Xingyun Technology signs 921.6 million yuan computing power deal; Texas Instruments raises prices for third time this year

Advanced Micro Devices CEO Lisa Su said chip demand remains ahead of supply and expects it to stay very high for the next several years, speaking to reporters in Taipei on October 6. Su said AMD has increased supply through 2026 and plans a much bigger jump in 2027, with capacity now planned three to five years ahead. Her Asia trip included meetings with Foxconn and TSMC, followed by talks with memory makers in South Korea, as AI chips depend on large amounts of memory that constrain how many AMD can ship. Stifel raised its AMD target to $700 on October 5, arguing AI buildouts are limited by chip supply. AMD's forward non-GAAP P/E of 83.38x and forward price-to-sales ratio of 20.25x both sit at more than double their 5-year averages, while analysts expect EPS to rise about 82% in 2026, 107% in 2027 and 45% in 2028.

Applied Materials and Intel announced on October 6 that they are expanding their partnership on next-generation AI chipmaking, working together on transistors, the wiring that connects them, and advanced packaging, including Intel's Foveros technology for stacking chips in 3D. Teams will work out of Applied's EPIC Center in Silicon Valley and Intel's research campus in Oregon, with Intel CEO Lip-Bu Tan saying the goal is to move new technology from the lab into high-volume U.S. production faster. Intel CFO Dave Zinsner has said packaging can reach 40% gross margins, and some deals could be worth billions of dollars a year, though industry reports linking Intel to potential advanced packaging deals with major cloud customers have not been publicly confirmed. Intel trades at a forward non-GAAP P/E of 73.94x, about 20% above its 5-year average, and a forward price-to-sales ratio of 9.37x versus an average of 3.17x, while carrying about $21 billion in net debt. Applied Materials trades at a forward non-GAAP P/E of 41.47x and a forward price-to-sales ratio of 12.29x, and unlike Intel it is net cash positive, with management expecting packaging revenue to grow more than 70% in calendar 2026.

Amazon.com is reportedly considering moving roughly $8 billion of Nvidia Grace Blackwell chips into a special-purpose vehicle financed by outside investors and then leasing the hardware back, a structure that would make part of its AI buildout more asset-light as capital spending is expected to reach $220 billion this year. AWS revenue rose 37% to $42.2 billion in the second quarter, its fastest growth in more than four years, and contract backlog reached $496 billion, while CEO Andy Jassy said the company still lacks enough capacity to meet demand. Trailing-12-month free cash flow swung to negative $7.6 billion from positive $18.2 billion a year earlier as infrastructure spending accelerated. Amazon trades at 23.64 times forward earnings, and Tigress Financial recently raised its target to $385 from $315, arguing earnings from the investment cycle are approaching an inflection point where they grow faster than operating capital. The proposal also raises the question of residual chip value, since Grace Blackwell will eventually be superseded by Vera Rubin while Amazon assumes semiconductor generations remain useful for at least five years.

GlobalFoundries announced FDX Fusion, a next-generation FD-SOI platform to be developed and manufactured in Dresden for highly integrated, energy-efficient Physical AI chips spanning sensing, compute, control and connectivity functions. The platform blends seven nanometer class digital performance with advanced RF, analog and embedded memory on a single chip aimed at edge and automotive, industrial and consumer AI systems. The launch fits GlobalFoundries' broader shift away from weaker smart mobile demand toward automotive, industrial, communications and AI applications, though it does not immediately change the near term risk that higher investment and oversubscribed niches like SiGe could still coexist with only high 80s percent fab utilization. Among recent announcements, a US$2 billion, five year silicon interposer deal with TSMC reinforces GlobalFoundries' role in AI hardware supply alongside FDX Fusion, with the Dresden FDX roadmap and Malta advanced packaging capacity sitting squarely against the key catalyst of moving its revenue mix toward higher value, AI linked processes. GlobalFoundries' narrative projects $10.4 billion revenue and $1.7 billion earnings by 2029, requiring 14.5% yearly revenue growth and a roughly $1.0 billion earnings increase from $716.0 million today, while the most optimistic analysts already assumed about US$11.7 billion of revenue and US$2.3 billion of earnings by 2029.

Nvidia is in early-stage discussions to acquire or increase its investment in Reflection AI, a U.S. startup focused on developing open-weight models, the Financial Times reported, citing people familiar with the discussions. The structure of a potential deal could take several forms, including a so-called acqui-hire that would let Nvidia hire staff and license technology while avoiding antitrust scrutiny tied to a full acquisition, and deal terms currently under discussion were not available. Nvidia, the world's most valuable company, has already invested $800M into Reflection, which was last valued at $25B in a March funding round. Nvidia, whose CEO Jensen Huang has in the past called for an open-weight ecosystem for AI, could also make an additional equity investment in Reflection or reach an agreement to supply more chips or computing power to the company. Multiple people with knowledge of the matter said a deal could be reached in the coming weeks, while cautioning that the companies could also walk away from the talks; Nvidia and Reflection did not respond to the Financial Times' requests for comment.