Samsung Electronics Foundry Expected to Reach 100% Utilization Within the Year... Countdown to Profitability Samsung Electronics' foundry division is understood to be targeting 100% utilization within the second half of this year. Current utilization is estimated at 70 to 80%, and given the order backlog and the state of contract negotiations, the mood inside and outside Samsung is that hitting the target is all but certain. With the foundry lines, which endured a prolonged slump at below 50% utilization from 2024, now approaching full operation after roughly a year, a turning point also appears to be forming in the division's chronic loss making structure. According to industry sources on the 3rd, behind the rise in utilization are high bandwidth memory (HBM) base dies and expanding demand for advanced products centered on large US Big Tech customers. With 4nm applied to sixth generation HBM (HBM4), the analysis is that a structure has formed in which the memory boom feeds directly into foundry utilization. On top of that, 2nm orders are expanding among major cloud service providers (CSPs) and artificial intelligence (AI) and high performance computing (HPC) customers, and project discussions with large customers such as Broadcom are continuing. Utilization is the core indicator of how much a fab is actually running. Contract chipmaking carries a heavy fixed cost burden because of the scale of capital investment, and the longer lines sit idle, the more losses snowball. The industry explanation for the foundry division's chronic losses last year is that utilization on mature node lines such as 4nm and 5nm fell below 50%, leaving it unable to cover fixed costs. Conversely, once utilization returns to a normal track, incremental revenue flows straight through to profit. Moves toward customer diversification are also notable. Samsung's foundry has long been criticized for heavy dependence on specific customers, but recently Big Tech firms including Qualcomm, AMD and Google have begun sitting at the negotiating table. Tesla is also reported to be placing next generation chip volumes with the 2nm process, thickening the medium and long term order pipeline. The common view, however, is that turning these into actual large scale mass production contracts requires 2nm yields to settle in the 70% range, and how quickly yields, currently estimated in the 60% range, improve from here is cited as the point to watch. A restructuring of the business mix is proceeding alongside. The advanced node share of revenue is expected to exceed 50% this year, and the AI and HPC share is expected to expand from the high teens last year to more than 30% this year. In the second half, mass production of mobile products on the second generation 2nm (SF2P) process begins. Capacity expansion is also picking up pace. Taylor Fab 1 in Texas is preparing to start operations this year as planned, and Taylor Fab 2 is to break ground within the year with mass production targeted for 2030. With customer inquiries about the 1.4nm process increasing, plans to secure additional fabs are also under review. This trend was also evident on the second quarter earnings conference call held on July 30. Samsung said that day that "utilization is improving year on year across all nodes" and that "leading edge nodes at 8nm and below have reached maximum levels through maximizing sales centered on products with strong high growth demand." It projected that 2nm project wins this year would more than double from last year, and on the timing of a return to profit said it was "difficult to state an exact point" while noting it "should be possible in the near future." Analysts expect the non memory businesses, including foundry and System LSI, to turn profitable as early as the third quarter and by the fourth quarter at the latest. Some also see the pace of earnings improvement running faster than the market expects once wafer price increases are reflected on top of normalized utilization. Parts of the industry, however, offer a more cautious view that whether 2nm yields exceed 70%, rather than reaching 100% utilization itself, will be the watershed that leads to full scale orders from large customers such as Qualcomm and AMD. One industry official said, "Utilization normalization is a leading indicator that the foundry business is passing breakeven, but confirming a real structural improvement requires watching until 2nm yields stabilize and large customers actually convert to mass production."

Samsung Electronics Foundry Expected to Reach 100% Utilization Within the Year... Countdown to Profitability Samsung Electronics' foundry division is understood to be targeting 100% utilization within the second half of this year. Current utilization is estimated at 70 to 80%, and given the order backlog and the state of contract negotiations, the mood inside and outside Samsung is that hitting the target is all but certain. With the foundry lines, which endured a prolonged slump at below 50% utilization from 2024, now approaching full operation after roughly a year, a turning point also appears to be forming in the division's chronic loss making structure. According to industry sources on the 3rd, behind the rise in utilization are high bandwidth memory (HBM) base dies and expanding demand for advanced products centered on large US Big Tech customers. With 4nm applied to sixth generation HBM (HBM4), the analysis is that a structure has formed in which the memory boom feeds directly into foundry utilization. On top of that, 2nm orders are expanding among major cloud service providers (CSPs) and artificial intelligence (AI) and high performance computing (HPC) customers, and project discussions with large customers such as Broadcom are continuing. Utilization is the core indicator of how much a fab is actually running. Contract chipmaking carries a heavy fixed cost burden because of the scale of capital investment, and the longer lines sit idle, the more losses snowball. The industry explanation for the foundry division's chronic losses last year is that utilization on mature node lines such as 4nm and 5nm fell below 50%, leaving it unable to cover fixed costs. Conversely, once utilization returns to a normal track, incremental revenue flows straight through to profit. Moves toward customer diversification are also notable. Samsung's foundry has long been criticized for heavy dependence on specific customers, but recently Big Tech firms including Qualcomm, AMD and Google have begun sitting at the negotiating table. Tesla is also reported to be placing next generation chip volumes with the 2nm process, thickening the medium and long term order pipeline. The common view, however, is that turning these into actual large scale mass production contracts requires 2nm yields to settle in the 70% range, and how quickly yields, currently estimated in the 60% range, improve from here is cited as the point to watch. A restructuring of the business mix is proceeding alongside. The advanced node share of revenue is expected to exceed 50% this year, and the AI and HPC share is expected to expand from the high teens last year to more than 30% this year. In the second half, mass production of mobile products on the second generation 2nm (SF2P) process begins. Capacity expansion is also picking up pace. Taylor Fab 1 in Texas is preparing to start operations this year as planned, and Taylor Fab 2 is to break ground within the year with mass production targeted for 2030. With customer inquiries about the 1.4nm process increasing, plans to secure additional fabs are also under review. This trend was also evident on the second quarter earnings conference call held on July 30. Samsung said that day that "utilization is improving year on year across all nodes" and that "leading edge nodes at 8nm and below have reached maximum levels through maximizing sales centered on products with strong high growth demand." It projected that 2nm project wins this year would more than double from last year, and on the timing of a return to profit said it was "difficult to state an exact point" while noting it "should be possible in the near future." Analysts expect the non memory businesses, including foundry and System LSI, to turn profitable as early as the third quarter and by the fourth quarter at the latest. Some also see the pace of earnings improvement running faster than the market expects once wafer price increases are reflected on top of normalized utilization. Parts of the industry, however, offer a more cautious view that whether 2nm yields exceed 70%, rather than reaching 100% utilization itself, will be the watershed that leads to full scale orders from large customers such as Qualcomm and AMD. One industry official said, "Utilization normalization is a leading indicator that the foundry business is passing breakeven, but confirming a real structural improvement requires watching until 2nm yields stabilize and large customers actually convert to mass production."

Source:https://x.com/jukan05/status/2084179469754110380