Who Wins in a High-Rate Market? Meta Flashes Golden Cross as Micron Gets $3,000 Target

A so-called golden cross, a technical signal often interpreted as a shift into an uptrend, has appeared in Meta Platforms shares. Micron Technology, meanwhile, won a higher price target on expectations that a shortage of high-performance memory will persist for two more years. The AI infrastructure spending boom is also broadening beyond GPUs into memory, optical networking and custom chips.

Meta Flashes Golden Cross; Stock Rose a Month Later in All Prior Cases

Bloomingbit Market Insight reported on October 7 that Meta shares formed a golden cross the previous day. The pattern occurs when the 50-day moving average, a measure of short-term momentum, rises above the 200-day moving average, which tracks the longer-term trend. In practical terms, Meta’s roughly two-month average share price has moved above its roughly 10-month average.

This is Meta’s seventh golden cross. In each of the previous six cases, the stock was higher one month later. The average one-month gain was 9.3%, and the median was 8.2%. The smallest gain was 0.1%, while the largest was 22.1%.

After two weeks, the shares were higher in five of the six cases, with an average gain of 4.6%. After three weeks, they were up in all six cases, with an average gain of 5.3%.

Micron is one of the clearest beneficiaries of a shortage in high-performance memory tied to AI expansion. D.A. Davidson on October 7 raised its 12-month price target on Micron by 43% to $3,000 from $2,100, while maintaining its buy rating.

The new target is about 187% above Micron’s October 6 closing price of $1,045.56. Micron shares have climbed 266% this year and 463% over the past 12 months.

D.A. Davidson analyst Gil Luria wrote that investors are only beginning to understand Micron’s value and that the market could eventually assign the stock a higher price-to-earnings multiple.

The key issue is memory supply and demand. AI models require large amounts of high-performance memory to process more data and handle longer context windows. D.A. Davidson projects that demand will continue to outstrip supply in 2027 and 2028.

Micron is aiming to secure 50% of its revenue through long-term strategic customer agreements. A larger share of long-term contracts could improve earnings visibility and bolster pricing power. LSEG data shows that 46 of the 50 analysts covering Micron rate the stock buy or strong buy.

GPU or ASIC? Marvell Is Going After Both

Marvell Technology is drawing attention as a company poised to benefit whether the AI market remains centered on Nvidia’s GPUs or shifts toward custom AI semiconductors, or ASICs, developed by major cloud providers.

At its investor day on October 6, Marvell set a fiscal 2031 revenue target of $70 billion to $90 billion. That marks an aggressive goal compared with fiscal 2026 revenue of $8.2 billion. The company said revenue could grow at a compound annual rate of 55% to 60% over the next five years.

Marvell believes it can expand its custom semiconductor business as cloud giants including Google, Meta and Microsoft develop their own AI accelerators and networking chips. At the same time, it can supply optical communications and high-speed networking technology for Nvidia-based AI systems.

As AI clusters grow larger, the technology linking chips, servers and data centers becomes as important as the performance of the GPU or ASIC itself. Marvell stands out because it is not betting on a single AI chip, but supplying the connective infrastructure for AI systems.

SpaceX Discusses Raising as Much as $40 Billion to Secure AI Chips

SpaceX is discussing plans to raise as much as $40 billion to buy Nvidia AI chips. Options under consideration include $10 billion in bank loans and $30 billion of investment-grade corporate bonds.

If completed, the financing would rank among the largest debt raises ever tied to AI infrastructure investment. Talks remain at an early stage, and a deal may not materialize.

SpaceX’s ability to raise that amount is tied to computing lease agreements. Anthropic has reportedly signed a contract to use computing capacity at the Colossus 1 data center in Memphis. Google and AI startup Reflection AI have also been cited as customers.

Recent AI infrastructure financing has often been structured around the long-term cash flow generated by GPUs rather than the chips themselves. A special purpose vehicle, or SPV, secures the GPUs, and interest and principal are repaid with computing fees paid by customers. In effect, Wall Street is placing more weight on the quality of customer contracts attached to the chips than on the resale value of the hardware.

Musk’s ‘Terafab’ Marks a Vertical-Integration Bet on Chips

Elon Musk has said companies he controls will build and operate the large semiconductor plant planned in Texas, known as Terafab. TSMC could lease part of the site if it chooses, but it would not run the factory.

Terafab is part of a vertical-integration strategy aimed at securing a stable supply of chips for Tesla’s autonomous-driving and robotics operations, as well as AI chips for data centers used by SpaceX and xAI.

The plan has three main goals: reducing production delays caused by chip shortages, speeding design changes and the shift to mass production, and strengthening long-term bargaining power on price and volume with external suppliers.

The project’s success, however, will depend less on construction than on yield. If advanced chips cannot be produced at scale with stable yields, an in-house manufacturing base could instead become a major cost burden.

Park Shin-young, New York correspondent, Hankyung.com, nyusos@hankyung.com