The launch of OpenAI's latest flagship model, Astra, has once again sparked enthusiasm for AI computing infrastructure, with memory chips emerging as one of the first sectors to benefit.

Last Friday, US semiconductor stocks rallied broadly, and by Monday, the momentum had extended to Asian markets, with major memory chip makers like Samsung Electronics, SK Hynix, and Kioxia Holdings seeing notable gains.

ChatGPT-6 Astra was rolled out in two phases last Thursday and Friday. OpenAI President Greg Brockman called it "the world's most intelligent and well-aligned model." Since Astra is designed for more complex AI tasks, the market is increasingly betting that continuous iteration of frontier models will drive up computing demand, supporting demand for core hardware such as HBM, DRAM, and GPUs.

This expectation quickly translated into share price gains. Micron Technology rose more than 6% last Friday, SanDisk jumped nearly 12%, and SK Hynix's Nasdaq depositary receipts climbed over 8%. On Monday, Asian markets picked up the baton, with SK Hynix gaining about 8%, Samsung Electronics up over 5%, and Kioxia rising roughly 10%.

Institutions believe that as AI model capabilities continue to improve, computing demand is still expanding rapidly, and bottlenecks in memory and networking segments could become more pronounced, positioning the "memory chip trade" to potentially re-emerge as a key theme in AI infrastructure investing.

AI Computing Expansion Puts Memory Chips Back in the Spotlight

The core of the rally sparked by Astra is not just the release of a single model, but the market's renewed awareness that continuous improvements in AI model capabilities will still translate into higher computing and memory requirements.

Charu Channa, Chief Investment Strategist at Saxo Bank, noted that Astra's release further supports sustained growth in AI spending. As frontier models increasingly demand more computing power, bottlenecks in memory and networking segments are widening, which will benefit memory chip makers such as SK Hynix and Samsung Electronics.

Analysts at Goldman Sachs and Morgan Stanley project that global AI capital expenditure will reach $1.3 trillion to $1.5 trillion by 2027, with more than half going toward memory infrastructure. If this trend continues, storage products like HBM and DRAM are expected to benefit consistently from AI computing expansion.

Meanwhile, Tim Moe, Chief Asia Equity Strategist at Goldman Sachs, reiterated his year-end target of 12,000 points for the Korean stock index, noting that earnings expectations in the Korean market continue to be revised upward, with this week's upgrade reaching 3.2%. He also emphasized that foreign investors' holdings in Korea's semiconductor sector remain significantly below historical averages.

Previous discussions about a semiconductor "supercycle" focused more on whether AI capital spending can be sustained and whether chipmakers can maintain high growth and high margins. Now, as newer AI models continue to push computing demand higher, memory is once again becoming a key beneficiary in the AI infrastructure trade.

The stronger the model, the higher the computing demand behind it, and the greater the need for memory. Astra's release is reigniting this investment logic.