South Korea’s major IT services companies posted mixed second-quarter results, yet the progress of their artificial intelligence (AI) businesses — the key future growth driver — is showing a different trajectory. Samsung SDS and LG CNS are accelerating monetization with concrete achievements in GPU-as-a-Service (GPUaaS), AI data centers, and physical AI. Hyundai AutoEver, by contrast, presents a stark contrast, with its role in the group’s physical AI strategy, business scale, and monetization timeline still shrouded in uncertainty.

According to industry sources on Sunday, Hyundai AutoEver recorded Q2 revenue of ₩1.25 trillion (approximately $874.6 million) and operating profit of ₩90.5 billion (approximately $63.3 million), up 20% and 11.3% year-over-year, respectively, both beating market consensus. The enterprise IT business drove the results, with segment revenue reaching ₩1.04 trillion (approximately $726.1 million), up 27.9% from a year earlier. This was fueled by growing cloud demand and a higher proportion of high-value projects, including the group’s next-generation enterprise resource planning (ERP) implementation. The results also benefited from approximately ₩20 billion (approximately $14.0 million) in enterprise IT revenue carried over from the first quarter.

Samsung SDS posted Q2 revenue of ₩3.72 trillion (approximately $2.6 billion) and operating profit of ₩231.8 billion (approximately $162.1 million), up 5.9% and 0.7% year-over-year, respectively. While cloud revenue grew 17% to ₩779.4 billion (approximately $545.0 million), operating profit growth was limited due to upfront investments for expanding external business in public and financial sectors, as well as intensifying competition. The IT services segment’s operating margin also declined 0.6 percentage points year-over-year.

LG CNS reported Q2 revenue of ₩1.52 trillion (approximately $1.1 billion), up 4.2%, but operating profit fell 9.2% to ₩127.9 billion (approximately $89.4 million), missing market expectations. The decline was attributed to contract schedule delays for certain affiliate projects pushed to the second half, along with increased investment in new AI platforms and physical AI initiatives. Song Kwang-ryun, CFO of LG CNS, said during the earnings conference call on July 31: “The deferred projects are proceeding normally, and as contract signings and project execution ramp up in the second half, they will contribute to earnings improvement.”

CompanyQ2 RevenueYoY ChangeQ2 Operating ProfitYoY ChangeSamsung SDS₩3.72 trillion+5.9%₩231.8 billion+0.7%LG CNS₩1.52 trillion+4.2%₩127.9 billion-9.2%Hyundai AutoEver₩1.25 trillion+20%₩90.5 billion+11.3%

Note: Based on each company’s earnings release.

Industry observers note that short-term earnings alone are insufficient to gauge mid-to-long-term prospects, as the AI business scorecards of these companies tell a different story. All three firms are supporting AI infrastructure expansion and AI transformation (AX) initiatives for their respective group affiliates, while positioning physical AI — which integrates robotics with manufacturing and logistics systems — as their next growth pillar.

Samsung SDS disclosed the operational status and order pipeline for its GPU services and AI data center business during the earnings call. According to the company, it launched NVIDIA B300-based GPUaaS through Samsung Cloud Platform (SCP) in March, and demand has been so strong that remaining available capacity is now limited. The company expects B300 to reach full operation starting in the third quarter. CEO Lee Jun-hee stated: “Having confirmed robust demand for the latest GPU models, we will continue to expand supply going forward.”

The company also secured contracts in its AI data center business. Samsung SDS announced it won its first order in the DBO (Design, Build, Operate) business — where it takes on end-to-end data center design, construction, and operations for enterprises — in the first quarter, with multiple additional business opportunities materializing in the second quarter. The company plans to expand its AI infrastructure capacity from the current 110 megawatts (MW) to 230MW by 2029 and over 800MW by 2031. Samsung SDS also unveiled its digital asset business plans. In May, it acquired a 4% stake in Dunamu, and is exploring opportunities in stablecoin infrastructure, virtual asset-based financial SI, and next-generation payment services. CEO Lee said: “The equity investment in Dunamu is a strategic move to enter the digital asset infrastructure business. Our goal is to combine Dunamu’s blockchain operational expertise with Samsung SDS’s IT services, AI, cloud, and security capabilities to strengthen our digital financial infrastructure business.”

LG CNS is expanding its business scope around enterprise AI platforms, AI computing infrastructure, and physical AI. After launching its enterprise agentic AI platform “AgenticWorks” in the first quarter, the company introduced “PhysicalWorks,” a robot learning and operations platform, in the second quarter. In July, it also launched “XPUWorks,” which provides AI computing resources such as GPUs and neural processing units (NPUs) on a subscription basis.

For physical AI, LG CNS plans to build business cases this year through group projects and field validation, then expand to external customers starting next year. The company recently signed a ₩189.7 billion (approximately $132.7 million) contract with LG Electronics to supply GPU infrastructure and PhysicalWorks for humanoid robot learning. It also completed a first-phase proof of concept (PoC) with Kurly, a South Korean e-commerce company. The company’s strategy is to focus on field validation and platform advancement this year, then aggressively expand robot transformation (RX) and robot data factory construction projects for manufacturing and logistics customers starting in 2027. Overseas, LG CNS plans to complete Indonesia’s AI data center — its first overseas order won by a South Korean company — in the second half, while also expanding its push into the North American manufacturing AX market.

Hyundai AutoEver, in contrast, is expected to benefit from Hyundai Motor Group’s physical AI investments, but its specific role, business scale, and monetization timeline have yet to be revealed. The group’s robot training and validation facility, the Robot Metaplant Application Center (RMAC), is scheduled to begin operations in August, but Hyundai AutoEver’s scope of work and related revenue figures have not been disclosed.

Securities analysts acknowledge that the group’s AI and robotics investments could present growth opportunities for Hyundai AutoEver, but emphasize the need for more concrete business plans. Shin Yoon-chul, an analyst at Kiwoom Securities, said: “Contrary to aggressive market expectations, the specific scale, timing, and funding plans for Hyundai AutoEver’s new businesses have not been fleshed out. Despite successive meetings between Hyundai Motor Group and NVIDIA’s top executives, visibility into the new business remains murky.”

Meanwhile, POSCO DX saw Q2 revenue and operating profit decline 14.8% and 42.3% year-over-year, respectively, due to investment adjustments by its parent group. POSCO DX has declared itself an “AI-native company” and plans to pursue business around physical AI and agentic AI as its two pillars. However, external customer expansion — which stood at just 3% in the second quarter — remains a challenge.

An industry official said: “SI companies are enjoying tailwinds in their earnings amid the AI transformation era. Nearly all firms are increasing related investments, so strong results are expected for the time being. But with competition intensifying, strategies for winning customers have become more critical than ever.”