Samsung Foundry, the contract chipmaking division of Samsung Electronics, has set a 2029 target of capturing 66% of high-performance computing (HPC) wafer orders and quadrupling its overall customer base, according to Seoul Economic Daily.

The five-year roadmap, as reported in the Seoul-based business daily, frames the unit's bid to narrow the gap with Taiwan Semiconductor Manufacturing Company (TSMC). The targets position HPC as a central pillar of Samsung Foundry's external order book.

The two quantitative anchors—a 66% HPC share and a 4× customer count—set concrete external benchmarks Samsung Foundry has published in recent years.

What does the 66% HPC target cover?

High-performance computing orders span four end-product families:

• Data-center CPUs (x86 and Arm-based server processors)
• AI training and inference accelerators (GPU and custom tensor-processing designs)
• Data-center networking processors for top-of-rack switches and fabric ASICs
• Hyperscaler custom silicon (in-house accelerators and Arm-server chips)

These workloads carry higher per-wafer pricing, longer design-win cycles measured in years rather than quarters, and concentrate demand on the most advanced process nodes.

Why is customer count tracked separately from revenue?

Foundry customer counts and revenue share move on different axes. A mobile SoC customer absorbs high wafer volumes at modest die prices. An HPC customer procures fewer wafers on advanced nodes at premium per-die prices. Quadrupling the customer roster therefore signals a broader design-win portfolio, not a fourfold revenue lift.

Strategic implications for the division include:

• Reduced revenue concentration on a small set of mobile clients
• Expanded participation in AI accelerator and hyperscaler custom-silicon programs
• Higher design-start density at leading nodes, improving fab utilization rates
• Greater geographic diversification of customers across U.S., European, and Asian fabless designers

Where does Samsung Foundry stand in the merchant market?

Samsung Foundry's node portfolio spans legacy 28nm and 14nm capacity through 3nm gate-all-around (GAA) production, with 2nm GAA in risk production through 2025. The division has held the number-two position in global merchant foundry revenue for several years, trailing TSMC by a wide margin.

Intel Foundry Services occupies a smaller third position, with selective external customers concentrated in advanced packaging and U.S. defense programs. Samsung Foundry operates fabs in South Korea and at a Taylor, Texas site, with additional capacity under construction in both countries.

How large is the HPC opportunity?

Third-party analysts expect the HPC segment to grow faster than the overall foundry market through 2028, driven by hyperscaler capital expenditure on AI infrastructure and the broader build-out of data-center capacity. Global foundry revenue tracked by industry research firms has surpassed $100 billion annually in recent years, with HPC accounting for an expanding share.

Converting HPC design starts currently held by TSMC—particularly in AI accelerator and hyperscaler custom-silicon categories—remains the central execution challenge for Samsung's 2029 target.

What details does the headline omit?

Seoul Economic Daily's headline does not specify:

• The baseline HPC share from which the 66% figure derives
• The customer-count base year used for the fourfold expansion
• Whether Samsung announced the targets at an investor event, in regulatory filings, or via press release
• Whether external customers have committed to the implied HPC wafer volume