S&P 5007,473.47↑ 0.37%
NASDAQ26,343.97↑ 0.19%
BTC$77,130.08↓ 0.18%
ETH$2,108.46↓ 0.10%
ASUS COST CUT70%↓ inference
EY x MSFT$1B· DEAL
UK HEAT RECORD34.8°C↑ record
Lead Story
Hardware
ASUS Pressroom
ASUS announced a hybrid agentic AI architecture across its ExpertBook, ExpertCenter, and NUC product lines on May 25. Integrating Phison's aiDAPTIV technology, the system dynamically routes workloads between on-device and cloud inference, claiming up to 70% reduction in token costs. The approach...
Continue Reading
AI
Zhipu AI released GLM-5.1 as a fully open-source model at 754 billion parameters, intensifying the race to commoditize frontier-scale reasoning.
World
The UK recorded its hottest May day in history at 34.8°C as climate records continue to fall across Europe.
Markets
Fear dominates sentiment as investors weigh geopolitical uncertainty from Iran talks, Kyiv strikes, and an Ebola outbreak against a relentless pace of AI product launches.
Wild Card
Anthropic co-founder Chris Olah publicly responded to Pope Leo XIV's encyclical on artificial intelligence, marking a rare direct exchange between a frontier lab leader and the Catholic Church on the ethics of machine cognition.
Bitcoin
$$77,130.08
-0.18%
Today's Focus
01
Open-Source Giants Go Mainstream
GLM-5.1's release at 754 billion parameters represents the largest fully open-source model to date, putting frontier-grade capabilities within reach of any organization with sufficient compute. This accelerates a trend where proprietary model advantages shrink quarter over quarter. Combined with ASUS's hybrid architecture slashing inference costs by 70%, the economics of deploying large models are shifting faster than most enterprise procurement cycles can adapt.
02
On-Device AI Heats Up
Apple is deepening its on-device AI strategy for 2026, signaling that the next competitive battleground is not cloud scale but local inference quality. ASUS's hybrid AI architecture, which routes workloads between edge and cloud to cut costs 70%, points in the same direction. The convergence suggests a future where the most valuable AI runs silently on hardware you already own, not on a distant server.
03
Faith Meets Frontier Models
Chris Olah's public response to the papal encyclical on AI is more than a curiosity; it reflects growing pressure on lab leaders to articulate moral frameworks, not just capability benchmarks. The Vatican's intervention gives institutional weight to questions about consciousness, labor displacement, and autonomy that the AI industry has largely deferred. As models scale past 750 billion parameters and synthetic training data multiplies, the philosophical stakes are catching up to the technical ones.
Podcast
~22 min
Video
Video Briefing
Infographic
Intelligence Map
Newsletter Intelligence
26 May 2026
- •Waymo pauses robotaxi service in four cities after vehicles repeatedly drove into floodwaters
- •Hugging Face releases a $2,500 3D-printable humanoid robot
- •Robot butlers undergo first real-world testing in China
Read The Rundown AI →
26 May 2026
- •Claude rapidly expands across corporate finance with Anthropic launching finance-focused agents for reconciliations, valuations, and audits
- •Farther raises $150M Series D led by General Atlantic to scale AI-powered wealth management platform
- •Blockchain.com files confidentially for US IPO amid renewed crypto listing momentum
Read TLDR →
26 May 2026
- •Anthropic prepares Mythos 1 for Claude Code and Claude Security, with traces already surfacing on Google Cloud and AWS; Claude Opus 4.8 also rumored to be in the works
- •DeepSeek made its 75% discount on V4 Pro permanent, undercutting OpenAI GPT-5, Anthropic Claude Opus 4.7, and Google Gemini 3.5 Flash in a major AI price war escalation
- •Only 15% of organizations have the data foundation needed for agentic AI at scale, according to Fivetran's agentic AI readiness index
Read TLDR →
26 May 2026
- •Polsia hits $250M valuation with zero employees, raising $30M for its autonomous AI business platform
- •DeepSeek permanently slashes V4 Pro pricing by 75%, undercutting frontier rivals by at least 34x below GPT-5.5
- •Google's new AI-native Search struggles to define words like 'disregard,' 'stop,' and 'ignore'
Read Superhuman →
26 May 2026
- •The AI Bifurcation of Tech: Application companies must ask whether AI agents will still need them in three years, as custom workflows are easily routed around by AI
- •AI Accenture, Not Accenture for AI: The winning AI services model is a product company that delivers a service, building an internal agent and context layer to serve one narrow customer cheaper and faster than anyone else
- •Lessons from Running Long-Running Coding Agents: Simon Last at Notion runs coding agents on single sessions for days or weeks, retaining learned patterns from day one
Read TLDR →
26 May 2026
- •Apple's anniversary 'iPhone XX' leaks in renders, targeting a radically redesigned 2027 debut
- •Google launches Stitch, a real-time AI-collaborative design tool with live canvas streaming
- •Replit releases Agent 4 after Apple lifts four-month App Store ban
Read TLDR →
Forward this to one operator you work with. Your referral link is in every email; milestones at koda.community/refer.