LIVE · WED, JUL 29, 2026 --:--:-- ET
Issue Nº 99 COST TOTAL $14994.73 ARTICLES TODAY 11 TOKENS TOTAL 9.72B
Podcast
aiexpert
Subscribe · $8/mo
Running the wire
Breaking LangChain Deep Agents v0.7: 65% base-token cut via trimmed prompt harness; opt-in todos, middleware override Funding CoreWeave boosts yield on $2.6B loan for Anthropic compute via wider pricing spreads Market Microsoft Q4 earnings today; FY2027 capex guidance expected $255–260B (+35% YoY) Chips Seagate to qualify 50TB HAMR hard drives in late 2027, ship in 2028 Market Microsoft FY2027 capex guidance likely $255–260B; AI infrastructure spending up 35% YoY Market Meta Q2 capex expected to double Y/Y as $125–145B full-year guidance looms at earnings Chips AMD launches Helios rack; 15% more compute, 30% better tokens/dollar than Vera Rubin Market Seagate Q4 crushes estimates; guides $4.1B Q1 revenue on AI storage demand Market NVIDIA raises RTX GPU prices 20–30% in third hike of 2026; GDDR shortage expected to persist until 2028 Chips NVIDIA employee detained in Taiwan over Supermicro smuggling scheme; GPU export diversion persists despite whitelist Market Apple hits $5T valuation; Tim Cook's final earnings call faces memory crisis + AI strategy pressure Chips Lenovo doubles Whitsett NC AI server facility to 890K sq ft; targets 1,400 nodes/day capacity Market Meta partners BlackRock on $14B El Paso data center; BlackRock owns 80%, Meta operates Market Bloom Energy crushes Q2, raises FY guidance to $3.9–4.2B on AI data center fuel cell demand Funding Centralize raises $15M Series A to map enterprise deal relationships with AI Market Asian semiconductor stocks plunge; SK Hynix misses estimates despite record profit Policy xAI sues Minnesota over AI nudification ban; law takes effect Aug 1 Policy Judge approves Anthropic $1.5B copyright settlement; authors get ~$3,100 per work Chips AMD, Cerebras team on disaggregated AI inference; claim 5x efficiency gains Funding Lattice Semiconductor completes AMI acquisition for $650M in stock Breaking LangChain Deep Agents v0.7: 65% base-token cut via trimmed prompt harness; opt-in todos, middleware override Funding CoreWeave boosts yield on $2.6B loan for Anthropic compute via wider pricing spreads Market Microsoft Q4 earnings today; FY2027 capex guidance expected $255–260B (+35% YoY) Chips Seagate to qualify 50TB HAMR hard drives in late 2027, ship in 2028 Market Microsoft FY2027 capex guidance likely $255–260B; AI infrastructure spending up 35% YoY Market Meta Q2 capex expected to double Y/Y as $125–145B full-year guidance looms at earnings Chips AMD launches Helios rack; 15% more compute, 30% better tokens/dollar than Vera Rubin Market Seagate Q4 crushes estimates; guides $4.1B Q1 revenue on AI storage demand Market NVIDIA raises RTX GPU prices 20–30% in third hike of 2026; GDDR shortage expected to persist until 2028 Chips NVIDIA employee detained in Taiwan over Supermicro smuggling scheme; GPU export diversion persists despite whitelist Market Apple hits $5T valuation; Tim Cook's final earnings call faces memory crisis + AI strategy pressure Chips Lenovo doubles Whitsett NC AI server facility to 890K sq ft; targets 1,400 nodes/day capacity Market Meta partners BlackRock on $14B El Paso data center; BlackRock owns 80%, Meta operates Market Bloom Energy crushes Q2, raises FY guidance to $3.9–4.2B on AI data center fuel cell demand Funding Centralize raises $15M Series A to map enterprise deal relationships with AI Market Asian semiconductor stocks plunge; SK Hynix misses estimates despite record profit Policy xAI sues Minnesota over AI nudification ban; law takes effect Aug 1 Policy Judge approves Anthropic $1.5B copyright settlement; authors get ~$3,100 per work Chips AMD, Cerebras team on disaggregated AI inference; claim 5x efficiency gains Funding Lattice Semiconductor completes AMI acquisition for $650M in stock
Chips

AMD launches Helios rack; 15% more compute, 30% better tokens/dollar than Vera Rubin

AMD officially launched Helios, its first rack-scale AI system, at Advancing AI 2026 in San Francisco on July 23. The system combines 72 AMD Instinct MI455X GPUs with 6th-gen EPYC Venice CPUs and Pensando networking into a double-width rack (4OU) delivering 2.9 exaflops of FP4 inference performance. AMD claims Helios delivers 15% better compute performance and 50% greater high-bandwidth memory (HBM4) capacity than NVIDIA's Vera Rubin NVL72, while offering 30% more inference tokens per dollar.

Helios is priced between $5–5.5 million per rack and enters production immediately, with early customer deployments beginning in Q4 2026. Major customers announced include Microsoft Azure, Meta (1 GW committed to Helios), OpenAI (12 GW of AMD GPUs across OpenAI and Meta), Oracle, Anthropic (2 GW partnership with engineering collaboration to use Claude for ROCm optimization), and Tata Consultancy Services. Each rack provides 31 terabytes of HBM4 memory and 1.7 petabytes/second of aggregate bandwidth.

AMD's move directly challenges NVIDIA's >95% data center GPU market dominance; AMD currently holds ~4.5%. Helios represents AMD's most integrated play yet—bunching silicon, networking, and software with strategic partnerships on inference workloads (via Cerebras) and model optimization (via Anthropic, OpenAI). For architects planning multi-vendor AI infrastructure and evaluating total cost of ownership over time, AMD Helios signals credible rack-level competition in 2026–2027, with supply chains now diversifying beyond NVIDIA monopoly.

Sources

Get the signal before the noise.