• Apple has joined the board of the Ultra Accelerator Link consortium
• The link is a key technology that binds GPUs, not unlike synapses on neurons
• UALink is emerging as the biggest rival to Nvidia’s proprietary NVLink

Back in June 2024, we reported how a number of big tech names had banded together to form the Ultra Accelerator Link (UALink) Promoter Group, a strategic move aimed at reducing Nvidia‘s dominance in the AI accelerator market.

Directly competing with Nvidia’s proprietary NVLink technology, UALink seeks to develop a new industry standard for high-speed, low-latency communication for scale-up AI systems in data centers. It already has the backing of Intel, AMD, Google, Microsoft, Meta, HPE, Cisco, and Broadcom, but now Apple has joined the UALink board too.

Added support for DLSS 4 with Multi Frame Generation for GeForce RTX 50 Series graphics cards, which boosts FPS by using AI to generate up to three times per traditionally rendered frame – enabled with GeForce RTX 50 Series on January 30th. DLSS 4 also introduces faster single Frame Generation with reduced memory usage for RTX 50 and 40 Series. Additionally, you can now choose between the CNN model or the new Transformer model for DLSS Ray Reconstruction, DLSS Super Resolution, and DLAA on all GeForce RTX graphics cards today. The new Transformer model enhances stability, lighting, and detail in motion.

Mukesh Ambani’s Reliance is planning to build what could become the world’s largest data center in Jamnagar, India, with a capacity of three gigawatts to capitalize on surging AI demand.

The facility would dwarf the current largest data center, Microsoft’s 600-megawatt site in Virginia, Bloomberg reported Friday. The project could cost between $20 billion to $30 billion, the report added.

Ambani raised more than $25 billion in 2020 from a group of investors including Meta, Google, Silver Lake, General Atlantic, KKR, Mubadala and PIF to fund the growth of Reliance’s retail and telecom ventures that now dominate the country. Reliance is India’s most valuable company.

Ambani aims to power the facility primarily with renewable energy from an adjacent green energy complex that will produce solar, wind and hydrogen power.

Ambani is buying chips from Nvidia for the data center, the report added. Nvidia and Reliance announced a partnership to build infrastructure for AI applications in India in October.

The Jamnagar project comes as OpenAI, SoftBank and Oracle this week pledged up to $500 billion for AI infrastructure in the United States through their Stargate Project.

Good morning! Let’s play Connections, the NYT’s clever word game that challenges you to group answers in various categories. It can be tough, so read on if you need clues.

What should you do once you’ve finished? Why, play some more word games of course. I’ve also got daily Strands hints and answers and Quordle hints and answers articles if you need help for those too, while Marc’s Wordle today page covers the original viral word game.

Google is bringing smart home controls in Gemini to everyone. The Google Home extension in the Gemini app is adding a few new features, in addition to letting you adjust your smart lighting, thermostat, speakers, and other compatible devices as long as they’re connected to your Google account.

Google first previewed the extension last November. With it, you can use natural language to control your smart home when interacting with Gemini, such as saying “The sun is too bright in the living room” to close your smart blinds. But now, Gemini can also carry out multiple requests, like “Turn the armchair light on too, but dim the kitchen lamp.” You’ll be able to use the Google Home extension to ask Gemini about the status of your devices too, such as whether you’ve left your porch light on.

Additionally, Google will let you control “non-sensitive” smart home devices, like your lights, from your phone’s lock screen. Other updates include the ability to adjust the volume, pause, and resume media on smart speakers, displays, and TVs within the Gemini app, as well as an updated thermostat control design that matches the one inside Google Home. Gemini will also automatically open the Google Home app for security-related actions for cameras and locks (it previously only linked you).

The launch of the Google Home extension follows a big update to Gemini, which lets it perform more complex tasks across multiple apps. You can try out the integration for yourself by signing into Gemini with the same account you use for Home and turning on the Google Home extension. It launches today but is rolling out “over the coming weeks.”

Tesla has announced that its redesigned Model Y SUV is coming to the U.S., Canada, and Mexico in March, with a starting price just shy of $60,000.

The news comes just two weeks after Tesla first revealed the new-look Model Y and said it was coming to China and other Asian markets, also in March. Thursday’s announcement means the company is effectively launching the revamped SUV simultaneously around the globe — a departure from the multiple-month gap between the Asian and North American launch of the Model 3 sedan refresh in late 2023 and early 2024.

The redesigned Model Y is being launched at a crucial time for Tesla, which delivered fewer vehicles in 2024 than it did in 2023. Tesla has repeatedly warned investors that it is in between “two major growth waves” coming off the success of the Model Y, and promised that it will roll out mysterious new models meant to be built on existing production lines. Those models will likely be cheaper than Tesla’s current offerings (which start in the low $40,000 range), but it’s not clear by how much.

CEO Elon Musk has implied that those new models, plus the so-called Cybercab that was teased last October, will help bridge the company’s evolution from an automaker into a robotics and AI player.

But at the same time, Tesla’s vehicle lineup has been aging. Tesla has now refreshed each of its core vehicles — the Model S and 3 sedans, and the Model X and Y SUVs — but has only launched one truly new model in the last four years, the Cybertruck. While it became the best-selling electric truck in the U.S. in 2024, the Cybertruck did little to boost the company’s bottom line last year, and it does not seem to be the runaway hit Musk hoped for.

The new-look Model Y could offer some relief, though it is coming in at a higher price point than the existing versions. The starting price for the so-called “Launch Series” special edition, which is an all-wheel drive variant, is $59,990. That gets buyers a 320-mile range battery and it includes Tesla’s most advanced driver assistance software, which it calls “Full Self-Driving (Supervised)” — typically an $8,000 option. The older Model Y currently starts at $44,990 for a 337-mile rear-wheel drive version.

The most noticeable changes to the new Model Y come on the exterior, where the bubbly front fascia has been ditched in favor of a more cinched nose with a thin light bar that stretches across the hood. The rear of the vehicle also now has a light strip that stretches the full width.

Inside the refreshed SUV, Tesla has added a configurable light strip that rims the cabin. There’s a new rear-passenger touchscreen, and some quality-of-life upgrades like powered rear seats and an improved suspension.

• Perplexity AI has released a mobile app for Android
• The Perplexity Assistant offers voice, text, and camera-based interactions for tasks such as booking rides and identifying objects
• The assistant integrates with apps and leverages real-time information and task automation

AI conversational search engine Perplexity is going mobile on the Google Play Store with a new Android app. Peeplexity’s app pitches itself as a kind of digital Swiss Army knife that can manage tasks for you, including making reservations and identifying objects through your phone’s camera. Best of all, the app is free and speaks 15 languages.

By leveraging Perplexity’s own search engine, the assistant can also tap into real-time web information, so it’s not just regurgitating pre-programmed answers. This should, in theory, make it smarter and more versatile than many of its competitors. To juggle all of those abilities, Perplexity can maintain context across multiple tasks. That means it won’t double-book you and will remember what you like and don’t like.

Netflix plans to offer couch co-op and party games that it will stream over the cloud to TVs, co-CEO Greg Peters said as part of the company’s Q4 2024 earnings announcements this week. The company has offered cloud gaming as a beta to a “subset” of subscribers since 2023, so this news from Peters indicates that the company is going to continue to invest in it.

Peters didn’t say exactly when the co-op and party games might be available. But he did say that “we think of this as a successor to family board game night or an evolution of what the game show on TV used to be.”

Netflix will also continue to focus on “more narrative games based on Netflix IP” — Peters says those games are “consistent fan favorites and we’ve got a lot in the library to work with there.”

OpenAI says that it might store chats and associated screenshots from customers who use Operator, the company’s AI “agent” tool, for up to 90 days — even after a user manually deletes them.

OpenAI has a similar deleted data retention policy for ChatGPT, its AI-powered chatbot platform. However, the retention period for ChatGPT is only 30 days, which is 60 days shorter than Operator’s.

OpenAI says its policies around data retention for Operator are designed to combat abuse. “As agents are a relatively new technology, we wanted to make sure our teams have the time to better understand and review potential abuse vectors,” an OpenAI spokesperson told TechCrunch. “This retention period allows us to enhance fraud monitoring and ensure the product remains safe from misuse, while still giving users control over their data.”

OpenAI announced Operator on Thursday and released it in a research preview for subscribers to the company’s $200-per-month ChatGPT Pro plan. Operator is a general-purpose AI agent with a built-in browser that can independently perform certain actions on websites.

OpenAI claims that Operator can automate tasks like booking travel accommodations, making restaurant reservations, and shopping online. There are several task categories users can choose from within the Operator interface, including shopping, delivery, dining, and travel.

Operator captures screenshots of its built-in browser to help it understand how and when to take actions in apps, like when to use buttons and which forms to complete. To be clear, Operator doesn’t capture screenshots when it gets “stuck,” like when the tool needs a password. OpenAI calls this “take over” mode.

Still, some users may be wary of volunteering screenshots of their online activities to a company that may keep them for upwards of three months. OpenAI notes that, as with ChatGPT, Operator data may be accessed by “a limited number of authorized OpenAI personnel” and “trusted service providers” for purposes like investigating abuse and handling legal matters.

Quordle was one of the original Wordle alternatives and is still going strong now more than 1,000 games later. It offers a genuine challenge, though, so read on if you need some Quordle hints today – or scroll down further for the answers.

Enjoy playing word games? You can also check out my NYT Connections today and NYT Strands today pages for hints and answers for those puzzles, while Marc’s Wordle today column covers the original viral word game.