According to the latest foundry supply tracker report from Counterpoint Research, the global foundry 2.0 market generated $96.6 billion in revenue in the second quarter of 2026, representing a 25% year-over-year increase and an 11% quarter-over-quarter rise.

The market's growth rate accelerated compared to the previous quarter, driven primarily by sustained demand growth for AI GPUs and the expansion of AI infrastructure, including custom ASICs, server CPUs, networking chips, and power management ICs.

Brady Wang, Associate Director at Counterpoint Research, stated: "With pure-play foundries, non-memory IDMs, and OSAT vendors all achieving double-digit revenue growth, the impact of the AI investment cycle has extended across every link in the industry value chain. While TSMC remains the primary beneficiary of the AI demand cycle, mature-node foundries, non-memory IDMs, and OSAT vendors are also seeing more growth opportunities. As AI processors grow larger in size and increase in design complexity, advanced packaging remains the key constraint."

Counterpoint Research estimates that the supply-demand gap for TSMC's CoWoS will remain at approximately 20% in 2026, and is expected to narrow to about 10% in 2027 as its internal capacity expands to 185,000 wafers per month (KWPM). Extended test times and increased test steps are prompting OSAT vendors to gradually become strategic second-source suppliers within the AI semiconductor ecosystem.

Pure-Play Foundry Market Maintains Growth

In the second quarter of 2026, the overall pure-play foundry industry generated $54.9 billion in revenue, up 29% year over year.

Among them, Taiwan Semiconductor Manufacturing (TSM) reported revenue of $40.2 billion, up 34% year over year and 12% quarter over quarter, accounting for approximately 42% of the total foundry 2.0 market, up from 38% in 2025.

Although Taiwan Semiconductor Manufacturing's year-over-year growth rate slowed from 41% in the first quarter of 2026 due to a higher base, its quarter-over-quarter momentum remained relatively solid. Counterpoint Research expects Taiwan Semiconductor Manufacturing to achieve full-year 2026 revenue growth of approximately 43%. This implies year-over-year revenue growth of nearly 50% in the second half of 2026, contingent on whether the N2 process ramp-up, wafer price increases, and CoWoS capacity expansion proceed as planned.

CoWoS capacity constraints have limited AI product shipments, prompting Taiwan Semiconductor Manufacturing to increase back-end investment while customers simultaneously develop alternative packaging solutions. While this will not change Taiwan Semiconductor Manufacturing's leading position in customers' advanced products, manufacturing capacity across advanced/mature nodes and packaging is undergoing a structural reallocation.

Foundry 2.0 Market Revenue Share, Q2 2026 Source: Counterpoint Research Foundry Tracker, Q2 2026

Excluding Taiwan Semiconductor Manufacturing, pure-play foundries' revenue grew 18% year over year, up from 9% in the first quarter of 2026.

Among them, SMIC (Semiconductor Manufacturing International Corporation) delivered strong performance with 36% year-over-year growth. SMIC benefited from growing domestic AI chip demand, and its N+2 capacity is expected to remain tight in 2026 with high capacity utilization rates. Meanwhile, Hua Hong and Nexchip saw revenue growth of 17% and 21% year over year, respectively.

Taiwan region's second-tier foundries showed mixed performance. United Microelectronics Corporation (UMC) and Vanguard International Semiconductor reported year-over-year revenue growth of 9% and 20%, respectively, and quarter-over-quarter growth of 13% and 9%, respectively, mainly driven by demand for power management ICs (PMICs), MCUs, and other peripheral chips, which boosted capacity utilization rates. However, supply-demand tightness in mature nodes is concentrated in AI-related applications such as PMICs and silicon interposers, rather than being a broad industry-wide shortage, as consumer market demand remains weak.

While continuing to optimize manufacturing capabilities through advanced node capacity allocation, Vanguard is taking on more silicon interposer demand, while UMC is benefiting from improved demand in PMIC and silicon photonics areas. GlobalFoundries posted relatively modest revenue growth, as weakness in automotive and smartphone demand offset growth in its communications infrastructure and data center businesses, the latter mainly driven by silicon photonics and silicon germanium demand for AI optical interconnects.

William Li, Senior Analyst at Counterpoint Research, noted: "As customers seek to diversify their supply sources, Samsung and Intel Foundry are also benefiting. Thanks to demand for SF4/SF5 processes and rising wafer prices, Samsung maintained approximately 4% market share in the second quarter of 2026, ranking second; meanwhile, Intel's EMIB-T packaging has been adopted by Google for its next-generation TPU. However, neither supplier can currently fully replicate Taiwan Semiconductor Manufacturing's combined advantages in advanced node scale, yield, and packaging execution capabilities."

Non-Memory IDMs Benefit from Industrial Market Recovery and AI Power Demand

In the second quarter of 2026, non-memory IDM vendors collectively achieved 19% year-over-year revenue growth.

Among them, Intel Foundry grew 31% year over year, followed by STMicroelectronics at 26%; Texas Instruments at 23%; NXP at 19%; and Infineon at 15%. Part of the growth came from a lower comparison base, as these vendors were still in inventory adjustment mode in the first half of 2025.

Counterpoint Research expects the recovery momentum for non-memory IDM vendors to continue into the second half of 2026, mainly benefiting from increased rack-level power density in AI data centers and the transition to higher-voltage power delivery architectures, which is driving demand for PMICs, MOSFETs, and silicon carbide (SiC)/gallium nitride (GaN) devices.

However, if costs cannot be passed through to downstream customers, rising foundry and packaging costs could pressure IDM vendors' profit margins.

Advanced Packaging Drives OSATs to Become Key Links in the AI Supply Chain

In the second quarter of 2026, the OSAT industry maintained its growth trajectory, with revenue rising 22% year over year and 12% quarter over quarter to $12.6 billion.

Market growth was primarily driven by AI processor demand for advanced packaging, wafer probing, and final testing. ASE Technology (ASX) reported 24% year-over-year revenue growth, while Amkor grew 26% year over year. Other OSAT vendors also achieved year-over-year growth of over 20% in the second quarter.

In the foundry 2.0 market, ASE Technology ranks second only to Taiwan Semiconductor Manufacturing in scale, gradually becoming the second supplier after Taiwan Semiconductor Manufacturing in advanced packaging, while also achieving growth in its turnkey solutions business. Amkor is expanding its 2.5D packaging capacity; Powertech Technology is pursuing more opportunities in CPU packaging and developing panel-level packaging technology.

Despite the diversification trend in back-end packaging and testing, most advanced AI computing chip wafers are still manufactured by Taiwan Semiconductor Manufacturing, so its market position remains solid.

As chip complexity continues to increase, the importance of final testing and burn-in testing is growing. However, test requirements vary across different architectures, meaning test cycles for different products will not grow in sync, and suppliers need to customize and optimize for different architectures. Therefore, in addition to existing capacity, yield stability and process integration capabilities are increasingly becoming important competitive advantages for OSAT vendors.

Outlook: Back-End Capacity and Compute Diversification Will Shape H2 2026 Market Trends

As AI demand expands from GPUs to custom ASICs, server CPUs, and power components, Counterpoint Research expects the foundry 2.0 market to maintain double-digit year-over-year growth in the second half of 2026. Advanced packaging, substrates, and test capacity will remain critical constraints for AI deployment.

At the same time, the growth of agentic AI workloads is also driving up CPU demand per AI rack, further increasing demand for advanced-node wafers and CPU packaging.

The main risks facing the market include CoWoS and substrate capacity bottlenecks, cost pass-through of rising wafer and packaging costs, continued weakness in consumer market demand, and uncertainty in trade policy.

William Li further stated: "AI demand is no longer limited to a single product. GPUs, custom ASICs, server CPUs, and power chips are all competing for limited wafer, packaging, and test capacity, so growth opportunities are expanding from Taiwan Semiconductor Manufacturing to mature-node foundries, IDMs, and OSAT vendors. In the second half of 2026, suppliers that can secure large-scale packaging and test capacity are expected to capture more demand from new AI investments."

Beyond 2026: Trillion-Dollar Packaging Bottleneck Reshapes Foundry 2.0 Market Landscape

Looking further ahead, packaging rather than logic node scaling is becoming a key area of industry competition. Counterpoint Research estimates that over the next five years, more than 130 million GPUs and AI ASICs will ship with on-compute memory using advanced packaging technology, generating nearly $2 trillion in compute-related revenue.

To achieve this, three interlocking bottlenecks must be overcome: the Memory Wall, the Performance Wall, and the Copper Wall. HBM relies on CoWoS silicon interposers, and the resulting cost, yield, and capacity limitations are driving the market to explore more alternatives.

Intel is combining EMIB bridge packaging with ZAM and XBM memory architectures to create a packaging solution beyond Taiwan Semiconductor Manufacturing's CoWoS. Qualcomm's High Bandwidth Compute (HBC) stacks LPDDR memory on logic dies, bypassing 2.5D silicon interposers, with its first products expected to launch in mid-2027. NVIDIA's NVHBM technology moves the memory controller into custom HBM base dies designed by NVIDIA and manufactured by memory vendors on demand, further expanding its application in custom accelerator projects.

For the foundry 2.0 ecosystem, value is gradually shifting toward logic base dies, bridge structures, and substrates. This presents new opportunities for foundries that can combine advanced logic processes with packaging capacity, as well as OSAT and substrate suppliers that can achieve scale qualification for these new architectures.

Neil Shah, Research Vice President at Counterpoint Research, stated: "In the next phase of AI infrastructure buildout, chip packaging will be no less important than wafer manufacturing. As memory moves closer to compute, the integrated manufacturing, packaging, and testing model represented by foundry 2.0 has evolved from concept to actual industry practice. Foundries and OSATs that can now lock in base die, bridge structure, and substrate capacity for 2027 to 2028 architectures will be better positioned in the nearly $2 trillion AI computing market opportunity."