South Korean Chipmakers Emerge in Ranking of Companies with Biggest Global AI Revenue Growth

Why It Matters

The rapid revenue expansion validates HBM as a strategic asset for AI compute, yet its high energy footprint forces chipmakers and customers to confront mounting carbon‑emission pressures. Understanding this trade‑off is essential for investors and policymakers shaping the future of AI infrastructure.

Key Takeaways

• SK Hynix AI revenue doubled to $70.4B by 2025.
• Samsung AI-linked sales grew 24% to $94.3B.
• HBM chips boost AI but raise semiconductor emissions fivefold.
• SK Hynix carbon intensity 246 tCO₂e per $M, far above Apple.
• US firms still dominate AI supply chain revenue growth.

Pulse Analysis

The AI boom has turned high‑bandwidth memory into a linchpin of modern compute. HBM chips, prized for moving massive data streams in nanoseconds, enable the training of large language models that power chatbots, recommendation engines and scientific simulations. SK Hynix’s revenue surge—from $34.5 billion in 2022 to an estimated $70.4 billion by 2025—reflects its aggressive capture of this niche, while Samsung leveraged its scale to lift AI‑linked sales to $94.3 billion. Investors see these figures as proof that memory manufacturers are no longer peripheral suppliers but core contributors to AI value chains.

However, the environmental cost of HBM production is stark. Manufacturing HBM consumes up to five times the energy per gigabyte compared with conventional DRAM, driving semiconductor‑related emissions toward a projected 247 million metric tonnes of CO₂e by 2030. Samsung’s 2024 Scope 1‑3 emissions hit 41 million tonnes, translating to a carbon intensity of 539 tCO₂e per $1 million revenue, while SK Hynix’s intensity sits at 246 tCO₂e per $1 million—both far above Apple’s 37 tCO₂e benchmark. This disparity forces industry players to weigh short‑term revenue gains against long‑term sustainability mandates and potential regulatory scrutiny.

Beyond Korea, the AI supply chain remains heavily U.S.-centric. Companies like CoreWeave, Nvidia and AMD dominate revenue growth, illustrating that while memory producers fuel the hardware stack, cloud and GPU leaders capture the lion’s share of AI spend. The concentration raises strategic questions about supply resilience and diversification. As hyperscalers spread orders across multiple chipmakers, Korean firms may need to balance rapid scaling with greener manufacturing processes to retain their competitive edge in a market increasingly sensitive to both performance and carbon footprints.