LIVE · THU, JUL 09, 2026 --:--:-- ET

Issue Nº 79 COST TOTAL $14740.92 ARTICLES TODAY 10 TOKENS TOTAL 9.40B

Running the wire

Chips Intel Preps Nova Lake-S Dunlow Workstation CPUs With Up to 28 Cores at 95W TDP Market Meta Launches Muse Spark 1.1 Paid API; Charges $1.25/$4.25 Per Million Tokens Breaking OpenAI Rolls Out GPT-5.6 Sol, Terra, Luna With 54% Coding Token Efficiency Funding Mercor acquires Deeptune: AI agent simulation platform; $2B ARR, Mag Seven as customers Market GeForce NOW expands Toronto region with RTX 5080 servers; Blackwell rollout now complete globally Funding Kraken Technology becomes Europe's maritime defence unicorn—$175M Series B from NATO, defence investors Policy OpenAI publishes National Security Principles; expands Daybreak cyber access to 9 allied nations Chips Hugging Face transformers backend now reaches native vLLM speed on Qwen3 MoE, dense models Research NVIDIA releases Nemotron Post-Training v3 Prompt Atlas—10T tokens, synthetic personas for agentic AI Breaking Google AlloyDB ships proxy models for semantic queries—23,000x faster, 6,000x cheaper Market SK Hynix Debuts on Nasdaq With $1 Trillion Valuation; Raises $29B for U.S. Expansion Breaking HalluSquatting attack tricks AI agents into running malicious code via hallucinated repos; 85% success rate Funding Europe VC funding hits $24B in Q2 2026—strongest quarter in 4 years as UK leads Breaking SpaceXAI releases Grok 4.5; Musk claims Opus-class performance at half the cost Funding SambaNova raises $1B Series F at $11B valuation; JPMorgan deploys inference chips on-prem Market AI companies deploy $200m+ in 2026 midterms; Leading the Future + Public First Action back 40 candidates Market Gartner: AI servers hit 258 TWh by 2027, surpassing conventional hardware for first time Breaking Grok 4.5 launches at $2/$6 per million tokens, undercutting Opus 4.8 on cost and speed Chips Imec roadmap: 2.5D optical I/O cuts AI inference interconnect power from 1.25kW to <200W Breaking SpaceXAI releases Grok 4.5 at $2/$6 per MTok; claims parity with Opus 4.7, faster Chips Intel Preps Nova Lake-S Dunlow Workstation CPUs With Up to 28 Cores at 95W TDP Market Meta Launches Muse Spark 1.1 Paid API; Charges $1.25/$4.25 Per Million Tokens Breaking OpenAI Rolls Out GPT-5.6 Sol, Terra, Luna With 54% Coding Token Efficiency Funding Mercor acquires Deeptune: AI agent simulation platform; $2B ARR, Mag Seven as customers Market GeForce NOW expands Toronto region with RTX 5080 servers; Blackwell rollout now complete globally Funding Kraken Technology becomes Europe's maritime defence unicorn—$175M Series B from NATO, defence investors Policy OpenAI publishes National Security Principles; expands Daybreak cyber access to 9 allied nations Chips Hugging Face transformers backend now reaches native vLLM speed on Qwen3 MoE, dense models Research NVIDIA releases Nemotron Post-Training v3 Prompt Atlas—10T tokens, synthetic personas for agentic AI Breaking Google AlloyDB ships proxy models for semantic queries—23,000x faster, 6,000x cheaper Market SK Hynix Debuts on Nasdaq With $1 Trillion Valuation; Raises $29B for U.S. Expansion Breaking HalluSquatting attack tricks AI agents into running malicious code via hallucinated repos; 85% success rate Funding Europe VC funding hits $24B in Q2 2026—strongest quarter in 4 years as UK leads Breaking SpaceXAI releases Grok 4.5; Musk claims Opus-class performance at half the cost Funding SambaNova raises $1B Series F at $11B valuation; JPMorgan deploys inference chips on-prem Market AI companies deploy $200m+ in 2026 midterms; Leading the Future + Public First Action back 40 candidates Market Gartner: AI servers hit 258 TWh by 2027, surpassing conventional hardware for first time Breaking Grok 4.5 launches at $2/$6 per million tokens, undercutting Opus 4.8 on cost and speed Chips Imec roadmap: 2.5D optical I/O cuts AI inference interconnect power from 1.25kW to <200W Breaking SpaceXAI releases Grok 4.5 at $2/$6 per MTok; claims parity with Opus 4.7, faster

Breaking Thursday, July 9, 2026 at 01:04 PM

Google AlloyDB ships proxy models for semantic queries—23,000x faster, 6,000x cheaper

Google announced general availability of AlloyDB AI functions, PostgreSQL-compatible database operations for semantic search, filtering, and ranking alongside a two-phase proxy model pattern that addresses the per-row LLM cost problem at scale. The proxy model inverts the typical database-to-LLM relationship: instead of calling an external model for every row decision, the database learns from a sample using a frontier model, then applies that judgment locally at database speed.

The architecture works in two phases. Phase 1: a PREPARE statement sends a data sample to a frontier model (e.g., Vertex AI) to train a lightweight local proxy. Phase 2: the query executes using the local proxy at database speed, with fallback to the frontier model if confidence is too low. For ai.if semantic filtering in preview, Google reports throughput of 100,000 rows per second with proxy models, compared to row-at-a-time processing of ~4 rows per second—a 23,000x improvement. The cost reduction reaches 6,000x by eliminating per-row API calls and prompt overhead.

The release also adds smart batching (GA for ai.if and ai.rank), grouping multiple rows into single model calls and delivering 2,400x throughput improvement over row-at-a-time. AlloyDB now includes ai.generate, ai.if, ai.rank, ai.forecast, ai.summarize, ai.agg_summarize, and ai.analyze_sentiment. A managed MCP server for AlloyDB lets agents query databases through Model Context Protocol without teams running custom infrastructure.

For platform teams, the distillation-at-query-time pattern is architecturally significant. It applies beyond AlloyDB: any database calling external models per-row faces the same cost and latency wall. AlloyDB positions itself as a PostgreSQL-compatible database where structured queries, semantic search, vector search (up to 10 billion vectors via ScaNN), and LLM-powered analysis coexist. Note: the 23,000x and 6,000x figures are from internal testing of ai.if in preview. Run benchmarks on your own data before committing to production.

Source: infoq.com →