Nintendo brought Virtual Boy games to Switch and Switch 2 owners through its online service earlier this year. The collection lets players experience the red-tinted stereoscopic titles without digging out the original bulky headset. Yet something important stayed missing from the official offering. The service provides no dedicated controller, and standard Switch pads force awkward compromises that dull the precision the old hardware once delivered.

The original Virtual Boy controller is one of Nintendo’s most eye-catching designs from the 1990s, and there are still quite a few floating around. Anyone who has used the peripheral on its stand will recognize the two large grips protruding from the body. One directional pad is on the left, while another is on the right. The red A and B buttons cluster next to the right pad, while the grey Start and Select buttons hang around near the left one. Plus, the rounded L and R triggers are quite distinct. Some games used to use a dual-pad layout to create the illusion of depth in what was otherwise a beautiful 2D environment, but when you translate all of that to a current analog stick or single pad, you end up with dead zones or jittery responses that really mess up the vibe.

Nintendo Switch 2: Choose Your Game Bundle

• Kick off the fun with a Nintendo Switch 2 system, your choice of a select digital game, and a savings* of up to $29.99!
• Includes choice of either the Mario Kart World, Donkey Kong Bananza, or Pokémon Pokopia digital game download
• One system, three play modes: TV, Tabletop, and Handheld

RetroOnyx’s new circuit board is game-changing. This board can be inserted directly into an original Virtual Boy controller and replaces almost everything inside it. The ESP32 microcontroller is at the heart of it all, handling all wireless communications and button translation. What about the remaining original hardware? Nothing gets affected. The controller case itself has not been touched; there are no permanent cuts or new holes. It is entirely possible to revert this thing to its original state.

Installing this new board only involves a few simple tools and some soldering. Screw the shell open, remove the old board after you’ve added some new solder points to the battery terminals, and that’s all. The new board is installed with the power switch matched up with the existing slider and a small 3D printed plug that sends a light pipe into the old cable exit hole. Nothing really complicated. After that, reassembly takes only a few minutes if you’ve positioned the board correctly, and the result? Every physical button and pad remains in its factory position and spacing.

On a Switch or Switch 2, the upgraded controller appears as a normal Pro Controller. Pairing is a typical Bluetooth procedure; just follow the usual steps. A simple button push at power-on allows you to select the operating mode; the default is Virtual Boy on Nintendo Switch online, but you may hold the button for a second to switch to PC use or some experimental mappings that resemble vintage SNES or N64 layouts. Once you’ve paired everything up, the left directional pad appears as the left pad on the Switch side, while the right directional pad appears as the right analog stick. Face buttons swap over in a fairly straightforward manner to preserve muscle memory, and L and R triggers simply map to their modern counterparts as is.

Additional button combinations will enable a host of features that the Switch is actually anticipating. Holding specific buttons on the controller will access the home menu, increase the volume, modify the brightness on the emulated display, initiate a rewind (if supported), or even start a capture. The ESP32 handles all of these commands on its own, with no additional software required on the console, and best of all, the latency is low enough that the games feel extremely responsive. Most gamers agree that after you get past the first few minutes of adjusting, the entire experience feels very natural.

The same hardware breathes new life into the original Virtual Boy system, but this time wirelessly. All you need is a separate BlueRetro adapter that fits into the console’s controller port and pairs with the improved controller. As a result, depending on which of the two receivers is nearby, one controller can now serve as hardware for both the original 1995 system and the current Nintendo products. However, if Nintendo decides to enable haptic rumble on the Virtual Boy collection, subsequent updates may include haptic modules inside the grips.

RetroOnyx sells the board as a kit for $99, but you must provide your own donor controller and battery pack. They’ll occasionally have assembled versions available if they can find secondhand controllers. This project is actually pretty cool, because it fills a gap that Nintendo left open when they released all those games without any accompanying hardware, and it demonstrates how far a well-designed replacement board can extend the life of a really cool, but otherwise largely forgotten peripheral that would otherwise collect dust on a shelf.

Roblox announced Thursday a new feature called “Build,” allowing users to design games from their mobile devices using AI. 

The Build feature lets anyone turn simple text prompts into a basic game without any programming experience. For example, if a user types, “Let’s make a cozy adventure game set in a dense forest,” the new feature will generate an initial version of the game, which users can then modify and share with friends.

“Powered by a broad set of AI models, including both open-source and proprietary Roblox models, Build handles gameplay mechanics, environment, characters, visual style, sound, and more,” the company wrote in its blog post.

Companies like Google, Microsoft, and Tencent have built similar tools. However, AI-powered game generation has raised concerns among developers and players, with critics arguing that reducing the barriers to game development via text prompts could lead to an influx of low-quality and repetitive games. This may also increase competition on the platform, as creators are required to compete not only with other developers but also with AI-generated content that can be produced far more quickly. 

These concerns are reflected in this year’s Game Developers Conference State of the Game Industry survey, which found that 52% of game industry professionals believe generative AI is having a negative impact on the industry.

To address this, Roblox plans to rank these AI-generated games based on player retention, similar to the system used for other games on the platform. If a game is not played, it won’t be featured as prominently.

“Our discovery systems are designed to highlight games with long-term retention, which doesn’t include AI slop. The quality of games on the homepage isn’t changing: If no one plays it — no one can find it. The goal across these new tools is to continue to accelerate creation across all experience levels,” the company added.

The Build feature will enter public alpha testing on July 28, available to users in New Zealand aged nine and older who have verified their age. Users aged 16 and up will have the opportunity to publish their creations to a global audience. There will be a free, basic version available along with paid options.

Beyond the Build feature, Roblox is also working on developing AI agents that will assist creators in playtesting and providing analytics. These features are anticipated to roll out in the upcoming months.

The new feature highlights Roblox’s ongoing investment in AI, including an AI foundation model for generating 3D game assets and an AI chatbot for supporting developers through the game-building process. Additionally, Roblox is developing a “new scene-generation model” capable of creating entire editable and playable 3D scenes from a single text prompt.

Additionally, the announcement comes shortly after Roblox disclosed plans to discontinue “Roblox Connect,” the avatar-based video-calling feature introduced in 2023.

Moonshot announced a $2bn raise at $12bn valuation in May.

China is showcasing its AI prowess despite efforts by Washington, as Moonshot AI’s new Kimi K3 boasts a performance close to OpenAI and Anthropic’s latest models.

At 2.8trn parametres, Kimi K3 is the largest open-weight AI model available in the market today. The multimodal model is designed for frontier intelligence across long-horizon coding, knowledge work and reasoning.

Across benchmarks, K3 only trails behind OpenAI’s newest GPT-5.6 Sol and Anthropic’s Fable 5, while outperforming them in some coding and general agent tasks. The company said that K3 performed “competitively” with Fable 5 and “substantially outperformed” Opus-4.8, GPT-5.6 Sol and GPT-5.5.

K3 is also the cheapest of the three, at around $0.94 per Artificial Analysis intelligence index task and $15 per million output tokens, while GPT-5.6 Sol max costs around $1.04 and Claude Fable 5 makes a significant leap to around $2.75 per task.

Although cheaper than the American juggernauts, Moonshot’s new model marks a major leap in price compared to its Chinese contemporaries. DeepSeek’s V4 Pro costs around $0.04 per index task according to Artificial Analysis rankings, while MiniMax’s M3 costs around $0.12 for the same.

Moonshot unveiled its new model just months after announcing a $2bn raise from Chinese food delivery company Meituan’s VC arm Long-Z Investments, alongside Shuimu Capital, China Mobile and CPE Yuanfeng. The May round valued the start-up at around $20bn.

The launch marks the latest in escalating tensions between US and China over AI leadership. US curbs on semiconductor exports to China over recent years has pushed China to focus on its own chip capabilities, with Moonshot AI president Yutong Zhang telling the audience at the World Economic Forum earlier this year: “We knew we didn’t have the luxury to simply scale up compute…That forced us to focus on fundamental research and efficiency.”

The US is taking a more restrictive approach to sharing advanced home-grown AI models by only allowing approved bodies to access GPT-5.6, Fable and Mythos – much to OpenAI’s disappointment.

Lawmakers from the country, meanwhile, are also urging the Trump administration to ban US companies from buying memory chips from Chinese semiconductor developers.

Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.

I’ve uncovered a pretty cool way to save almost $400 off a ThinkPad by combining two separate promo codes.

Right now, you can get the Lenovo ThinkPad T14 Gen 7 AMD for $1,324.39 (was $1719) at Lenovo. The laptop is already discounted down to $1,471.55. But then, enter coupon code THINKWEEKEND, which saves you $77.45, and code LENOVO10DEAL to save an extra $147.16. That’s almost $400 off one of our top-rated business laptops.

I don’t know how long this loophole will last though, so use them while you can. Better still, you can run this coupon combo across a range of other laptops and desktops, too.

Business laptops can sometimes blur together, but Lenovo‘s ThinkPad range has always stood out for me, earning its reputation by focusing on reliability, security, and practical features that make everyday work easier.

Powered by AMD‘s Ryzen 5 PRO 230 processor, the ThinkPad T14 has more than enough performance for office work, web browsing, video meetings, spreadsheets, and multitasking.

You get 16GB of DDR5 memory and a 512GB PCIe Gen4 SSD, as well as a fingerprint reader for fast sign-ins, a 5MP RGB and IR webcam, a physical privacy shutter, and Windows 11 Pro, making it well suited to business tasks.

The 14-inch WUXGA IPS display has a 1920 x 1200 resolution with an anti-glare finish and 400 nits of brightness, making it easier to work in bright offices or while traveling on a train or plane.

A backlit keyboard, integrated Radeon 760M graphics, Wi-Fi 7, and Bluetooth 5.4 round out modern specifications that mean the laptop should continue to be perfectly fine for many years.

Although Lenovo markets the ThinkPad T14 primarily toward professionals and small businesses, it also makes plenty of sense for students and anyone wanting a durable, lightweight laptop that puts productivity ahead of flashy, but entirely unnecessary, extras.

If you need a dependable laptop that can handle almost anything you throw at it, these combined discounts make the ThinkPad T14 Gen 7 AMD an easy recommendation.

For more top-performers, see our guide to the best business laptop. And for more savings, see our round-up of the best business laptop deals around right now.

Also consider: More ThinkPad deals

AI is reshaping tech careers. But it will not kill off the value of a STEM degree, according to Google DeepMind chief Demis Hassabis.

He made the comments at a London business conference, in a video published on Wednesday. Knowing the fundamentals of software gives you an edge with AI, he told the audience, Business Insider reported.

“You absolutely needed to lean into STEM and computer science,” Hassabis said. He framed AI as the next programming language, after machine code, C and Python. The future, he suggested, may be plain English.

Fundamentals still matter

Even so, the basics do not go away. “You’re still going to need to know about architecting things and best software engineering practices,” he said.

“Those people who understand the deep technical, they’ll be able to use these tools 10 times more effectively than people who don’t have that technical knowledge,” Hassabis added.

He also made a case for the humanities. “The time is now for the humanities like philosophy, economics,” he said. “I think we really need them in the world we’re about to enter.”

A wider pushback on vibe coding

Hassabis is one of several tech leaders pushing back on the idea that vibe coding makes coding degrees pointless.

Geoffrey Hinton, often called the godfather of AI, made a similar case to Business Insider in December. A mid-level programming job “is not going to be a career for much longer, because AI can do that,” he said. But he argued a computer science degree is worth far more than coding. It will stay useful “for quite a long time,” he said.

Affirm chief Max Levchin has said much the same. Computer science fundamentals separate good code from “garbage,” he argued on a podcast. Microsoft’s Brad Smith and others have offered similar reassurance to anxious graduates.

OpenAI’s Sora may have shut down, but Google apparently thinks there’s still interest in a tool that lets you star in your own AI videos. On Thursday, the tech giant announced an update to Google Vids that will allow you to create a custom digital avatar that looks and sounds like you based on a selfie and a voice recording you upload.

In addition, Google said it’s bringing its multi-modal AI model Gemini Omni to Vids, letting you create videos using a combination of a written prompt and reference images you upload. Omni then mixes those inputs together to create the AI video you want. It can also be used to do things like swap out the background or fix the lighting in a video recorded on your phone, or add effects.

Plus, Omni now supports step-by-step edits, meaning you can make changes to your video as you go instead of starting over from scratch.

The updates push Google Vids beyond its original role as an AI-assisted workplace presentation tool to become more of an all-in-one video creation platform. By making Vids a part of Google Workspace, the company is telegraphing its use as a business tool for things like company updates or training videos, but personalized avatars and conversational edits could put it in closer competition with other AI video startups and tools like HeyGen, Synthesia, Captions, D-ID, and others.

Google notes that the new AI avatars will be tied to the account holder’s likeness, tied to their Google account, and watermarked invisibly with SynthID. (I suppose that means no one will be using the tool to make bizarre AI videos of Google CEO Sundar Pichai, the way that OpenAI CEO Sam Altman had let users do with Sora when it was available!)

The company also says that access to personal avatars is limited to users in certain regions who are aged 18 or older.

Ryan Beiermeister has joined Founders Fund as a partner, she announced on Monday. Beiermeister is well-known in Silicon Valley for a number of reasons. For one, prior to this role, she spent about two years as VP of Product Policy at OpenAI as it became a household name, shortly after ChatGPT became the fastest-growing app in history.

That career choice ended abruptly in February when she was reportedly fired after objecting to a planned ChatGPT feature called “adult mode,” which was going to allow adults to use the chatbot for erotica. The Wall Street Journal reported that her firing involved an accusation by a male colleague of sexual discrimination, although Beiermeister called any allegation that she discriminated against anyone “absolutely false.” In March, OpenAI reportedly scrapped plans for adult mode.

More recently, Beiermeister has become well-known in Silicon Valley for her skillful strategy in a Founders Fund YouTube show called “Mafia.” The game involves discovering which players are secret Mafia killers before those players can “kill” the rest of the players.

Beiermeister played the game against OpenAI’s Sam Altman, Anduril’s Palmer Luckey, Figma’s Dylan Field, Flexport’s Ryan Petersen, Founders Fund’s Trae Stephens, and several others.

One of the most intense scenes in Episode One involved her and Altman each saying that if they were found dead, it would mean the other was the killer. Those who knew the history laughed.

Some commented on Twitter that maybe the whole Mafia game was really a job interview for her. The game, according to the firm’s chief marketing officer and the game’s MC, Mike Solana (who brought the game to the firm), is often played at Founders Fund retreats.

However, it wasn’t. “Though she is an excellent Mafia player, that wasn’t part of her interview process. She has been close with Trae Stephens since they worked together at Palantir and has been friendly with our team for years,” a Founders Fund spokesperson told TechCrunch.

Though the way Beiermeister played the game — coolly, making analytical observations and arguments about who might be Mafia — couldn’t have hurt her prospects.

Still, Beiermeister has known Trae Stephens for at least a decade. Prior to her role at OpenAI, and at Meta before that, she spent her formative years at Palantir, the big data company founded by the VC firm’s founder, Peter Thiel. Stephens also worked at Palantir in its early days.

Beiermeister says she’s most interested in backing the kinds of startups that Founders Fund is known to gravitate toward.

“The companies that will define the next twenty years are being built in the categories where product engineering is hardest and the stakes are highest — AI infrastructure and agentic systems, defense, energy, climate, biotech, the regulated frontier,” she wrote in a LinkedIn post. “To the founders in these domains, especially if you don’t fit the standard mold: I want to talk to you and my inbox is open.”

• HP India has been fined 1.42 billion rupees ($14.7 million) for two separate cases
• Self-reporting ultimately landed it with lighter fines
• Both cases relate to government tender manipulation between 2017 and 2020

India’s Competition Commission (CCI) has accused HP India and some related resellers of coordinating bids for Indian government contracts on the Government e-Marketplace.

According to the CCI, the company and certain partners manipulated government tenders by predetermined or communicated bid prices, submitting deliberately uncompetitive bids to create the appearance of competition and controlling discounts.