Korea’s ‘father of HBM’ sees 1,000x AI memory surge as Google’s TurboQuant faces real-world tests

Alphabet’s Google has unveiled its KV cache quantization compression technology, TurboQuant, promising dramatic reductions in memory usage for AI inference. While the innovation has captured global attention, South Korea’s academic and industrial sectors remain skeptical about its practical feasibility, even as they firmly expect AI inference to continue driving substantial growth in memory demand.

Continue reading

ASE Technology sees seasonality fade as AI demand drives steady 2026 growth

Global outsourced semiconductor assembly and test (OSAT) leader ASE Technology Holding (ASEH) held an earnings call on April 29, reporting a robust first quarter of 2026 that broke from the traditional seasonal slowdown. Growth was driven primarily by sustained strong demand for leading-edge advanced packaging (LEAP) technologies, alongside a simultaneous rebound in wire bonding orders — together fueling an 87% year-on-year surge in quarterly net profit after tax, the second highest on record for the period.

Continue reading

SpaceX’s IPO gives Musk unchecked control via ‘Texas fortress’ strategy

As SpaceX prepares for a public debut at a valuation targeting US$1.75 trillion, new regulatory filings have revealed a corporate governance structure that positions Elon Musk as an unassailable leader. According to Reuters, the company’s IPO filing includes a provision stating that Musk cannot be removed from his roles as CEO and chairman of the board without his own consent.

Continue reading

Taiwan OSAT Powertech lifts capex to US$1.6bn, targets AI packaging growth

Memory packaging and testing provider Powertech Technology posted net profit of NT$1.84 billion (US$57 million) for the first quarter of 2026, its second-highest for the same period, and raised its full-year outlook. The company increased its planned 2026 capital expenditure from NT$40 billion to NT$50 billion and expects broad price increases for logic and memory products in the second quarter of 2026, supporting sequential revenue gains and high single-digit to low double-digit annual growth.

Continue reading

Google’s split TPU chips signal shift from universal to specialized AI accelerators

Google has unveiled its eighth-generation Tensor Processing Units (TPUs) by splitting them into two distinct chips: the training-focused TPU 8t and the inference-optimized TPU 8i. This move goes beyond simply designing two chips; it reflects Google’s capability to break down bottlenecks across different stages of the model lifecycle, restructuring chip design, interconnects, memory, scheduling, and software stacks.

Continue reading