112 articles
The SOXX semiconductor index has fallen 25% from its highs, with leveraged chip ETFs down 62%, even as AI infrastructure deal-making continues. GlobalFoundries struck a CHIPS Act partnership with the Commerce Department, AMD partnered with Cerebras, and Nvidia and SK hynix signed new AI accelerator supply agreements.
Micron and SanDisk are among 2026's top-performing chip stocks on AI memory demand, but both have fallen sharply from June record highs. Figures dated and sourced.
Semiconductor fundamentals are pulling away from Wall Street sentiment: Amkor Technology reported record earnings and SK hynix secured new AI-memory supply agreements even as leveraged AI-infrastructure ETFs fell sharply from highs. GlobalFoundries picked up a $300 million CHIPS Act award plus a government equity stake, while Skyworks and Qorvo named joint leadership ahead of their merger and Kalray struck an AI-networking partnership with Bull.
The SOXX semiconductor index fell roughly 25% in late July 2026, with leveraged chip ETFs down 62% from highs, even as Amkor posted better-than-expected results and SK hynix signed long-term AI-memory deals. Washington backed GlobalFoundries with a $300 million CHIPS award and a 1% equity stake, while Skyworks and Qorvo finalized merger leadership — consolidating suppliers just as AI infrastructure demand keeps climbing.
Amkor Technology reported a record second quarter on surging AI packaging demand, even as it guided Q3 below consensus. Across the chip supply chain, capacity investment is accelerating: Nvidia has taken a $5 billion stake in Safe Superintelligence, the U.S. government holds a CHIPS-linked equity stake in GlobalFoundries, and Skyworks and Qorvo are consolidating into one company.
Nvidia and SK Hynix signed a $500 billion memory supply deal, joined by Samsung-Broadcom's $200 billion MOU and a roughly $950 billion South Korea-US AI summit package. The U.S. government took an equity stake in GlobalFoundries as agentic AI tools from Nvidia, Cadence, Siemens, Synopsys and Silvaco move into chip design itself, while Skyworks and Qorvo merge.
Samsung and Broadcom signed a $200 billion MOU while Nvidia and SK Hynix struck a $500 billion memory deal, both in late July 2026. A South Korea-US AI summit added a roughly $950 billion package the same week. EDA firms are embedding Nvidia's agentic AI into chip design as Skyworks and Qorvo consolidate RF/analog manufacturing.
TSMC lifted its 2026 capital spending target on July 16 alongside record second-quarter results, underscoring how tight AI accelerator supply remains. Applied Materials shares are up 121% year-to-date as chipmakers pour money into new capacity.
TSMC lifted its 2026 capital expenditure guidance to $52 billion-$56 billion after diligence with customers, while SK Hynix listed on the Nasdaq to fund memory expansion. The moves, paired with ASML data showing memory's share of sales rising to 51%, point to more AI accelerator supply working through the chain ahead of an expected 40%-50% jump in hyperscaler data center spending.
Samsung's semiconductor profit rose 48-fold and SK Hynix's Q1 profit jumped 406% on AI chip demand, pushing both firms past $1 trillion in market value. The windfall is lifting South Korea's GDP but the Bank of Korea warns gains are concentrating narrowly in chipmakers, creating a K-shaped economy.
Micron says its entire High Bandwidth Memory production for 2026 has already been sold, signaling AI accelerator demand is outrunning memory supply. The sellout comes as Sanjay Mehrotra leads Micron through the AI-driven memory upcycle, positioning HBM as a chokepoint for AI chip scaling.
Groq closed a $650M funding round while Nvidia simultaneously acquired its founder and key technical team members. The structural anomaly — raising capital while core talent exits to the dominant incumbent — signals a leadership vacuum and elevated execution risk over the next 12-18 months. Nvidia stands to internalize Groq's custom inference silicon architecture learnings.
SOXX, the iShares Semiconductor ETF, is up 79% year-to-date and 152% over one year as of June 5, 2026. New silicon from Inspire Semi and Phison targets vertical markets and memory efficiency, while Chinese chipmakers pursue AI independence under U.S. export pressure. The competition is shifting from raw compute to ecosystem depth and supply chain control.
The iShares Semiconductor ETF gained 79% year-to-date through June 5, 2026, before a ~10% single-day sell-off exposed structural pressures. Rising ASIC design costs are blocking new entrants from competing in AI accelerator development. Supply concentration at Nvidia and HBM suppliers compounds deployment risk for AI model operators.
NVIDIA unveiled more than 10 products and partnerships at GTC Taipei, spanning edge AI, data center, robotics, and autonomous vehicles. The lineup includes the Vera CPU, DGX Station for Windows, RTX Spark PCs, Cosmos 3, and a new Agent Toolkit. All major ODM partners—Foxconn, Quanta, Super Micro, and others—are positioned to fulfill demand.
NYSE is partnering with NVIDIA, HPE, and streaming platform Redpanda to rebuild its market infrastructure on AI-optimized silicon, including NVIDIA's Vera CPU and BlueField-4 STX data processing units. Financial exchanges are expected to accelerate AI-native infrastructure spending in H2 2026. Legacy market data vendors including Broadridge, SS&C, and FIS face competitive pressure as exchanges internalize more of their own infrastructure.
Supermicro released a coordinated multi-product ecosystem of DCBBS blueprints covering both the NVIDIA Vera Rubin NVL72 and HGX Rubin NVL8 platforms. The launch bundles Quantum-X800 InfiniBand networking, DLC-2 liquid cooling, software, and context memory storage into a single infrastructure stack. The simultaneous release signals the Rubin platform is moving from development into active production procurement.
Nvidia's Vera Rubin and China's Zhenwu V900/J900 represent diverging national bets on AI compute supremacy. ASIC design costs have nearly doubled since FinFET adoption in the mid-2010s, concentrating frontier development among well-capitalized players. A U.S. 2027 ban on Chinese-origin rare earth materials in defense systems is fracturing global supply chains further.
Netlist, positioning itself as an AI memory semiconductor player, relies on patent litigation settlements rather than product revenue to fund operations. A single adverse court ruling could cut off operating cash flow entirely. The model raises urgent questions about how litigation-driven strategies slow innovation across the AI hardware stack.
Ambiq Micro, a maker of ultra-low-power AI accelerators for IoT and edge devices, reported a net income loss of $38.35M while trading at a 20x price-to-sales ratio. The financial strain threatens R&D continuity at a company whose low-power chip architecture is central to edge AI deployment. Any revenue slowdown could trigger sharp multiple compression, cutting investment in next-generation silicon.