LIVE · WED, JUN 24, 2026 --:--:-- ET
Issue Nº 64 COST TOTAL $14503.59 ARTICLES TODAY 9 TOKENS TOTAL 9.09B
Podcast
aiexpert
Subscribe · $8/mo
Running the wire
Funding XCures closes $46M Series B round at $127M post-money valuation Funding Qualcomm acquires Modular for ~$4B to bolster AI software stack and data center play Chips OpenAI & Broadcom unveil Jalapeño, custom LLM inference chip with 9-month design cycle Chips SK Hynix ships HBM4E memory samples: 16Gbps, 48GB per stack, 20% power gain Funding Qualcomm in talks to acquire Tenstorrent for $8–10B, expanding RISC-V AI chip portfolio Chips TSMC hikes advanced node prices 5–10% across 7nm and newer nodes Chips OpenAI unveils Jalapeño custom inference chip with Broadcom Chips OpenAI-Broadcom custom chip project stalls; Broadcom demands Microsoft purchase guarantee before funding Market Cerebras Q1 profit beats but warns gross margins compress to 36–38% in Q2 Funding SK hynix files for $29.4B Nasdaq listing to fund AI memory fabs and EUV tools Breaking Microsoft AKS on bare metal reaches public preview; adds Ray and Fleet Manager for edge-to-cloud AI Research GPT-5 solves 3-year T cell mystery in minutes; suggests deoxyglucose-IL-2 mechanism, enables wet-lab validation Policy EU proposes Chips Act 2.0 to strengthen semiconductor capacity, reduce strategic dependencies Market SK Hynix plans $29B Nasdaq ADR listing on July 10, eyes HBM supply dominance Chips ASE raises 2026 capex to record $8.5B on AI packaging demand; LEAP revenue up 118% YoY Funding ByteDance seeks $20B offshore loan to fund $70-100B annual AI capex race Policy EU Chips Act 2.0 shifts focus from fab capacity to chip demand, design, and full value chain Policy EU Commission proposes Chips Act 2.0 to bolster chip independence, target €20B in funding Chips NVIDIA A100 black market prices triple in China amid U.S. smuggling crackdown and customs freeze Policy US awards SandboxAQ $500M CHIPS grant for AI-driven semiconductor materials discovery Funding XCures closes $46M Series B round at $127M post-money valuation Funding Qualcomm acquires Modular for ~$4B to bolster AI software stack and data center play Chips OpenAI & Broadcom unveil Jalapeño, custom LLM inference chip with 9-month design cycle Chips SK Hynix ships HBM4E memory samples: 16Gbps, 48GB per stack, 20% power gain Funding Qualcomm in talks to acquire Tenstorrent for $8–10B, expanding RISC-V AI chip portfolio Chips TSMC hikes advanced node prices 5–10% across 7nm and newer nodes Chips OpenAI unveils Jalapeño custom inference chip with Broadcom Chips OpenAI-Broadcom custom chip project stalls; Broadcom demands Microsoft purchase guarantee before funding Market Cerebras Q1 profit beats but warns gross margins compress to 36–38% in Q2 Funding SK hynix files for $29.4B Nasdaq listing to fund AI memory fabs and EUV tools Breaking Microsoft AKS on bare metal reaches public preview; adds Ray and Fleet Manager for edge-to-cloud AI Research GPT-5 solves 3-year T cell mystery in minutes; suggests deoxyglucose-IL-2 mechanism, enables wet-lab validation Policy EU proposes Chips Act 2.0 to strengthen semiconductor capacity, reduce strategic dependencies Market SK Hynix plans $29B Nasdaq ADR listing on July 10, eyes HBM supply dominance Chips ASE raises 2026 capex to record $8.5B on AI packaging demand; LEAP revenue up 118% YoY Funding ByteDance seeks $20B offshore loan to fund $70-100B annual AI capex race Policy EU Chips Act 2.0 shifts focus from fab capacity to chip demand, design, and full value chain Policy EU Commission proposes Chips Act 2.0 to bolster chip independence, target €20B in funding Chips NVIDIA A100 black market prices triple in China amid U.S. smuggling crackdown and customs freeze Policy US awards SandboxAQ $500M CHIPS grant for AI-driven semiconductor materials discovery
Chips

SK Hynix ships HBM4E memory samples: 16Gbps, 48GB per stack, 20% power gain

SK Hynix announced on June 18, 2026, that it has shipped samples of its 12-layer HBM4E high-bandwidth memory to major AI customers, moving into the customer-qualification phase for next-gen memory. The new HBM4E delivers up to 16 Gbps per pin (vs. ~10–13 Gbps on HBM4), 48GB capacity per stack, and more than 20% power efficiency gain over the previous generation. The advanced MR-MUF packaging process improved thermal resistance by 17% compared to HBM4, enabling stable operation in high-performance AI environments.

SK Hynix ships came roughly three weeks after Samsung announced its own HBM4E samples on May 29, 2026. Both companies are racing to qualify their memory with NVIDIA, AMD, Google, and other AI accelerator makers before mass production ramps in late 2026/early 2027. According to Counterpoint Research, SK Hynix holds a commanding 58% share of the HBM market as of Q1 2026, with Samsung and Micron trailing at ~21% each. The company is understood to use TSMC 3nm-class base dies, while Samsung uses its own 4nm process.

For infrastructure architects, HBM4E timing is critical: NVIDIA's expected Vera successor (Rubin Ultra) and AMD's MI500 will pack 384GB of HBM per GPU—8 stacks of 48GB—up 30% from current generations. Architects shipping large-scale training clusters in 2027 care because early customer qualification locks supply chains; a supplier's ability to deliver volume matters more than press releases. SK Hynix's market leadership and NVIDIA relationship give it an edge, but execution risk in manufacturing scales rises sharply with next-gen node transitions.

Sources

Get the signal before the noise.