South Korean tech giants are among the top firms globally that generated the most revenue growth from the artificial intelligence (AI) boom last year, found a new report.
To continue reading, subscribe to Eco‑Business.
There's something for everyone. We offer a range of subscription plans.
- Access our stories and receive our Insights Weekly newsletter with the free EB Member plan.
- Unlock unlimited access to our content and archive with EB Circle.
- Publish your content with EB Premium.
Tech giant Samsung Electronics and semiconductor firm SK Hynix entered the ranking of 20 publicly listed firms that posted the fastest AI-related revenue growth across the supply chain, surveyed by investing research platform BestBrokers.
SK Hynix, a leading South Korean memory chip producer, more than doubled its AI-linked revenue as it rode a wave of demand for high-bandwidth memory (HBM), a type of ultra-fast computer chip that can move massive amounts of data in a very short time and considered one of the tightest bottlenecks in scaling advanced AI systems, according to the research.
Its revenue expanded from US$34.5 billion in 2022 to US$70.4 billion in 2025, reflecting a compound annual growth rate (CAGR) of 26.84 per cent.
That growth underscores the central role of HBM chips in training and deploying large language models and other compute-intensive AI applications, as hyperscalers and chip designers scramble for reliable supply, suggested the report.
Image: BestBrokers
Samsung Electronics, the world’s largest memory manufacturer, also recorded substantial gains in AI-related sales, though from a much larger revenue figure. The conglomerate’s AI-linked revenue rose from US$76.2 billion to US$94.3 billion in the same period, translating to a CAGR of 7.36 per cent.
However, HBM chips — the very product powering SK Hynix and Samsung’s AI revenue surge — can consume up to five times more energy per gigabyte during production than standard memory, according to research site Silicon Analysts.
Semiconductor manufacturing emissions are projected to rise by roughly one-third, reaching 247 million metric tonnes of carbon dioxide equivalent by 2030, driven largely by surging demand for HBM chips, based on data from Canada-headquartered firm TechInsights Inc.
Samsung registered 41 million tonnes of carbon dioxide equivalent in Scope 1 to 3 emissions in 2024 — the highest among seven major global technology companies — with a carbon intensity of 539 tonnes of carbon dioxide equivalent (tCO₂e) per US$ million in revenue, based on a report by the Institute for Energy Economics and Financial Analysis (IEEFA) released in December. SK Hynix, recorded a carbon intensity of around 246 tCO2e per US$ million in revenue.
For context, Apple’s carbon intensity is just 37 tCO₂e per USD million. SK Hynix recorded 246 tCO₂e per USD million, also significantly above global peers, according to IEEFA.
US tech giants dominate AI supply chain
The BestBokers report groups Samsung and SK Hynix alongside major United States and global firms involved in AI chips, cloud infrastructure and software platforms, painting a picture of an AI supply chain increasingly dominated by a handful of large incumbents.
Across the entire AI supply chain, American AI cloud-computing company CoreWeave ranks first in AI revenue. The Texas-based graphics processing unit cloud operator went from US$15 million in revenue at the end of 2022 to US$5.13 billion in 2025, a CAGR of 586.88 per cent.
Traditional cloud providers could not absorb graphics processing unit demand fast enough, and CoreWeave captured the overflow, effectively building a multi-billion-dollar infrastructure business in the time it takes most startups to find product-market fit, said the study.
Meanwhile, California-headquartered tech company Nvidia also leads in AI computing, with its data centre revenue surging from US$14.6 billion in 2022 to US$167.9 billion in 2025 because it dominates the market for the specialised chips used to train AI models.
Fellow US chipmaker AMD has risen as a serious rival, with its data centre business growing from US$6.1 billion to US$16.6 billion, as major tech giants deliberately spread their chip orders across multiple suppliers to avoid over-reliance on any one company.
Marvell Technology, another US firm, grew from US$2.5 billion to US$5.8 billion by making the specialised “glue” chips that connect large groups of AI processors together so they can work as one.
Broadcom grew more slowly, reflecting its broader, more diversified chip business, while Dutch firm ASML, which makes the highly specialised machines used to manufacture all these chips, also posted steady growth, underscoring that the entire AI chip industry depends on its equipment to function.
