Hashtag Trending

AI Hardware Arms Race & Humanoid Robot Olympics: Apple Chips, OpenAI 'Jalapeno,' Meta Models, and Robot Prices Plummet
Jim Love, Marcel Gagné, and retired John Pinard cover tech news and a deep dive on robots. They note Linux's 35th anniversary and how the name changed from "Freax" to Linux. They discuss Apple's AI-focused hardware advances (integrated CPU/GPU/neural processing and shared memory) enabling large local model inference, and OpenAI's "Jalapeno" inference chip built with Broadcom, developed in about nine months, delivering 1.5–1.9x more AI work per watt, raising questions for NVIDIA's CUDA moat. They touch on Meta's strong new Muse Glimmer/Spark models despite Reuters reporting failed internal AI rollout, morale issues, and metrics like 220% more AI-assisted coding with 40% reaching production alongside slower issue resolution. Bill Gates warns AI's growing capability could drive permanent job loss and dual-use security/bio risks. They then review China's World Humanoid Robot Competitions, rapid performance gains, autonomous team sports, and one-shot robot learning, plus falling robot price points—from $8K household task bots to Unitree's ~$5,900 humanoid and a ~$1,300 wheeled home helper—arguing autonomy is advancing fast.
00:00 Welcome and Introductions
00:48 Linux Turns 35
02:29 Apple Goes All In
06:18 OpenAI Builds Chips
10:41 Meta Models and Missteps
17:25 Why AI Rollouts Fail
20:33 Bill Gates Sounds Alarm
24:59 AGI and Persistent Agents
33:03 Humanoid Robot Olympics
34:56 Robot Sprint Breakdown
35:49 Humanoid Games Highlights
37:04 Soccer Bots And Teamwork
39:19 From Sports To Warfare
41:04 One Shot Robot Learning
46:33 Robots Making Robots
49:11 Home Robots Price Drop
52:14 MicroDuck And Cheap Bots
53:46 Unitree R1 Under 6K
58:55 Nori Budget House Helper
01:04:18 Autonomy And Price Point
01:05:55 Optimus 2027 Outlook
01:07:42 Weekly Wrap And Signoff

OpenAI has unveiled Jalapeño, its first custom AI inference chip, and the early results could challenge one of Nvidia's biggest advantages.
OpenAI says Jalapeño delivers between 1.5 and 1.9 times more AI work per watt than the systems it tested against. SemiAnalysis watched benchmark runs inside OpenAI's lab and reported strong performance-per-watt results against Nvidia Blackwell. But the bigger story may be software: OpenAI is using Codex to write optimized kernels, potentially weakening the nearly 20-year CUDA software moat that has helped protect Nvidia from competitors.
Also on this episode of Hashtag Trending:
A Reuters investigation reveals how Meta's ambitious AI-driven restructuring plan fell apart. AI-assisted code increased dramatically, but gains in actual product features were far smaller, while technical incidents increased and employee morale deteriorated.
Canada is investing more than $500 million in 64 new research chairs as it tries to turn decades of brain drain into brain gain. The real test will be whether Canada can keep the intellectual property, companies and economic benefits created by those researchers.
And Twitter is back. Sort of. Operation Bluebird has launched a new social network using the Twitter name and the familiar language of tweets and retweets. Elon Musk's X says it still owns the trademarks and is suing.
Hashtag Trending is hosted by Jim Love.
00:00 Headlines
00:27 OpenAI unveils the Jalapeño AI chip
01:39 Can OpenAI challenge Nvidia's CUDA moat
03:02 Meta's AI restructuring plan backfires
04:50 Canada makes a major brain-gain push
06:33 Can Canada turn research into Canadian industry
07:38 X battles a new company called Twitter
09:18 Wrap-up

Meta will pay at least $12.1 billion—and potentially as much as $17.1 billion over ten years—to settle claims that Facebook and Instagram harmed young users.
Jim Love examines why the record settlement may still be a smart move for Meta. The company admits no wrongdoing, limits its legal exposure and avoids a verdict that thousands of other plaintiffs might have used in their own lawsuits.
TIME also reports that Sam Altman expects OpenAI to have an internal system he would call artificial general intelligence by the end of 2026. What would qualify as AGI, and why would it matter more than passing the Turing Test? We separate confirmed reporting about OpenAI's delayed Astra models and cybersecurity concerns from unconfirmed rumours involving persistent memory and continual learning.
Also in this episode:
The Justice Department and FBI disable QScan and QTRouter, two hacking platforms allegedly used by Chinese state-sponsored attackers against NASA, the Federal Reserve, the U.S. Senate and hundreds of other organizations.
The Environmental Protection Agency proposes removing federal minimum public-participation requirements for minor-source air permits, potentially allowing some data centre generator permits to be approved without public notice or comment.
00:00 Today's tech headlines
01:22 Meta's record teen-safety settlement
04:22 Altman says AGI could arrive this year
08:06 FBI disrupts China-linked hacking platforms
10:44 EPA proposes permit-notice changes
14:11 Wrap-up and where to listen
Hashtag Trending for Thursday, August 27, 2026, hosted by Jim Love.

Apple is making a much bigger bet on local AI. Jim Love looks at the new M6 Mac mini, Apple's first 2nm processor, with Neural Accelerators built into every GPU core, dual 16-core Neural Engines and significantly faster processing. At the high end, the new M5 Ultra Mac Studio can offer up to 80 GPU cores and 512GB of unified memory, potentially allowing enormous AI models to run entirely on the Mac.
The performance comes with a price. The new M6 Mac mini starts at $899 U.S., compared with the original $599 starting price of the M4 model. But Apple did announce one major price cut: its famous polishing cloth drops from $19 to $9.
Also on Hashtag Trending:
A new investigation warns that a catastrophic failure of the aging U.S. electrical grid could leave parts of the country without power for 18 months or more. Large transformers are often custom built, difficult to replace and increasingly hard to source. An earlier FERC analysis identified roughly 30 particularly critical substations and concluded that losing the right nine simultaneously could potentially trigger a nationwide blackout.
Then, nearly 6,000 senior executives were asked what AI has actually done for their companies. More than 90 percent reported no impact on employment and 89 percent reported no improvement in labour productivity. Yet executives still expect AI to reduce employment. Separate research finds that AI-related layoffs can make employees less receptive to the very technology companies hope will increase productivity.
And Linux turns 35. Linus Torvalds introduced his new operating system in 1991 by calling it "just a hobby" that wouldn't become big or professional. Today Linux powers servers, cloud infrastructure, Android and every one of the world's Top500 supercomputers.
Hosted by Jim Love.
00:00 Headlines and intro
00:33 Apple M6 and the local AI push
03:45 Apple's $9 polishing cloth
04:24 America's aging power grid
06:06 AI productivity reality check
08:07 Linux turns 35
10:07 Wrap-up and where to listen

Sam Altman is warning that artificial intelligence could become concentrated in the hands of a few companies, models or individuals. But there's an uncomfortable tension in that warning: OpenAI and Anthropic both rely heavily on proprietary models, while Chinese companies are aggressively releasing increasingly capable open-weight alternatives.
Jim Love also looks at billionaire Mark Cuban's proposal for tackling wealth inequality by giving workers equity in the companies they help build, with Nvidia CEO Jensen Huang offering a similar philosophy on employee compensation.
Then, a developer discovers that AliExpress was silently using Web Audio and a range of hardware and software signals to create a fingerprint of visitors' devices and send encrypted results back to Alibaba servers. Brave blocked the technique, while Chrome allowed it in the researcher's testing.
And Apple is cutting more than 200 jobs in Siri and Vision Pro-related teams just days before the company's CEO transition, despite recently reporting its strongest June quarter ever.