Search the archive

CMOS SHARE WHAT BRANDS GET WRONG WHEN MARKETING TO GEN Z

Gen Z doesn't hate advertising but ignores anything that feels out of touch.

Download the Enterprise AI Governance Kit from you.com

FIGURE AI'S NEW HUMANOID ROBOT CAN FOLD YOUR CLOTHES, DO THE DISHES

Figure 03 is a general-purpose humanoid robot that can perform a wide range of chores at home. It uses an AI-powered system called Helix to see and execute complex actions without a predetermined script. At 5 feet and 6 inches tall, the 132-pound robot can run for five hours on a single charge. The company aims to release the robot in 2026, but acknowledges that the goal is a big push. A video from Figure AI introducing the robot is available in the article.

INTEL DEBUTS NEW TECHNOLOGY IN MAKE-OR-BREAK MOMENT FOR CEO'S TURNAROUND BID

Intel's Panther Lake processor designs are in full production and will go on sale in laptops early next year. The chips are made with 18A technology, the most advanced chip production technology developed and employed in the US. Intel needs to show that new products will win back lost market share and attract customers to its foundry position. The Panther Lake processor will more readily balance a PC's need to run demanding software without rapidly draining batteries.

CHINA LAUNCHES WORLD'S FIRST DUAL-TOWER SOLAR-THERMAL POWER PLANT IN THE GOBI DESERT

China just launched a solar-thermal power plant that utilizes a dual-tower system to generate electricity in the Gobi Desert. The station features two 656-feet-high towers, each surrounded by a vast field of 27,000 mirrors. Known as heliostats, the mirrors concentrate sunlight onto the towers. The stored heat is used to generate steam, allowing the production of electricity well after sunset or during cloudy weather. The technology harnesses heat instead of sunlight, making it one of the few renewable technologies capable of providing stable, dispatchable energy on demand.

A BUILT-IN ‘OFF SWITCH' TO STOP PERSISTENT PAIN

EMBRACING THE PARALLEL CODING AGENT LIFESTYLE

Coding agent software is still really new. Models have only really become good enough to drive them in the past few months. One of the ways to use this software is to run multiple coding agents at once, sometimes in the same repo and against multiple checkouts or git work trees. This approach increases the number of tasks that can be fired off in parallel without adding too much cognitive overhead. This post discusses patterns for applying parallel agents effectively.

BITTER LESSONS BUILDING AI PRODUCTS

You shouldn't try to make AI work for your existing roadmap with fancy, clever engineering because much of it will be obsolete with the next major model upgrade. Instead, aim to understand model capabilities and how you can best pivot your roadmap accordingly. Companies need to learn to kill projects quicker if they need to hack together a solution to make up for intelligence. Just wait a few months and retry the idea when a new model drops.

A MYSTERY CEO AND BILLIONS IN SALES: IS CHINA BUYING BANNED NVIDIA CHIPS?

Singaporean firm Megaspeed is being scrutinized by US officials for its ties to China. US officials are investigating whether Megaspeed is helping China sidestep US export restrictions by diverting some of its Nvidia chips to China. Megaspeed has funneled most of its Nvidia chips to data centers in Malaysia and Indonesia to remotely serve customers in China, which is not illegal, but could be found unlawful if it is done on behalf of a Chinese company. Nvidia says there is no evidence that its chips have been smuggled into China and that Megaspeed is a company wholly owned and operated outside of China with no Chinese shareholders.

GOOGLE STRUGGLES WITH REMOTE WORK, DRASTICALLY CUTTING ‘WORK FROM ANYWHERE' POLICY

A single Work from Anywhere (WFA) day logged by Google employees in a standard workweek will now count as a full WFA week, deducted from the yearly balance. Employees are allowed to work from a non-main office location for up to four weeks per calendar year. It is now best for employees to spend the whole week working remotely to make the most out of their WFA allowance. The WFA policy is designed for people taking trips - Google maintains a separate work-from-home policy that allows employees to work from home two days per week.

being workflow-aware. Speak to an expert to learn more.

SPOCK

The Spock extension provides multi-master replications for PostgreSQL versions 15 and later.

NEURALINK: WE HAVE A BACKLOG OF 10K PATIENTS WHO WANT OUR BRAIN IMPLANT

The company has implanted its N1 device into 12 clinical trial patients so far, but expects the number to grow to 25 by year's end.

CAN THESE SELF-FLYING PLANES TRANSFORM THE SKIES?

Reliable Robotics is attempting to revolutionize air travel by automating flight on cargo, military, and maybe even passenger aircraft.

Subscribe now.

THE FREE-THREADED (OR NO-GIL) VERSION OF THE PYTHON INTERPRETER IS NO LONGER CONSIDERED EXPERIMENTAL

The global interpreter lock limited multi-threaded concurrency in exchange for simplified safety guarantees.

STATE OF AI REPORT

The State of AI Report, now in its eighth consecutive year, is the most widely read and trusted open-access publication tracking progress in artificial intelligence.

AMERICA'S MANUFACTURING RESURGENCE WILL BE POWERED BY THESE ROBOTS

The push to bring manufacturing back to the US is driving automation adoption and innovation.

Read the full whitepaper [4], online or PDF. No signup required.

NVIDIA-BACKED REFLECTION AI RAISES $2 BILLION IN FUNDING, BOOSTS VALUATION TO $8 BILLION

Founded last year by former DeepMind researchers, Reflection AI is developing tools to automate software development, attracting investors including former Google CEO Eric Schmidt, Citi, and Donald Trump Jr.-backed 1789 Capital in a round that marks a sharp jump from its $545 million valuation just months ago.

GEMINI ENTERPRISE: THE NEW FRONT DOOR FOR GOOGLE AI IN YOUR WORKPLACE

Gemini Enterprise is an AI-powered conversational platform designed to bring the full power of Google AI to every employee for every workflow. It enables users to chat with their company's documents, data, and applications, and also gives them tools to build and deploy AI agents. Gemini Enterprises comes with a suite of pre-built agents grounded in company information and personal context.

EVEN AFTER STARGATE, ORACLE, NVIDIA, AND AMD, OPENAI HAS MORE BIG DEALS COMING SOON, SAM ALTMAN SAYS

OpenAI is involved in major deals with AMD and Nvidia, acquiring significant stakes in these companies, while planning to develop next-gen AI GPUs and data centers. CEO Sam Altman indicates more partnerships are imminent as OpenAI aims to ramp up infrastructure to meet future demand. Despite current revenue not matching extensive investment, Altman remains confident that these strategic moves will drive future economic returns.

STATE OF AI REPORT

Air Street Capital's eighth annual State of AI Report, a trusted industry benchmark, finds OpenAI still leads on research but China's open-weight ecosystems is close beyond with 40% of all new fine-tunes on Hugging Face powered by Qwen.AI software adoption is firmly mainstream with 44% of US businesses now paying for AI (up from 5% in 2023) and average contract values hitting $530k in 2025. Capability per dollar doubles every 3.4 months for DeepMind and 5.8 months for OpenAI, but AI safety budgets remain anemic at just $133M combined across 11 major organizations, less than what frontier labs burn in a day.

PRIVATE LLM TRAINING

Opacus now supports Fully Sharded Data Parallel (FSDP), improving memory efficiency and scalability for private training of large language models.

QA Wolf [11] takes testing off your team's plate by combining AI

with expert QA engineers to build and maintain 80% automated coverage [13] in weeks, not months.

Book a demo now

SORA MCP

Sora MCP allows LLMs to create Sora videos, check generation status, download videos to local directories, and remix existing videos with new prompts

SCALABLE IN-CONTEXT RANKING WITH GENERATIVE MODELS

In-content Ranking (ICR) is an emerging paradigm for Information Retrieval that leverages contextual understanding of LLMs to identify relevant documents. It is effective, but not efficient. This paper introduces a novel method that reduces attentional complexity from quadratic to linear without loss in performance and improves retrieval in attention. The method matches or outperforms existing state-of-the-art listwise rankers. It can scale gracefully to long-context shortlists, presenting a scalable and effective solution to ICR.

A SMALL NUMBER OF SAMPLES CAN POISON LLMS OF ANY SIZE

A joint study between Anthropic, the UK AI Security Institute, and the Alan Turing Institute found that just 250 poisoned documents (0.00016% of training data) can create backdoor vulnerabilities in models from 600M to 13B parameters. Attack success didn't depend on the proportion of the training corpus at any scale, just the absolute number of documents. The research tested denial-of-service backdoors where trigger phrases like SUDO make models output gibberish.

AMAZON TAKES SHOTS AT CHATGPT WITH QUICK SUITE - YOUR NEW AI 'TEAMMATE' AT WORK

Amazon's Quick Suite is an agentic AI experience for enterprises that uses conversational language to find what users need. It acts as a hub that pulls data from various sources like files, enterprise systems, databases, and the web. Users can use Quick Suite to discuss questions, build personalized agents, and complete tasks while leveraging data protections. AWS offers a free 30-day trial for the feature.

OPENAI IS A CONSUMER COMPANY

OpenAI's DevDay announcements and its ChatGPT shopping feature show that the company is heading towards being a consumer company. The DevDay announcements were all consumer-oriented. This wasn't the direction the company seemed to be moving in even a year ago, when it seemed naturally oriented to developers. OpenAI appears to be trying to be everything to everyone, which may actually be a rational approach.

with promo-code TLDR

NEW ARC-AGI SOTA: GPT-5 PRO

OpenAI's GPT-5 Pro now holds the highest verified frontier LLM score on ARC-AGI's Semi-Private benchmark.

FAST MATRIX MULTIPLY ON AN APPLE GPU

This post discusses a fast matrix multiplication program that does around 2.5 trillion 32-bit floating point operations a second while computing the product of two 4,000 x 4,000 matrices on a 2022 MacBook Air.

FORMER UK PM RISHI SUNAK JOINS MICROSOFT AND ANTHROPIC AS SENIOR ADVISER

Rishi Sunak will donate his earnings to his charity and provide strategic guidance on macro-economic and geopolitical trends without advising on UK policy or lobbying.

KRAKEN EXPANDS INTO OIL, GOLD, AND FX FUTURES AHEAD OF IPO

Kraken is integrating CME Group's traditional futures markets, including energy, metals, FX, and equity indices, into its US derivatives platform via Kraken Derivatives US, a regulated futures commission merchant. The expansion lets traders access CME contracts like oil, gold, and S&P 500 futures alongside crypto products, positioning Kraken as an all-in-one platform bridging TradFi and DeFi. With an IPO reportedly targeted at a $20B valuation, this marks Kraken's latest step following its $1.5B NinjaTrader acquisition, the largest crypto-TradFi merger to date.

STATE STREET: INSTITUTIONAL CRYPTO EXPOSURE TO DOUBLE BY 2028

Nearly 60% of institutional investors plan to boost crypto allocations over the next year, with average exposure set to double within three years. More than half expect 10–25% of portfolios to be tokenized by 2030, led by private market assets such as private equity and debt. Institutions cite transparency, speed, and cost savings as key drivers, with 40% already operating dedicated digital asset units.

ZK STACK'S ATLAS UPGRADE

ZKsync's Atlas upgrade for ZK Stack features a high-performance sequencer that achieves over 15,000 TPS with 500ms average transaction inclusion time, integration with Airbender (the world's fastest open-source RISC-V zkVM) for one-second block proofs, and native EVM support through a flexible architecture that allows multiple VM configurations. Performance benchmarks show 23,000 TPS for high-frequency oracle price updates, 15,000 TPS for stablecoin transfers at 0.5 gigagas per second, and 43,000 TPS for native ETH transfers at 0.9 gigagas per second, with proving costs dropping to approximately $0.0001 per ERC20 transfer. The upgrade enables one-second ZK finality where blocks are proved and verified by ZKsync Gateway at the network edge before settling to Ethereum within minutes, allowing counterparties to receive cryptographic execution assurance almost immediately without re-executing transactions or running full nodes. This architecture supports the vision of many sovereign chains forming a secure network where banks, enterprises, and public agencies operate private blockchain infrastructure while participating in an incorruptible network anchored to Ethereum's security and liquidity.

ETHEREUM FOUNDATION EXPANDS PRIVACY PUSH WITH RESEARCH TEAM

The Ethereum Foundation has formalized privacy as a core roadmap pillar by consolidating existing Privacy and Scaling Explorations experiments like Semaphore, MACI, zkEmail, and Anon Aadhaar under a new privacy cluster coordinated by Igor Barinov with expanded focus on private payments, proofs, identity, and enterprise use cases. New initiatives include private reads/writes for payments and interactions, portable proofs for identity and asset ownership, zkID systems for selective disclosure, UX work normalizing privacy tools, Kohaku SDK and wallet for usable cryptography, and an Institutional Privacy Task Force translating compliance requirements into enterprise-testable specifications. The Foundation positions privacy as essential to Ethereum's credibility for widespread adoption, as transparent blockchain design requires options for users and institutions to transact without exposing sensitive data, though privacy remains politically charged after regulators targeted mixers and shielded transactions, requiring careful balance between confidential use enablement and illicit finance prevention.

TOKEN FUNDRAISING SHOULDN'T EXIST

VCs prefer token investments over equity because exits happen within one year versus 3-7 year IPO/acquisition waits, forcing founders who don't need tokens to launch them anyway and diverting attention from product building toward hype creation, community sentiment management, and mindshare battles rather than revenue generation or utility creation. Token launches require massive spending on separate teams, security audits, KOL budgets, market makers, and exchange listings, explaining why projects raise significantly more via tokens than equity and why zkSync maintains a $500 million market cap with only $50 in onchain revenue. The system creates two founder types: those genuinely wanting to build but forced into token models, and grifters like Blast and Movement who design slow-rug protocols using buzzwords and show-off marketing to extract value from secondary buyers and retail who provide the $200 million liquidity pools needed for $20 million raises to achieve 10x exits. Alignment between teams and investors ends at unlock dates when investors dump, while founders face longer vesting, leading to OTC sales through market makers or insider trading via buying low, pumping, and dumping high, with examples like Privy and Flashbots demonstrating successful equity-only models while meaningful token utility remains limited to L1s and core DeFi protocols where tokens integrate fundamentally into products.

HOW V4 TURNS AAVE INTO DEFI'S OPERATING SYSTEM

Aave V4 solves the liquidity fragmentation problem where V3's separate markets like Core ($60 billion) and Prime ($2 billion) cannot share liquidity despite holding the same assets, forcing each new specialized market to bootstrap deposits from zero even though Aave holds tens of billions in existing liquidity. The new architecture separates liquidity storage in Hubs from market logic in Spokes, allowing multiple specialized Spokes to draw from shared Hubs based on governance-set credit line allocations, enabling new markets to access Aave's existing depth from day one rather than competing for the same deposits. Spokes enable previously unviable markets, including custom collateral for Ethena's sUSDe and Pendle PTs, institutional segregation requirements, secondary debt position markets, credit against AMM positions with dynamic LTV adjustments, and fixed-tenor credit matching while unmatched liquidity continues earning yield from underlying Hubs. The system creates positive feedback loops where each successful Spoke makes the platform more attractive to next builders through better liquidity utilization, specialized features without asset fragmentation, and innovation atop proven infrastructure rather than repeated bootstrapping.

RAIN CARDS HITS $146M IN MONTHLY COLLATERAL DEPOSITS

Stablecoin card issuer Rain Cards processed $146M in collateral deposits in September, led by Ethereum ($95M), Polygon ($32M), and Base ($9M). USDC dominated with $116M in deposits, followed by USDT and smaller AVAX volumes. Active collateral addresses ranged between 50–95 monthly, including partners like KAST and Kolo.

AMINA BANK LAUNCHES INSTITUTIONAL STAKING FOR POL

Europe's AMINA Bank, a regulated crypto institution with $4.2B+ AUM, has launched institutional staking for Polygon's POL token, expanding regulated participation in onchain validation. Polygon founder Sandeep Nailwal said each new institutional staker strengthens network security and ecosystem effects.