• Samsung’s 2024 earnings were positive with robust growth compared to a tepid 2023. However, it is still far from peak performance.
• Samsung’s growth has been limited considering its Device Solutions (Memory and Foundry) missed the AI boom, a potential $30 billion-$45 billion in lost opportunity between 2022-2025 to competitors such as SK Hynix, TSMC in the AI servers & Qualcomm, Micron in the AI edge.
• Samsung still has strong fundamentals and scale. It needs to go back to drawing board and a turnaround is possible later this year, making sure the next AI wave is not missed.

Samsung, the world’s largest smartphone, consumer electronics, semiconductor solutions, and OLED display supplier reported its Q4 2024 and full year 2024 earnings last week.

2024 was a year for recovery and growth for those at the centre of the AI boom. This was in contrast to the tepid 2023 which caused the entire industry to decelerate. In 2024, Samsung benefitted a bit due to its broader business segments but overall, the company struggled to cash in on the first AI growth wave.

Rivals TSMC in foundry, SK Hynix and Micron in memory, NVIDIA a potential customer, and Apple in smartphones did exceedingly well, capturing more value.

> CY2024 growth looked good because of 2023 downturn but nowhere close to the highs of 2017-18 when the smart devices market peaked and 2021-22 was lifted by the post-pandemic growth spurt.

Source: Counterpoint Research

Memory Business: A “Bit” Volatile!

> Over the last decade, Samsung has been a tale of two segments – Device Solutions (DS) – Memory & LSI) and Mobile & Networks (MX) business contributing to average 65%-75% of the total revenues and almost 80%-95% of total operating profits.

> While MX (smartphones) is a stable business with minor ups and downs, the real cash cow has been the Memory business.

> There is a strong co-relation between Memory earnings and Samsung’s overall earnings. In many high quarters, Samsung’s Memory division contributed to anywhere from half of the brand’s overall operating profits. Other times, the division has also weighed on the company’s performance, driving negative operating margin during cyclical downturns.

Source: Counterpoint Research

> Samsung has enjoyed strong growth and scale over the past decade with a leadership position in supplying NAND and DRAM for smart devices and traditional servers. However, Samsung was not able to take full advantage of the AI server boom which was driven by NVIDIA. Samsung missed the boat and fell behind, failing to qualify as a supplier of HBM solutions for NVIDIA.

> Its rival, SK Hynix was the biggest beneficiary as it was “flexible” enough to work with NVIDIA to meet the requirements and team up with Samsung’s arch rival TSMC to integrate and package them. SK Hynix raked in almost $10 billion worth HBM in 2024.

Read here: Samsung’s HBM Saga

> Counterpoint’s MS Hwang highlighted in the above report that Samsung plans to launch revised HBM3e products in H1 2025 and HBM4 later in the year. The updated 12-layer HBM3e “must” enhance power efficiency and yield over existing products. The goal is to complete development by the end of Q1 2025 or early Q2 2025, with sales expected from late Q2 2025 to early Q3 2025 after customer qualification.

> At the same time, there are strong headwinds from the China market, where the homegrown CXMT is making significant strides to tap into Samsung’s China-based client roster.

Read here: The Growth of China’s CXMT

> The DeepSeek reality check could be a blessing in disguise for Samsung and other chipmakers that missed the AI boat. DeepSeek’s claims could slow AI infrastructure investment down or atleast make companies rethink, shifting focus to cost-cutting rather than immediate expansion. It is a speed-breaker in the semis demand rally, stretching out the hundreds of billions in AI infra investments over a much longer period. If it plays out like this, it could be a good opportunity for players like Samsung to play some catch up.

Read here: Will the DeepSeek Approach Reshape the Next AI Wave?

> To make the situation worse, Samsung is seeing mixed results in the Edge Gen AI race, losing slots within its flagship S25 series and most of the Qualcomm Snapdragon 8 series-based Gen AI smartphones to Micron’s DDR5X.

Read here: The Role of Memory in GenAI Smartphones

Foundry Business: Yielding to TSMC

> Meanwhile, Samsung Foundry has been suffering from attracting a strong client roster for its advanced nodes and have lost significant business to the rival TSMC. The yield and UTRs have been disappointing for its 3nm and other leading nodes.

> This has also negatively effected Samsung Foundry’s biggest internal customer which choses 100% Qualcomm Snapdragon for its latest flagship sixth generation Galaxy Z foldables and S25 series instead of flagship Exynos series. Having said that, Samsung LSI did well expanding mid- to high-tier Exynos series chip orders for the Galaxy A series which offsetted some decline.

MX (Mobile): Premiumization & AI Partnerships Keeps it Afloat

> Galaxy Smartphones, wearables and others continue to hold the fort steady for Samsung raking in a top line of $85 billion annually. Though it is a decline from the $115-billion peak in the 2013-2014 time frame when the smartphone market was just picking up globally.

Source: Counterpoint Research

> While the volumes have been flat and way lower than the peak in 2017-18, the growing “premiumization” trend across regions has helped Samsung maintain a healthy revenue run rate. Further, Samsung has reduced focus on sub-$150 smartphone segment which has hurt some volume but helped maintain better bottom lines.

> China is still an elusive market for Samsung, reducing the overall opportunity for Samsung. This is similar to how the US market is not accessible to China-based brands. However, Apple is the only brand that enjoys “true” global scale and head on competes with Samsung, which is almost a “zero sum game”.

> Samsung’s strong partnerships with Google and Qualcomm to bring industry-leading experiences has been working for the brand to differentiate and edge ahead of Apple. For Google and Qualcomm, it is also important to partner with Samsung considering its global scale in “Google Android” markets. The Gemini integration powering Galaxy AI optimized on Snapdragon is powerful but is also a question mark on its vertical efforts and performance of its in-house Exynos compute and its own memory which is relegated to mostly mid-tier Galaxy smartphones.

Wrapping Up:

• It’s a shame, Samsung, the industry leader, missed the most important and lucrative AI wave. It lost a $30 billion-$45 billion opportunity over the last three years!
• The miss has been not only in HBM but also on the Foundry side affecting its position in edge AI markets.
• However, we believe this was a low point for the South Korean vendor and did not overall hurt its earnings per se but only cause “opportunity loss” and reduced “mindshare”.
• With its scale, expertise, and determination, Samsung can capitalize on upcoming AI advancements, from Gen AI to Agentic AI to Physical AI.