Saturday, October 10, 2026
GlobalFoundries and TSMC announced a five-year, $2 billion agreement on Thursday under which GlobalFoundries will make silicon interposers, a component of TSMC's CoWoS advanced packaging, at its Malta, New York, plant. Volume production is due in the first half of 2028.
The Department of War's Office of Strategic Capital has issued a conditional commitment for up to $1.5 billion in long-term financing to Wolfspeed, a silicon carbide and gallium nitride chipmaker that emerged from bankruptcy a year ago. Nothing has been signed.
Micron Technology reported fiscal fourth-quarter revenue of $54.23 billion, up 379 percent from a year earlier, with a non-GAAP gross margin of 87 percent as AI demand pushed memory prices higher. The shares swung both ways the next day and a week later sat slightly below where they were before the report.
Marvell Technology raised its fiscal 2028 revenue forecast to about $20 billion at its investor day on Tuesday, above the $18.2 billion analysts expected, and introduced a first-ever fiscal 2031 revenue range of $70 billion to $90 billion. From fiscal 2026 revenue of $8.2 billion, the targets imply compound annual growth above 50 percent for five straight years, powered by custom AI chips, networking and optical interconnects for data centers.
The United States imported $420.8 billion of goods and services in August, the most in any month on record, and the trade deficit widened to $105.6 billion, the largest since March 2025. The last record was set by consumer goods rushed in ahead of tariffs. This one is made of capital goods. Four lines of hardware account for more than the whole rise in goods imports this year.
The world's chipmakers sold $159.7 billion of semiconductors in August, 144 per cent more than a year earlier, the Semiconductor Industry Association said on Monday, and passed $1 trillion for the year to date. In February the industry's forecast for all of 2026 was roughly $1 trillion. It is now more than $1.6 trillion. Micron's results show how much of the jump is price.
Huawei and Qualcomm announced a multi-year cross-licence on Monday covering 5G, computing, AI and networking, and Qualcomm agreed to buy some of Huawei's American patents. Neither company disclosed the price. Huawei called it its first revenue-positive agreement with Qualcomm, and Nikkei reported that Qualcomm will be the net payer for the first time. The purchase needs regulatory approval.
The Computer & Communications Industry Association and four other technology trade groups wrote to President Trump on Friday warning that broad Section 232 tariffs on semiconductors, robots and industrial machinery would raise the cost of building AI data centers and factories. Advanced chips bound for American data centers are exempt from the 25 per cent tariff imposed in January. A second phase under discussion may end that.
Spending on data center construction reached a seasonally adjusted annual rate of $85 billion in August, the Census Bureau said, up 73 per cent in a year. Spending on computer, electronics and electrical factories, the category that includes chip plants, has fallen to $52 billion from $126 billion in mid-2024. Without data centers, private nonresidential construction is down about 6 per cent on a year ago.
Chip stocks fell before Monday's open after the heads of Anthropic and OpenAI said the industry should slow the rate at which AI models improve. Nvidia was down 3 percent in premarket trading, Marvell and Arm 6 percent, and ASML 9.5 percent. The labs are mostly private. The public way to express a view on slower AI is to sell the companies that supply it.
Positron raised $875m on Thursday at a $5bn valuation, having been worth about $1.06bn in February. Its chips are built around memory capacity and bandwidth rather than raw compute, using commodity LPDDR5X rather than the high-bandwidth memory that constrains everybody else. That choice is the entire thesis.
Twelve disclosed AI chip funding rounds raised $5.37bn between late January and mid-August. Eight of them, and about 67 percent of the capital, went to companies building for inference rather than training. Venture money is a forecast, and this one is about where the permanent cost of AI is going to sit.
The constraint on the AI buildout has been described as electricity for two years. The nearer one is a procurement queue for hardware whose price and availability are now set by trade policy rather than by fabrication.
One policy pays you to break ground before the year ends. The other may raise the price of the equipment you would break ground with, and has not yet said by how much or on what. Waiting is the only move that answers both.