Nvidia and Intel sign “historic collaboration”

Dominant AI chipmaker Nvidia has signed what they are calling a “historic collaboration” with once-dominant, now struggling, personal computer chipmaker Intel. Lip-Bu Tan, Intel’s recently installed turnaround CEO, has quickly showed a flair for dramatic dealmaking. Last month he secured billions in federal grants by wowing President Trump in a private meeting (and promising 10% of the company’s shares to the US government.) Now he has closed a deal for $5 billion in funding from the most valuable public company in the world, tech powerhouse Nvidia. Intel gets a lifeline and a place in the new AI technology sector, a sector it had woefully lagged in, threatening its viability. Nvidia gets a storied American chipmaker, whose chips helped power the spread of personal computers, for a collaboration on chips and systems for data centers and for new AI-powered consumer computers. Nvidia CEO Jensen Huang gets to hedge his bets against a failure (or Chinese takeover) of his current exclusive chip foundry, Taiwanese TSMC. He also gets to set the roadmap for getting his chips into the devices of general consumers, without having to rely on other large companies like Microsoft who may be partial competitors with Nvidia.

Nvidia and Intel sign “historic collaboration”

Google puts Gemini into the Chrome browser

Mere weeks after winning a court decision that they would not be compelled to sell off their dominant Chrome browser (69% market share worldwide; second place Apple has 15%), Google has put the pedal to the metal and stuffed a plethora of AI goodies into Chrome. Reacting to competing AI-first browsers like Perplexity’s Comet, Google’s current Chrome update allows users to open an AI side-window to chat with Gemini. The AI will perform usual chatbot functions like responding to queries, but will also know what is in all your tabs and be able to assist you with tasks related to integrating information from them. It also has an interactive voice response feature, so you can converse with the AI aloud. The AI will integrate with Google productivity apps, so that you can use the AI to help you create and manipulate spreadsheets and documents. Coming real soon, so Google says, is getting the AI to act as an agent and perform multi-step tasks from a single request.

Google puts Gemini AI into the Chrome browser

Ireland becomes first nation to implement EU’s AI regulation law

Although the European Union’s Parliament passed it AI regulatory law last year, no country had yet implemented it - until now. Last Thursday, Ireland announced that it had designated several different government departments to manage the regulatory requirements of the AI Act for each of the sectors of society and government. The multiple regulatory agencies will be coordinated by a central body to be created within the next year. Ireland has apparently decided that rather than creating a new central AI agency first, it will split up AI regulation among existing government departments now, and kick the tricky coordinating part down the road a bit. This way it can claim to be first, without actually having to do the hardest work up front.

Ireland becomes first nation to implement EU’s AI Act

Nvidia and OpenAI sign $100 billion datacenter deal

AI chip powerhouse Nvidia has inked an agreement with OpenAI to put $100 billion into datacenter infrastructure that OpenAI can lease. This is the second blockbuster AI infrastructure deal from OpenAI in the span of 2 weeks. Last week, the company announced a $300 billion datacenter deal with Oracle, a deal so massive that Oracle’s stock price soared 40% in one day, briefly making Oracle CEO Larry Ellison the richest man in the world. The Nvidia agreement is actually larger than the Oracle pact in terms of datacenter capacity. Oracle is promising 4.5 gigawatts (GW) of computing capacity, while Nvidia is projecting 10 GW of compute. That’s over twice the compute for a third the price. Nvidia can do this because Ellison has to buy chips for his datacenters from Nvidia, and Nvidia… doesn’t. The margin on Nvidia high end AI chips is in the range of 70%. Now OpenAI CEO Sam Altman can negotiate down Oracle’s price by switching compute to Nvidia. Ellison’s vast fortune might even shrink a little if things go as poorly as they could for Oracle. Weep not for Ellison. In any likely scenario he will still have a net worth greater than several African countries.

Nvidia and OpenAI sign $100 billion datacenter deal

OpenAI Codex AI model beats 139 college teams in coding contest

For the first time, an AI model has outperformed all humans in a computer coding competition. Last Wednesday, OpenAI’s newly upgraded Codex AI model ranked first in the International Collegiate Programming Contest, besting 139 university teams from around the world. Codex was the only contestant to receive a perfect score, solving all 12 highly complex coding challenges within a strict 5-hour time limit. The highest-scoring human team solved only 11 of the 12. Google’s AI model solved 10, still rated as a Gold Medal performance. Caveat: the human competitors were college students, not the best computer coders in the world. No matter. OpenAI confidently poredicts it will be able to defeat all humans in a coding contest some time next year.

OpenAI Codex AI model beats 139 college teams in coding contest

Notion launches AI agents

Notion, the all-in-one workspace app, sometimes called the “Swiss Army Knife of productivity tools”, has just launched its first AI agent. The agent can create or modify Notion pages and databases, including Notion’s calendar, note-taker, task manager, project planner, and wiki builder, and can also link to outside tools such as email, Slack, and Google Drive. More integrations are promised soon. The agent can execute complex multi-step workflows with a single text prompt. Notion’s AI Agent is so simple to use, and yet so deceptively powerful, that Notion may break out of its current cultish niche among consumer apps and start penetrating enterprise clients in a much more substantial way than previously.

Notion launches AI agents

Google’s Gemini AI is coming to your TV

Google is upgrading its Google TV streaming manager to include its Gemini AI model. This will allow you to have a conversation with your TV about content choices, or even about the content itself. Google’s announcement highlights the ability to ask for suggestions for what to watch, to get summaries of a series’ prior episodes, or to generate educational content such as “Please explain how volcanoes erupt to my third-grader” which will pull YouTube explainer videos and give a brief text summary. Today’s so-called Smart TVs are actually pretty dumb. Google is planning to up the definition of “Smart” by including actual AI.

Google’s Gemini AI is coming to your TV

AI-operated convenience store opens in Vienna

Xpand, a startup developing “store in a box” modular retail shops, has partnered with AI robotic automation company Campo Group to open a fully autonomous convenience store in Vienna, Austria. The store is open 24/7 with no onsite staff. It offers a full line of convenience items, including a wide variety of food and beverages. Onsite cameras and other sensors feed into an AI system that monitors inventory, detects theft, and adapts ordering to match the preferences of the local customer base. This is a bold step in light of recent hilariously inept performances of AI models in running simple vending machines, as has been reported here. However, whoever makes this business model work will set an industry trend, and Xpand/Campo are betting that they are the team that will do it.

AI-operated convenience store opens in Vienna

Robotic beehives save the bees with AI

Bees are vital pollinators for over a third of our food plants, and they are dying. Approximately 40% of bee colonies collapse each year, brought down by severe weather due to climate change, mite infestations, and pesticides. Now Beewise, an AI beekeeping startup (yes, that’s a thing now) has accumulated $170 million of investor funding for commercial development of an AI-controlled robotic beehive. Beewise robotic hives monitor the bees with cameras, and AI is able to process the images and input from other sensors to determine any interventions needed to protect the bees’ health. The robotic hive can provide food, medicine, temperature regulation, as well as increased weatherproofing during severe storms. Colony loss is decreased by 70% over traditional wooden beehives, and labor costs are substantially lower. There are currently over 1,000 Beewise robotic hives installed, and that number is growing rapidly.

Robotic beehives save the bees with AI

Robot programmed to act like a 7-year-old girl comforts hospitalized kids

Hospitals are scary places for kids (ands most adults, too.) Now Robin, a semi-autonomous AI-powered emotional support robot programmed with the personality of a 7-year-old girl, is being used in over 30 hospitals in the US, including UMass Memorial Children’s Hospital. Reactions by staff and patients have been overwhelmingly positive. The robot is small, cute, with a computer-display face that can show a variety of emotions, and a triangular body designed to be easy to hug. It “remembers” patient names, favorite songs, and other personal details, and can display emotions from happiness, to empathy, and even sadness, depending on the content of the patient’s responses. Robin’s personality and reactions have been honed by thousands of interactions with patients over the years since it was invented by a graduate student in Armenia. Robin is also used in nursing homes, where it is reminiscent of a grandchild for elderly patients.

Robot programmed to act like a 7-year-old girl comforts hospitalized kids

Scientists use AI to design and build bacteria-killing viruses

Bacterial resistance to antibiotics is a growing problem worldwide, and new approaches are needed. Scientists at the Arc Institute have developed an AI system that can design and build bacteriophages, which are bacteria-killing viruses.

Starting with a known bacteriophage, effective against the bacterium E. Coli, the AI system suggested modifications which should make the bacteriophage even more effective. Multiple different modifications were implemented in different strains of the genetically engineered bacteriophages, and then tested for effectiveness in killing the target bacteria. After several rounds of this process, novel and highly effective bacteriophage strains were developed.

The scientists note that combining different AI-engineered bacteriophages in the treatment of a patient’s infection is likely to make it more difficult for the target bacteria to evolve comprehensive resistance to the bacteriophage therapy.

Scientists use AI to design and build bacteria-killing viruses

AI is democratizing medical expertise

Robert Pearl MD, former CEO of Kaiser Permanente Medical Group, is a staunch advocate for AI in health care. He made the case in his bestselling 2024 book, “ChatGPT MD” that AI had the potential to transform medical care for the better, and in particular to address the glaring deficiencies of the US healthcare system, which has been described as “neither healthy, caring, or a system.”

In a recent LinkedIn post he notes that in recent surveys, 78% of patients and 65% of medical professionals have recently turned to AI for answers to health-related questions, and 88% of all users rated AI-provided medical answers as “good” or “excellent.” In addition, multiple studies have shown that AI systems can equal or exceed the accuracy of experienced physicians in making diagnoses of medical conditions.

He now argues that patients are beginning to “climb the ladder” of medical expertise, moving from the current paradigm in which the physician is the sole expert and gatekeeper of medical knowledge, to a system of shared expertise and collaboration between physicians and patients.

He looks forward to a near-term future in which patients are assumed to be able to get accurate, up to date, and personalized information and suggestions for common medical conditions from always-available AI. Soon after, he expects that advances in wearable fitness watches and in home medical monitoring equipment will allow for continuous monitoring of chronic diseases such as diabetes, with personalized suggestions and alerts for the patient, and potential alerts to the patient’s care team for urgent results.

AI is democratizing medical expertise