When the underlying economics of infrastructure commitments meet the financing muscle of the companies supplying foundational compute, the traditional boundary between supplier and customer begins to blur.
China's National Day holidays have opened one of the fourth quarter's most important promotional windows for consumer electronics. For smartphones, the timing is especially significant: Apple, Xiaomi, Vivo, Oppo and Honor have all launched new models since September, while Huawei introduced its Mate 90 series on October 1.
As Elon Musk's Terafab ambitions advance, the realities of modern silicon manufacturing become increasingly apparent, prompting Musk to enter talks with industry mainstays like Intel and TSMC for assistance.
Singapore is building a semiconductor R&D hub by attracting both multinational investment and overseas research talent. Dr Yu-chieh Chien, a Taiwanese scientist at the Agency for Science, Technology and Research's Institute of Microelectronics (A*STAR IME), is one example.
As the number of AI computing transistors integrated within a single CoWoS package rises rapidly, DIGITIMES observes that moving massive volumes of data to compute chips fast enough is becoming increasingly critical to fully utilizing available computing power.
This excerpt from DIGITIMES analyst Luke Lin's podcast looks at TSMC's rumored Texas expansion, Intel's 14A and 18A process debate, and Qualcomm's high bandwidth compute (HBC) push as AI workloads drive demand for hybrid bonding and advanced packaging.
The rapid iteration of AI-capable hardware, following an industry-wide push toward integrating AI-processing capabilities into PCs, is bringing renewed attention to what AI could offer as its presence extends further into the notebook market. But beneath the expanding promise of these machines lies a less straightforward question: what is actually compelling buyers to bring them into their businesses and homes?
Nvidia's new AI agent security platform puts hardware at the centre of agent control, but it also forces enterprises to weigh stronger protection against higher infrastructure costs and deeper dependence on Nvidia's stack.
TSMC said the next phase of AI performance growth will depend less on shrinking a single chip and more on combining multiple chips, memory, and advanced packaging inside larger systems. The shift could reshape semiconductor competition, as demand for compute, bandwidth, power delivery, and cooling rises in tandem. The company outlined the trend at its heterogeneous integration forum during SEMICON Taiwan 2026, saying AI computing transistors in a single CoWoS package are set to rise more than 48-fold by 2029 from 2024 levels, while total HBM bandwidth is expected to increase more than 34-fold.
DIGITIMES Intelligence observes that AI applications are moving through a clear progression: from early classification AI focused on feature recognition, to generative AI capable of creating content, and toward a broader shift into the agentic AI era in 2027. Agentic AI is expected to multiply compute demand, while multi-step reasoning introduces new risks to task quality. Efficient control of total token consumption and reliable task-quality management will therefore become critical to commercial deployment.
South Korea's leading telecommunications operator, SK Telecom (SKT), recently established SK Hyper and SK Horizon in quick succession. The moves mark a pivotal shift for telecommunications AI, signaling a transition from internal operational optimization toward full-scale infrastructure monetization.
Xunfei Healthcare (iFLYTEK Medical), the healthcare AI arm of China's iFlytek, is using its Spark Medical Large Model to support diagnosis and medical record generation while tackling challenges including clinical workflow integration, differences in care settings, and model hallucinations. It also hopes to bring its experience from China to South Korea through local partnerships.
As HBM approaches half the cost of a GPU-HBM CoWoS package, can memory still be viewed as merely a passive component of AI computing? The answer goes back to the "memory wall," a challenge that has shaped nearly half a century of semiconductor development.
Garmin's decision to offer OLED options but no Micro LED version in its new flagship smartwatches, the fenix 9 and fenix 9 Pro, has revived debate over whether the emerging display technology is ready for sustained commercial competition. The shift comes about a year after Garmin launched the world's first Micro LED smartwatch, the fenix 8 Pro Micro LED in September 2025.
Micron unveiled the Micron Ventures Paradigm Fund, a US$250 million venture vehicle, in mid-August 2026, as the memory maker pairs more than US$25 billion in fiscal 2026 capex with a pledge to invest more than US$250 billion in the US by 2035. The fund is designed to give Micron an early look at technology shifts, position it for next-generation AI architectures, and strengthen its strategy to reduce memory-cycle volatility.
