and execution progress for nuclear EPC projects.