Sakhanda Wire
NVDA $150.96 -2.89% MSFT $508.29 +0.40% GOOGL $282.44 +0.80% META $616.38 -0.40% AMZN $267.74 +3.82%

Latest news

Older stories

TechCrunch

Maven Robotics raises $100 million for warehouse automation

Maven Robotics has emerged from stealth with a $100 million Series A led by RoboStrategy, LocalGlobe, Vine Ventures and XTX Markets Ventures. The startup plans to produce 250 third-generation robots and begin designing a fourth-generation platform.

The Verge

AI backlash exposes limits beyond software, Nilay Patel says

In a mailbag episode of Decoder, Nilay Patel and producers Nick Statt and Greg Ott discussed the strong reaction to Patel’s “software brain” video and the broader backlash against artificial intelligence. Patel said the current AI excitement is closely tied to systems’ growing ability to write software and manipulate databases, where results can be checked by running code through a compiler. He argued that verification becomes much harder in fields such as novel drug development, and warned against treating software as a universal model for the world.

The Decoder

Deepseek V4.1-Flash reduces memory demands for AI agents

Deepseek has released V4.1-Flash, a multimodal model designed to lower the cost of processing long contexts. The model has 552 billion parameters and supports contexts of up to one million tokens. Deepseek says its KV cache requires about one quarter of the fast-memory space used by Deepseek-V4-Flash, while the portion stored on SSD or host memory is reduced to roughly one eighth. Compared with Deepseek-V1, the global cache size per token is 437 times smaller.

The Decoder

Meta introduces Muse, an AI agent operated through WhatsApp

Meta has introduced Muse, an AI agent that users control through WhatsApp. The company says it can perform tasks such as booking travel, sending emails and shopping by opening a browser, completing forms and negotiating on a user’s behalf. For broader goals, Muse is designed to plan steps and manage time and resources. It can continue working when the app is closed and ask for approval before sending an email or completing a purchase.

The Decoder

Nvidia and Palantir plan AI-run supply chains

Nvidia and Palantir are working on an artificial intelligence system for managing supply chains, with Nvidia’s own operations set to become its first deployment. The company’s network covers millions of parts, thousands of suppliers and partners around the world. Nvidia says a single Vera Rubin rack includes 1.3 million components.

The Verge

Mathematician asks OpenAI to prove it did not use his work

Mathematician Andreas Thom is challenging OpenAI over the origins of data behind the company’s increasingly notable mathematical results. In a series of posts on Mastodon, Thom questioned whether conversations that he and colleagues had with ChatGPT before OpenAI’s public announcement may have contributed to the system’s performance.

MIT Tech Review

AI data centers expose a new grid architecture challenge

A transmission-line fault in Ashburn, Virginia, on July 22, 2026, abruptly removed more than 3 gigawatts of data-center demand from the grid. A similar incident two years earlier followed the failure of a single surge arrester and disconnected roughly 60 Virginia facilities, representing 1,500 megawatts. The incidents highlighted a reliability risk caused not by insufficient generation, but by many large facilities responding to disturbances in the same way.

The Decoder

AI extinction warnings reach US mainstream media

Jacob Coxon, a departing Anthropic researcher, has brought warnings about self-improving artificial intelligence to a broad US audience. He said on CNN that such systems could pose an existential threat to humanity, and his views were also discussed on Fox News. Several US politicians addressed the issue on social media, while Joe Rogan dedicated an episode to AI safety.

AI News

JD.com launches plan to expand physical AI in logistics

JD.com has launched a Physical AI Acceleration Plan to expand robotics and artificial intelligence across its logistics network. Announced at JDDiscovery 2026 in Beijing, the programme reiterates a five-year procurement target of 3 million robots, 1 million autonomous vehicles and 100,000 delivery drones. JD Logistics also introduced its industrial Wolf Robot series for warehousing, sorting, transport and delivery, including systems designed for temperatures down to minus 20 degrees Celsius.

The Decoder

Top AI-Spending Companies Cut Per-Employee Costs in August

The September 2026 Ramp AI Index reports that median AI spending per employee among the top 1 percent of US companies fell 9.7 percent in August, reaching $7,205. Spending by the top 10 percent and by the median company continued to rise. Ramp said the top group is a small and therefore more volatile sample that could be revised later. Some of the August decline may reflect seasonal vacations among engineers, according to chief economist Ara Kharazian.

MarkTechPost

DeepSeek-V4.1-Flash cuts cache demands for million-token contexts

DeepSeek AI has released DeepSeek-V4.1-Flash, a multimodal Mixture-of-Experts model with 552B backbone parameters, 196B Engram parameters and a 1M-token context window. It activates 8B parameters per token during prefill and 16B during decoding. The reported global KV-cache footprint is 890 bytes per token, about one quarter that of DeepSeek-V4-Flash and roughly 437 times below DeepSeek-V1.

MarkTechPost

LandingAI launches ADE Gen2 with DPT-3 document parsing models

LandingAI has released Agentic Document Extraction (ADE) Gen2, a rebuild of its document-intelligence platform around the DPT-3 model family. The service is generally available, with deployment options including US or EU cloud, customer VPCs on AWS, Azure or Google Cloud, Snowflake, and on-premises or air-gapped environments. Developers can begin in the ADE playground.