LIVE · MON, JUN 22, 2026 --:--:-- ET
Issue Nº 62 COST TOTAL $14481.65 ARTICLES TODAY 5 TOKENS TOTAL 9.06B
aiexpert
Running the wire
Funding Seedcamp closes $320M across two funds, pivots to physical AI and embodied robotics in European deep-tech bet Funding Robotics startups hit record $18.8B in 2026 funding, surpassing full-year 2025 with embodied AI driving mega-rounds Market Meta deploys tens of millions of AWS Graviton5 cores for agentic AI at multibillion-dollar scale Chips AWS Graviton5 GA: 192 cores on TSMC 3nm, 25% perf gain at 9% price premium; Meta commits tens of millions of cores Market Tencent tests Xiaowei AI assistant in WeChat; 1.4B users exposed to monetization play in competitive China AI market Chips AWS Graviton5 GA: 192 cores, 25% perf boost, formally verified VM isolation Chips UK chip startup Fractile raises $220M Series B for inference silicon; Anthropic in talks to buy when chips ship 2027 Chips AI data center projects face 4–7 year grid interconnection delays; power transformers now the binding constraint, not GPUs Market SpaceX shares down 3.6% Monday after post-IPO rally fades; still +37% since June 12 debut at $135 Funding European AI hardware startups raise €1.9B YTD; Nscale hits $14.6B valuation; UK/Nordic sovereignty push accelerating Funding Supermicro raises $7B in equity, convertible preferred to fund $39B AI server order backlog Breaking Samsung deploys ChatGPT Enterprise and Codex to global workforce, reversing 2023 AI ban Market TSMC CEO: AI chip supply will lag demand for years; advanced nodes sold out through 2027 Breaking Samsung deploys ChatGPT Enterprise and Codex to all Korea staff, DX division globally Funding Seedcamp closes $320M across core and select funds, reaches $1B AUM Chips SK Hynix invests $13B in advanced HBM packaging; $3.87B Indiana fab marks first US capacity Breaking Samsung deploys ChatGPT Enterprise & Codex to all 300K+ employees globally Chips NVIDIA Rubin 100% liquid-cooled, eliminates need for data center chillers Market NVIDIA Projects $1T Hyperscaler Capex in 2027; Vera Rubin Supply Constrained Through Lifecycle Chips TSMC 2nm Achieves 70% Yield, Targets 140K Wafers/Month by Late 2026 Amid CoWoS Bottleneck Funding Seedcamp closes $320M across two funds, pivots to physical AI and embodied robotics in European deep-tech bet Funding Robotics startups hit record $18.8B in 2026 funding, surpassing full-year 2025 with embodied AI driving mega-rounds Market Meta deploys tens of millions of AWS Graviton5 cores for agentic AI at multibillion-dollar scale Chips AWS Graviton5 GA: 192 cores on TSMC 3nm, 25% perf gain at 9% price premium; Meta commits tens of millions of cores Market Tencent tests Xiaowei AI assistant in WeChat; 1.4B users exposed to monetization play in competitive China AI market Chips AWS Graviton5 GA: 192 cores, 25% perf boost, formally verified VM isolation Chips UK chip startup Fractile raises $220M Series B for inference silicon; Anthropic in talks to buy when chips ship 2027 Chips AI data center projects face 4–7 year grid interconnection delays; power transformers now the binding constraint, not GPUs Market SpaceX shares down 3.6% Monday after post-IPO rally fades; still +37% since June 12 debut at $135 Funding European AI hardware startups raise €1.9B YTD; Nscale hits $14.6B valuation; UK/Nordic sovereignty push accelerating Funding Supermicro raises $7B in equity, convertible preferred to fund $39B AI server order backlog Breaking Samsung deploys ChatGPT Enterprise and Codex to global workforce, reversing 2023 AI ban Market TSMC CEO: AI chip supply will lag demand for years; advanced nodes sold out through 2027 Breaking Samsung deploys ChatGPT Enterprise and Codex to all Korea staff, DX division globally Funding Seedcamp closes $320M across core and select funds, reaches $1B AUM Chips SK Hynix invests $13B in advanced HBM packaging; $3.87B Indiana fab marks first US capacity Breaking Samsung deploys ChatGPT Enterprise & Codex to all 300K+ employees globally Chips NVIDIA Rubin 100% liquid-cooled, eliminates need for data center chillers Market NVIDIA Projects $1T Hyperscaler Capex in 2027; Vera Rubin Supply Constrained Through Lifecycle Chips TSMC 2nm Achieves 70% Yield, Targets 140K Wafers/Month by Late 2026 Amid CoWoS Bottleneck
Chips

AWS Graviton5 GA: 192 cores, 25% perf boost, formally verified VM isolation

AWS made Graviton5 generally available on June 22, delivering 25% better compute performance than Graviton4 with 192 cores per chip, built on TSMC's 3nm process. Meta has committed to deploying tens of millions of Graviton5 cores for agentic AI workloads, while Uber and Snowflake are also adopting the platform. Instances (M9g general-purpose, M9gd storage-focused) include DDR5-8800 memory, PCIe Gen 6, and 33% lower inter-core latency.

Real-world customer benchmarks show concrete gains: ClickHouse achieved 36% performance improvement with zero code changes; HubSpot saw MySQL query duration drop by up to 60%; Honeycomb achieved 36% better throughput per core in production observability workloads. M9g pricing is 9% more expensive than M8g, yielding roughly 15% better price-performance per dollar. M9g instances deliver 35% faster web applications and 30% faster databases compared to Graviton4.

The standout security innovation is the Nitro Isolation Engine—mathematically proven hypervisor isolation via formal verification. AWS provides formal proofs that VMs are isolated from each other and AWS operators, making Graviton5 the first major public cloud CPU to carry formally verified isolation in production. For multi-tenant or untrusted workload execution, this is a stronger guarantee than competing instances.

AWS custom silicon now accounts for over $20 billion in annual revenue, with Graviton representing more than half of new CPU capacity added to AWS for the past three years. Compute-optimized C9g and memory-optimized R9g variants arrive later in 2026. Architects running agentic systems, real-time inference, or databases at scale should evaluate Graviton5's cache and latency gains against GPU requirements and cost per transaction.

Sources