Valar Atomics confirmed the whisperings that it was raising a giant round led by Sequoia on Monday. The startup’s founder and CEO Isaiah Taylor announced that the company has raised $1 billion in equity, with Sequoia partner Shaun Maguire joining its board, and that it also secured a $200 million line of credit from Erebor and other banks.

While the nuclear startup did not disclose its valuation, Bloomberg reported that it was $6 billion.

Valar is building small modular nuclear reactors (SMRs) — essentially miniaturized, factory-built power plants intended to be cheaper and faster to build than traditional reactors. In June, it demonstrated its Ward 250 reactor successfully powering an Nvidia Blackwell system, the company said, and announced a deal with Nvidia to develop a waterless 30MW AI factory.

Valar says its reactors generate the engineering, manufacturing, and operational data needed to improve next-generation reactors, speeding its development cycles.

“It took two years to complete the NOVA core. It took seven months to take Ward 250 critical. With each reactor built, the tick rate will become smaller until Valar is producing tens, then hundreds, then thousands of reactors per year,” it said, adding that this funding will allow it to reach to its next milestone: manufacturing fleets of its SMRs.

It certainly isn’t the only nuclear startup pursuing such ambitions. Over the past few months, Antares raised $470 million, and X-energy raised $1 billion through its IPO.

As for Valar’s new round, other investors include Apandion Capital, Atreides Management, Conviction, Dream Ventures, HOF Capital, Point72 Ventures, Riot Ventures, Snowpoint Ventures, and Valor Equity Partners.

Apple’s iPhone production in India is getting more of a boost, as the government considers extending a tax exemption on manufacturing machinery and component imports until 2041.

India is a major location in Apple’s work to diversify its production outside of China. While it has already benefited from many incentives from the country, its supply chain stands to benefit even more.

A draft proposal seen by Reuters affects a tax exemption that India introduced in February. That incentive, applicable to foreign companies, was introduced after Apple lobbied against being taxed for iPhone machinery it owned in India but provided to contract manufacturers.

Under the original incentive, it was set to expire in 2031. Under the proposals, it would instead run until 2041.

At this time, the draft proposal still needs to pass through both houses of Parliament for approval, in a process that can take months, if not years.

Beyond machinery

The proposals don’t just extend the period of exemption on taxes for machinery. It also extends into the realm of component storage with regard to exports.

India will also exempt foreign companies from being taxed for storing and shipping parts used to manufacture components and products. However, only for hardware that is not intended for sale in India.

Instead, factories and warehouses would have to set up in customs-bonded areas. This refers to regions that are considered to be outside India’s customs border for taxation purposes.

Items imported and stored in these areas would not be subject to import taxes. That is, so long as the products and components made with them are then exported out of the country.

If those items are then sold within India, they would be subject to import taxes as normal.

For Apple, the change would be beneficial, since it wouldn’t pay import duties. The intention is to make India more attractive for producing iPhones intended for exports.

Apple already benefits from import duty cuts on July 9, affecting select components used to produce smartphones. The proposed change will bring those same taxes down to zero, for export-only production.

Data storage

While the bulk of the proposals surround manufacturing, there’s also some relating to another kind of storage. That of data.

In February, India introduced a tax exemption for foreign companies using data centers in the country, for providing services to users around the world. That tax exemption lasts until 2047.

However, the proposal extends the exemption to allow those data centers to be leased by Indian partners, rather than be owned outright. The change means it will lower the capital requirements for Indian partners, opening it up to smaller and mid-sized firms.

To Apple, it means a bit more flexibility if it wants to set up data centers in the country.

Presented by NTT DATA AIVista

At VB Transform 2026, NTT DATA AIVista CEO Bratin Saha joined VentureBeat CEO and editor-in-chief Matt Marshall to discuss the last-mile challenge of operationalizing frontier models in regulated production, where reliability, context, guardrails, and security determine whether AI delivers enterprise value. The conversation centered around the question facing every enterprise now pouring money into AI: how to convert that spending into real, tangible value.

“It’s not just a model, you’re building a system around the model,” Saha said. The last mile is the work of wrapping a frontier model in an enterprise’s own data, workflows, and guardrails.

In the end, regulated production turns on more than just technology, Saha said. Today, most enterprise AI projects fail during implementation because of poor integration, domain specialization gaps, lack of governance, and unclear ownership of outcomes. Last-mile specialization turns a capable foundation model into an enterprise agent shaped by domain-specific workflows, risk appetite, client classifications, regulatory interpretations, and institutional knowledge.

Why frontier models stall in enterprise workflows

Frontier models fall well short of production-grade accuracy on many real-world insurance workflows, Saha said, but last-mile specialization can lift them to the reliability enterprises need. Out of the box, those models struggle with the complexity of regulated workflows such as multinational insurance claims.

“These forms are pretty complex, often have handwriting, lots of checkboxes, and so on,” he said, and that complexity is why frontier models like Fable 5, Opus 4.8, and GPT-5.5 fall short out of the box.

Saha said the biggest gains come from specializing the entire AI system, not just the foundation model.

That system gets specialized with the customer’s data, workflow and, in many cases, the tribal knowledge that never made it into an operating procedure document.

“The biggest bang for the buck comes from the specialization and then these specialized guardrails,” he said.

The work has three components:

capturing the enterprise’s context and making it consumable by AI

running an ensemble of models so cost does not go through the roof

and adding specialized guardrails that check the model and force a redo when it gets something wrong.

What the last mile of agentic AI actually requires

None of this involves fine-tuning. VentureBeat’s latest enterprise survey found it ranked last among companies’ model-selection priorities.

Instead, the last mile centers on domain knowledge and undocumented workflows that companies would never expose publicly without losing their competitive edge.

“The last mile is about taking data that’s proprietary to you and using that to build a system around the model that can steer the model in the right way that can put the appropriate guardrails around it,” Saha said.

In the end, enterprise AI is about moving a workflow from point A to point B rather than deploying a technology, and NTT’s advantage comes from pairing AI experts with subject domain experts.

“The only reason is because we go and talk to those human workers and we say, ‘How do you actually do the work,’” he said. That expertise is then encoded into an agent.

Success in insurance, manufacturing, and other regulated industries relies on three things at once, he added.

“You need technology, you need the domain expertise, and you need the change management expertise,” he explained, adding that across his team’s clients, technology is not the bottleneck.

How enterprises turn AI investment into tangible value

For enterprises weighing large AI budgets, Saha’s said the payoff comes not from the model but from the work built around it.

“When you’re deploying AI in the enterprise, you’re not deploying a technology,” he said. “You are taking a workflow that exists and taking it from point A to point B.” The value is created by the workflow that gets moved, not the model that helps move it.

That reorders where money should go.

“Technology is not the bottleneck,” Saha said, pointing instead to the domain expertise and change management wrapped around the model, and to the discipline of commiting to all three together. Spending aimed only at the model leaves most of the return on the table.

Enterprises don’t have to choose between embedding AI into existing workflows and redesigning those workflows from scratch. NTT sees the two as successive stages of the same journey.

“We are starting with embedding in the workflow because it’s easier change management,” he said, noting that customers running mission-critical operations will not let a vendor rip out a working process midstream. “Once that happens, then we go into, how can we now reimagine this? And that really is where the biggest bang is.”

Where enterprise AI stays bespoke and where it becomes scalable

Keeping intelligence in the surrounding system rather than the model also preserves swappability and lets enterprises take advantage of open-weight and open-source models as they mature. Saha’s team runs an ensemble that mixes frontier and open-source models, and he expects the industry to lean on open weights wherever the cost of a mistake is low while reserving frontier reasoning for the cases that demand it.

“In many situations, especially in regulated industries where mistakes are very expensive, that last extra couple of percent matters,” he said.

The platform follows the same pattern: Guardrail generation and neurosymbolic models scale across customers, while capturing each organization’s tribal knowledge remains bespoke. Saha pointed to NTT DATA’s position as one of the world’s largest insurance third-party administrators as an advantage in acquiring that expertise.

“The ability to take that knowledge and trust that has been built over 20 years is very hard to replicate instantly, and I do think that is a durable aspect of what we have,” he said.

Sponsored articles are content produced by a company that is either paying for the post or has a business relationship with VentureBeat, and they’re always clearly marked. For more information, contact sales@venturebeat.com.

In my current season of toddler parenthood, I’ve sometimes felt that I could cut the tension between myself and the TV remote with a knife. As much as I try to keep screen time in check, sometimes I need 10 uninterrupted minutes to unload the dishwasher before a certain almost-2-year-old goes for the dog bowl.

That’s where audio players are handy. In a world where screens are increasingly hard to avoid, parents are looking for alternatives—and these players, in their various formats, are to the rescue. The devices mark a sweet spot between independent play and story time, letting kids listen to music, audiobooks, and interactive stories without a tablet in sight. Some audio players are geared toward younger kiddos who enjoy tactile toys and familiar characters like Elmo and Bluey, while other options are designed to grow with them to chapter books, podcasts, and beyond.

Over the past several weeks, my family has tested some of the biggest names in the category to see how they stack up. Here’s the verdict on which ones are best suited for different families and their needs.

Want more recommendations for WIRED-tested kids’ toys? Here’s our guide to Brain-Boosting Toys for Kids, Kid-Approved STEM Toys, and the Best Gifts for Lego Lovers.

Best for Kids of All Ages

The Yoto Player is simple enough for toddlers to operate independently while offering a broad enough content library to keep it relevant as kids grow. Stories and music live on credit card–sized cards that my daughter could swap in and out with her little hands, and the small pixel display on the front of the machine adds a visual element without becoming another distracting screen.

You can add each Yoto card to your child’s library through the device’s companion app. Once you do, you can stream the content directly from your phone or remotely send it to the Yoto. That means you’re not necessarily tied to bringing the speaker with you everywhere you go. I saw this capability as being especially useful on road trips—you can play stories through your car speakers or listen from your phone if the player gets left behind or its battery dies.

Yoto offers more than the stories and songs from its player cards. You’ll also get kid-friendly daily radio, draw-alongs, toothbrush timers, white noise, a wake-up clock, and a nightlight that activates when you flip the player screen-side down. I only see it getting more useful as my daughter grows, and if I’m being honest, there is a Nancy Drew story collection I am secretly eyeing for myself.

The incredibly useful feature is only present in specific headphone and earbud models.

Bluetooth multipoint reduces friction for multi-device setups

AYO Production/Shutterstock

Multipoint can create its own frustrations

Max Miller for Engadget

Hank Green, a novelist, comedian, and YouTuber with 3.2 million subscribers, recently apologized to his audience for his growing reliance on AI chatbots.

The controversy started when, in the middle of a video posted on the educational channel Complexly, Green incongruously used the phrase “I appreciate the pushback,” leading viewers to speculate that he’d written the script with a chatbot — and, in the process, accidentally included its response to one of his prompts.

In response to a since-deleted post on X, Green acknowledged that he’d produced the video “under a ton of pressure” and used ChatGPT “for research on this script.” At the same time, he said the “pushback” line was actually a response to the episode’s guest. 

Green then offered a more in-depth apology on Reddit, where he said he was “mortified” that he’d “let so many people down” and that he plans to reduce his video production as a result.

Green insisted that he’s only used ChatGPT to “locate papers and other resources for learning about topics,” and that the words and the “takes” have still been his. At the same time, he admitted it was fair to criticize him for “diluting” himself.

He also clarified that he’s “not a pure AI-hater,” while also listing a number of concerns about the technology, including its impact on climate change and “the speed at which these companies are trying to consolidate economic power.”

“Ultimately, what I am most scared of is ruining myself for people, but I have not been managing my impulses well,” he said. “You need to know that my words are mine. I don’t think that this hasn’t been true, but I’ve been moving so fast that my own process isn’t actually clear to me and I want to have it be a guarantee moving forward.”

It sounds like Green is going to take some time to figure out how to make that guarantee a reality. In the meantime, he’ll be pausing or posting less frequently to his various YouTube channels. Green said he wants to do more work like a recent meditative video “where the writing was the whole thing and I felt it all the way down.” Green also said he’ll probably make more videos with his “dumb unscripted straight to camera thoughts.”

 “But mostly I need to come to terms with the fact that the level of dopamine I’ve been getting from interacting with LLMs…with doing more and more and more and more…is not healthy for me or good for the world,” Green said. “It is careless, and has disconnected me from where people are on this.”

Several popular Samsung smart TV apps contain code that share the owner’s internet connection with strangers, potentially putting millions of Samsung smart TVs at risk of hijacking, according to new security research published on Monday. 

Some of these apps claim to have been installed on hundreds of millions of smart TVs in people’s homes, per the app developers.

At least one of the smart TV apps was a simple Pac-Man game that Samsung had endorsed and prominently featured in its “Editor’s Choice” section on customers’ TV screens.

These apps contain software that funnels outsiders’ web traffic through ordinary home and office internet connections, known as residential proxy networks (or “resproxies”), which are increasingly being linked to cybercrime. When opened, apps with resproxy code can turn the smart TV into an always-on tunnel for outsiders to funnel their web traffic through, known as an exit node — even when the app is no longer open. 

The security research by Norwegian cybersecurity company Mnemonic describes a perfect storm of problems that allows low-quality apps to proliferate across Samsung’s app store, containing code that puts users at risk of having their internet connections tapped by a rogue app. 

Many of these apps are bare-bone shells, made from only a few lines of code, and are designed solely to load content from another website, such as a game. While such smart TV apps load content from another server, any review of these apps sees only the few lines of code within, and not necessarily the content itself.

“What was reviewed is not necessarily what is running,” wrote Harrison Sand, an offensive security consultant at Mnemonic. 

After TechCrunch contacted Samsung with a request for comment about the research, the electronics giant said in an emailed statement that it was banning apps that share their users’ internet connections, and will remove apps that contain the functionality.

“We have already restricted new app registrations that incorporate such proxy functionalities on our Smart TV platform,” said a Samsung spokesperson. “We are currently implementing strict platform-wide developer policies explicitly banning residential proxy SDKs, and we are working to identify and remove all apps currently available in our store that contain these components.”

The move comes after LG said last month that it would ban apps that contain resproxy software after recent reporting found that around 42% of apps on the company’s app store enlisted a smart TV into a proxy network.

Inside a residential proxy network

The research also offers a rare look inside a residential proxy network.

Resproxy code can also be found in regular consumer phone apps, as well as other consumer electronics, like digital frames and Android streaming boxes, which then share that device’s internet connection. 

Any time a resproxy app or device connects to the internet, an outsider can also pay to use it. 

Resproxies are not inherently illegal. Some are used for evading censorship by routing internet traffic through ordinary looking residential homes. AI companies, for example, increasingly rely on resproxies to scrape data from multiple places on the internet in one go to train their AI models.

But cybersecurity companies say resproxies have gained a reputation for allowing hackers and spies to carry out cyberattacks and data breaches while hiding their malicious activity. 

Cybersecurity companies find resproxies challenging to tackle because the network traffic looks like it’s coming from an ordinary household, rather than a malicious hacker located overseas, as they might expect. 

Moreover, the network traffic that flows through a user’s device over resproxies is generally encrypted, which is generally impossible to unscramble and inspect.

By rooting a Samsung smart TV’s software, Mnemonic’s Sand gained deep access to the television’s internals and analyzed all of the network traffic that flowed in and out of the TV. This included any app that was sharing the smart TV’s internet connection with someone else. 

He found the Pac-Man game contained resproxy code from Bright Data, an Israel-based company that provides proxy networks touting access to millions of residential networks around the world. The company also has a marketplace for selling access to scraped datasets. These datasets are derived from a network of enlisted smart TVs as exit nodes, which are used to download large amounts of public data from the web from multiple sources at once, often to circumvent systems designed to prevent scraping.

Sand found that Bright Data’s resproxy code loaded when opening the Pac-Man game, but noted that this did not automatically turn the Samsung smart TV into an exit node. Sand said the resproxy code is dormant until the user accepts a consent screen, which immediately activates the resproxy code to run in the background until the user deletes the app.

Aside from the user themselves consenting to enlisting their device into a resproxy, Sand warned that a “simple code change on a web server” could instantly activate hundreds of millions of smart TVs into a potentially malicious botnet.

With access to the network data flowing through his smart TV, Sand could see that much of it appeared to suggest the resproxy network was used for large-scale scraping of LinkedIn profiles, and for collecting AI training data. Sand said he only saw a tiny percentage of what was routed over Bright Data’s network. 

Bright Data did not respond to a request for comment.

Apple TV‘s adaptation of Cape Fear saw a chilling finale recently, and it was as grisly as we’d expected it to be. Over the past few weeks we’ve followed Javier Bardem’s take on Max Cady, delivering a performance so scary it’s left some of us sleeping with the lights on.

Cady was particularly unhinged in Cape Fear‘s finale, which saw him killing his own family and taking the Bowdens hostage in one final altercation.

Many small businesses already tap into AI for market research, image and video creation, copywriting, and idea exploration.

Although this single-prompt approach is helpful, it has its limits. Enter AI agents.