If you haven’t already seen a job listing for a “product engineer,” you probably will soon. The job everyone’s suddenly hiring for, this role is like a cross between a product manager and an engineer (as the name suggests). And it’s a hiring trend worth paying attention to.

Companies are opening more of these roles every single month, but they’re struggling to fill them. The reason has almost nothing to do with engineers’ coding skills or years of experience.

The best career move you can make to prepare for these types of roles has almost nothing to do with getting more technical. Instead, it comes down to one of the fluffiest, most overused, and potentially cringiest words in all of tech: mindset.

Stick with me, I promise this goes somewhere useful.

The problem: We were trained to be task-takers

When I started out, my job looked like this:

Drive to an office. Sit through meetings that led to other meetings until a project manager handed me a task they’d already chopped into tiny pieces.

My job was to turn that task into code.

It took years for me to get good at a coding language and tech stack, and once I did, I executed that knowledge against specs that somebody else wrote.

You know what’s freakishly good at that exact job? I’ll give you a hint: It starts with A and ends with I.

Boris Cherny, the creator of Claude Code, recently said: “coding is basically solved,” and “the bottleneck is going to be good ideas.”

So if your entire value is “hand me a task and I’ll build it,” you’re in a footrace with the robots. I don’t like that for you.

The bad news… that is also good news

Many companies are flattening. Middle management is getting stripped out, for better or worse (mostly for worse), which means many of us are doing more with less.

This might sound like purely more work, but it’s also an opening for anyone who cares about what they’re building and can put on their manager hat. Companies are no longer just hunting for the strongest engineer in one narrow domain.

What’s rare, and what actually moves revenue, is an engineer who can spot the thing that’s quietly costing money and either flag it to leadership or just go fix it.

What this actually looks like

Being product-minded has NOTHING to do with your tech stack.

Here’s where to start:

Have an opinion and back it up. As a former engineering manager, the worst thing I ever heard was silence. I’d often ask the team what they thought because I doubted myself and wanted a gut check. I was grateful to the ones who said “nope, bad idea, here’s why.” Pushback is a gift.

Learn the domain, casually. Work for a plumbing company? You don’t need to become a plumber, but spend an hour on Reddit threads where plumbers vent. Now your ideas come from your potential customers.

Make experiments cheap and safe. This is where any engineer has massive leverage. Experiments are not free. A bad one loses customers and frustrates users. Tools like LaunchDarkly and Optimizely let you ship a change to 5 percent of users and roll it back the second it tanks. Learn them, or build a scrappy version yourself. A team that can quickly run safe experiments will out-learn everyone else in the building.

Be data-driven. Stop fighting about button colors. Pick a goal: making money, finding product-market fit, or making the product sticky so people come back. Then measure it. If your gorgeous redesign tanks time-on-site, it failed, no matter how good it looked to you. If the ugly version makes more money, ship the ugly version.

You don’t have to be the ideas person. Maybe you’re not a visionary. That’s fine. Organize a hackathon around an actual company goal. Pull up your company’s quarterly targets and build something against one of them. Don’t know what those targets are? That’s your first assignment.

Good ideas are the new bottleneck—and they always have been

When I was a manager, I asked myself one question every week: What’s the single most impactful thing I could do right now? The answer was almost never “write more code.” It was understanding a gnarly problem nobody had defined yet. Building a deck to spread knowledge that was in one person’s head. Getting the right three people in a room to actually make a decision we’d been putting off.

Code is cheap, and it always has been. We just couldn’t see it, because for decades the typing took so long that it felt like the hard part. It never was. The hard part was always knowing what’s worth building.

— Brian

In January 2025, Siobahn Day Grady launched the first AI research institute at a historically Black college or university. The institute aims to help expand AI skills for all students at North Carolina Central University, where Grady is an associate professor, through both AI research opportunities and skills training. Though the institute is the first of its kind, Grady hopes it could serve as a model for other HBCUs.

Read more here.

AI is increasingly used in the scientific research process. So does publishing need to change to keep up? Jiachen Liu recently co-authored a paper published on ArXiv arguing that the PDF should be replaced with an “Agent-Native Research Artifact” designed with AI in mind. In this interview with IEEE Spectrum, Liu lays out a provocative vision of AI-driven research and an infrastructure that captures—and learns from—details that often get left out of today’s papers.

Read more here.

Astronomers still don’t know exactly what dark matter is, but they can detect it—and so can you. With a small radio telescope and a few other pieces, you can create a DIY setup to gauge how fast hydrogen clouds are moving across the Milky Way. Feed those measurements into a spreadsheet, and you can see the same signals that have baffled the astronomical community for decades.

Read more here.

The amorphous corporate mascot will no longer haunt Copilot Voice.

Need to upgrade your phone but aren’t sure whether to splurge on a foldable or not? You’ve come to the right place.

We’ve compared Google’s newly announced Pixel 11 Pro to its new book-style foldable, the Pixel 11 Pro Fold, and noted the key differences that are likely to sway your decision. 

Once you’re finished here, make sure you visit our Pixel 11 vs Pixel 11 Pro guide, while our Pixel 11 Pro Fold vs Galaxy Z Fold 8 Ultra compares the foldable to one of our current favourites.

Finally, if you’re not sold on Google’s Pixel line-up, visit our best smartphone and best foldable phone round-ups instead.

Specs Comparison Table

	Google Pixel 11 Pro	Google Pixel 11 Pro Fold
Chipset	Google Tensor G6	Google Tensor G6
Colours	Canyon, Olive, Fog, Obsidian	Olive, Obsidian
Front Camera	42MP	10MP + 10MP
IP Rating	IP68	IP68
Operating System	Android 17	Android 17
RAM	12GB	16GB
Rear Camera	50MP, 48MP and 48MP 5x	48 MP, 10.5 MP and 10.8 MP 5x
Screen Size	6.3-inches	6.5 (external) and 8-inch (inner)
Storage	256GB, 512GB, 1TB	256GB, 512GB, 1TB
UK RRP	£1099	£1799
US RRP	$1099	$1999

Advertisement

Price and Availability

At the time of writing, both the Pixel 11 Pro and Pixel 11 Pro Fold are available for pre-order and will officially launch on August 20th. 

The Pixel 11 Pro is the cheaper of the two, with a starting RRP of £1079/$1099 for the 256GB handset. Alternatively you can opt for the XL iteration, which will cost you from an additional £200/$200. To learn the differences between the two, our Pixel 11 Pro vs Pixel 11 Pro XL guide will help. 

SQUIRREL_PLAYLIST_10208683

Advertisement

The Pixel 11 Pro Fold is the most expensive of Google’s line-up, with a starting price of £1799/$1999 for its 256GB model.

SQUIRREL_PLAYLIST_10208685

Pixel 11 Pro Fold has two displays

Let’s get the most obvious difference out of the way. While the Pixel 11 Pro is a regular “candybar” smartphone with just one screen, the Pixel 11 Pro Fold is a book-style foldable that’s equipped with two. 

Advertisement

Advertisement

Its external display is larger than the Pixel 11 Pro’s own, measuring at 6.5-inches compared to 6.3-inches. There are a few other differences with the screens too, as the Pixel 11 Pro Fold’s has a 19.5:9 aspect ratio whereas the Pixel 11 Pro’s a more standard 20:9 ratio. In addition, the Pixel 11 Pro is fitted with Gorilla Glass Victus 2 whereas the Pixel 11 Pro Fold has an “ultra-durable ceramic cover glass”. Otherwise, both displays promise to be Pixel’s brightest ever, with a peak of 3600 nits.

Open the Pixel 11 Pro Fold and you’ll find its mighty eight-inch panel. Like its external screen, the Pixel 11 Pro Fold’s internal display promises up to 3600 nits of peak brightness and HDR support too. It also benefits from a redesigned hinge that’s promised to be stronger yet lighter than the Pixel 10 Pro Fold’s own – but we’ll wait until we review the Pixel 11 Pro Fold to verify this.

Pixel 11 Pro promises slightly longer battery life

Google has never been known to equip its Pixel handsets with particularly mighty batteries, and the Pixel 11 series is no exception. However, the Pixel 11 Pro slightly bests the Pixel 11 Pro Fold, with a 4850mAh cell that promises over 30 hours of life. In comparison, the Pixel 11 Pro Fold has a slightly smaller 4806mAh capacity that should see around 24 hours between charges.

Otherwise, both support 30W fast charging and are Qi2.2-certified for up to 25W wireless charging too.

Advertisement

Pixel 11 Pro has a Pro rear-camera set-up

Although both handsets have three rear cameras, they differ with their exact offerings. The Pixel 11 Pro has a Pro camera system that’s made up of 50MP main, 48MP ultrawide and 48MP 5x telephoto lenses. It also supports Pro Zoom up to 120x and 8K video recording too.

Advertisement

In comparison, the Pixel 11 Pro Fold has an Advanced camera system that includes a 48MP main, 10.5MP ultrawide and 10.8MP 5x telephoto set-up. While it can’t quite reach 120x zoom, it does support Super Zoom of up to 30x.

That’s not where the camera differences end. While the Pixel 11 Pro has just one 42MP selfie lens, the Pixel 11 Pro Fold has two. Both are 10MP, but one sits on the external display while the other can be found on the inner panel.

As we’re yet to review either the Pixel 11 Pro nor the Pixel 11 Pro Fold, we can’t verify how well they can actually shoot photos. However, given that Pixel phones usually end up in our best camera phones list, and Google claims the Pixel 11 Pro in particular sports its best camera yet, we have pretty high expectations. 

However, and like all Pixel handsets, the hardware is just half the story when it comes to photographic ability, as Google also has a suite of photo editing software that’s genuinely brilliant. Alongside Camera Coach, which gives you live tips on how to take a better photo, there’s tools for post-production tweaks that allows you to move and even completely remove objects from images.

Advertisement

Advertisement

Both run on Tensor G6

Something that unites the entire Pixel 11 series is the fact all phones run on Google’s Tensor G6 processor. Like its predecessors, Tensor G6 prioritises AI capability over sheer performance and enables the running of Google’s extensive AI-powered toolkit. This toolkit is arguably one of the reasons to opt for a Pixel, as it covers everything from the aforementioned photo editing tools to productivity tasks like Circle to Search and access to voice assistant Gemini. 

Google promises that Tensor G6 will see faster web browsing and app launches compared to its predecessor, and will run on-device AI tasks up to 3.5x faster than before too. We’ll be sure to update this versus once we review the chip. 

Both have IP68 ratings

While it’s pretty standard for a smartphone to sport an IP68 rating, which means it’s dust-tight and can survive continuous submersion in water, the Pixel 11 Pro Fold is one of the only foldables to boast such a rating. 

Not only that, but the Pixel 11 Pro Fold is also made from multi-alloy steel with an aerospace-grace, high-strength aluminium alloy cover, while its ultra-durable back has a satin-finish, spacecraft-grade aluminium frame too. While we haven’t reviewed the Pixel 11 Pro Fold, it certainly sounds like a durable handset.

Advertisement

Early Verdict

Choosing between the Pixel 11 Pro and Pixel 11 Pro Fold will ultimately come down to how you want to use a smartphone. If you want a foldable phone, whether that’s because you want a larger screen for watching videos, gaming or even for casual working, then the Pixel 11 Pro Fold will appeal to you. On the other hand, if you aren’t sold on a foldable, and want a standard smartphone that’s fitted with a promising camera set-up, longer battery and AI smarts, then the Pixel 11 Pro is a no-brainer.

Advertisement

We’ll update this versus once we review both handsets.

Advertisement

South Korea’s hills once stood where steel skeletons now climb toward the sky. On a July morning last month, a small group of visitors stepped onto ground that crews had flattened from mountains. What rose around them was the first section of what SK Hynix calls the largest memory factory complex on Earth. Walls already reached halfway up a structure planned to match a 50-story apartment tower. Six cleanrooms will stack inside that single building, connected floor to floor. Production is set to begin in February.

The Yongin Cluster is where corporate representatives gave reporters a behind-the-scenes tour of the unfinished floors while presenting a somewhat ambitious $720 billion plan to increase overall capacity by 2034. Four factories will eventually occupy a large location of approximately 45 million square feet, which is roughly comparable to 600 soccer fields. The fundamental reason for the vertical form is more than just esthetics; roughly 70% of the country is mountainous, making it unfeasible to spread out like some chip facilities in the Arizona desert.

Myeong-jae Park, SK Hynix’s Head of HBM Design, conducted this tour and pointed out the steel and concrete taking shape, recounting the story of how the company placed an early bet on high-bandwidth memory about 7 years before AI processors came up and there was a race for it. They introduced the first commercial version of the technology in 2013, despite the fact that relatively few devices required this stacked design at the time, but they had a strong belief, which propelled the project forward. Fast forward to today, and such stacks now feed the processors that power the large AI models.

Two hours south of Seoul, they rolled out the red carpet for media to have a closer look at a site where work had already been completed. The M15X fab in Cheongju is a 33-floor building that opened in 2025. Visitors must first use an air shower to remove all outside dust before donning a full cleanroom suit. Once inside, the space is like to entering into a chamber the size of 13 football fields. Woo-jin Choi, head of packaging and testing, showed them the ropes as they strolled by these overhead hoist systems that shuttle wafer carriers thru the air; in this one building alone, there are 700 of these machines in use, with each costing up to $70,000. They can transport 25 wafers at once, and the company claims this is the world’s longest cleanroom track, at least in relation to the adjacent Icheon complex.

Choi came to a station where the layers of DRAM are all linked together to form finished high-bandwidth memory, and the process of producing the latest HBM4 takes four months and more than a thousand distinct processes. Now one of those procedures, called bulk reflow mold under fill, is truly unique to SK Hynix. Multiple dies are connected simultaneously under low pressure, and a protective liquid fills the spaces between the layers. Result? A stack that’s around 40 percent thinner than what they used to do it and better at managing heat, and you do that by squeezing in 380 billion tranistors into a single module. The same plant is already running HBM3E alongside the newer generation, and the next chip on the horizon is HBM5, which has been designed with specific clients in mind. Now, NVIDIA and Google have each requested customized versions that match their own processors. This shift is what is causing memory to become less interchangeable and generic than it was previously.

At SK Group headquarters in Seoul the following day, Chairman Chey Tae-won held up a finished HBM4 chip and a wafer that carried a handwritten note from Nvidia’s Jensen Huang: “Please make more.” Chey described the current demand in simple terms. “It’s like a war. Everybody wants the memory chips. Without that, they cannot produce their AI computing and AI chips.”

In another reshuffle of its video game division, Netflix is closing the first studio it purchased, Oxenfree team Night School, according to Stephen Totilo at Game File. Netflix acquired Night School Studio in 2021, and there it produced Oxenfree II: Lost Signals, the Black Mirror: Thronglets app, and Unhinged, a smartphone-controlled horror game that landed roughly six weeks ago.

Netflix also reportedly plans to shut down Helsinki studio Moonloot Games, which the company founded in September 2022 to build original mobile titles. Additionally, it’s slashing an undisclosed number of jobs from its in-house games team. This is all in the name of prioritizing kids gaming, party games, narrative experiences and mainstream titles like June’s FIFA World Cup: Launch Edition, a Netflix spokesperson told Game File.

The FIFA World Cup game was developed by a third-party studio, Refactor Games, which laid off 85 percent of its staff in early August. The layoffs followed reports that Refactor’s parent company, Delphi Interactive, pulled funding after the game’s release.

In a real salt-rubbing move, Netflix co-CEO Gregory Peters praised Night School’s game Unhinged and FIFA World Cup during the company’s Q2 2026 earnings call on July 16.

“We’ve been building some solid foundations, and now we’re seeing exciting positive signals that help inform and give us increased conviction in our future growth and the nature of that growth here,” Peters said just a few weeks ago. “So you mentioned the cloud-based strategy, those cloud-based TV games, we really see it working. FIFA and Unhinged became our two most successful cloud game debuts, really solid numbers that put it in the top tier of game performance for us.”

Night School co-founders Sean Krankel and Adam Hines were supremely optimistic about the Netflix acquisition when I talked with them at Summer Game Fest 2023, just prior to the release of Oxenfree II. Netflix decided to establish a video game business in 2021, and according to Krankel, that’s when executives wooed the Night School founders with lines like, “What can we do to unblock your team from making your dreams?” 

Netflix officially acquired Night School in September 2021. After the purchase, Night School hired more team members and moved into the Netflix offices. In 2023, Krankel and Hines talked about how cool it was to be able to fly in remote workers as needed, and to make Oxenfree II available in 32 languages on day one. The founders discussed how at ease they felt at Netflix.

“Our big concern was the autonomy aspect,” Hines said at the time. “We’ve all worked at bigger studios before, and have just seen and felt how long it would take to get decisions made, how the creative would kind of get choked out of things because there’s too many cooks in the kitchen. But just talking to Netflix a lot before we joined up, we felt really at ease, just like we were talking the same language about how to make games.”

Netflix rapidly expanded its video game division between 2021 and 2024, purchasing a handful of teams including Cozy Grove creator Spry Fox and mobile developer Boss Fight Entertainment, and staffing an internal AAA studio with veterans from Overwatch, Halo and God of War franchises. Then in 2024, Netflix hired former Epic Games executive Alain Tascan to lead its games business, and everything changed.

With this week’s closure of Night School and Moonloot, Netflix has now shut down or sold all but one of its studios. Spry Fox, at least, was able to purchase itself back from Netflix and is still making ambitious, adorable games. The sole survivors at Netflix are Helsinki-based Next Games, which focuses on couch-based experiences, and Netflix Games Studio, which mostly signs deals with external developers.

That group seems to be going strong, with a slew of streamable party games like Overcooked! All You Can Eat and Jackbox titles hitting Netflix in recent months.

Google is rolling out Gemini 3.7 Flash, a new version of its workhorse AI model that puts coding, agentic workflows and knowledge work at the center of the upgrade — while temporarily cutting API prices in half.

The release arrives just three weeks after the release of Gemini 3.6 Flash, an unusually short turnaround that Google attributes to developer feedback and algorithmic improvements.

For enterprise developers, the more consequential story may be the combination of those intelligence gains with lower inference costs: through the end of 2026, Gemini 3.7 Flash costs $0.75 per million input tokens and $3.75 per million output tokens.

Starting Jan. 1, 2027, pricing rises to $1.50 per million input tokens and $7.50 per million output tokens. That means the current discount is temporary, but it gives teams deploying high-volume coding and business agents several months to evaluate whether Google’s claimed reductions in retries and manual oversight translate into lower total operating costs.

The launch also underscores Google’s rapid iteration on its Flash line while its next flagship Pro model remains absent. Google did not provide a release date for Gemini 3.5 Pro with Thursday’s announcement, Reuters reported, despite the model having previously been described as undergoing partner testing. Axios similarly noted that 3.7 Flash arrives before the anticipated Pro release.

A three-week upgrade focused on getting work done

Google describes Gemini 3.7 Flash as its “most intelligent workhorse model yet for coding and agents.” The company says the model is better at adapting when it encounters roadblocks, clarifying intent when necessary and following instructions with greater fidelity.

Those improvements matter beyond benchmark scores. In an enterprise coding agent, a model that makes fewer unnecessary changes, recovers from errors and executes multi-step plans more reliably can reduce the number of human interventions needed to complete a task. The same principle applies to business agents operating across documents and applications, where an incorrect tool call or poorly interpreted instruction can derail an otherwise useful workflow.

Google says 3.7 Flash “thinks more diligently,” applying more effort to multi-step planning and tool calls. Its stated goal is more disciplined execution with fewer retries and less manual supervision.

That represents an interesting evolution from Gemini 3.6 Flash. Google’s developer documentation described 3.6 as reducing reasoning steps, conversational turns and tool calls compared with earlier models while attempting to limit execution-loop spiraling. With 3.7, the emphasis shifts toward putting sufficient effort into planning while improving the quality of execution — potentially a more useful optimization than simply minimizing the number of steps an agent takes.

Google DeepMind said in a post accompanying the release that 3.7 Flash shows gains in debugging and issue resolution, generates more functional web layouts and applications with fewer prompts, and improves reasoning and accuracy on real-world business workflows.

Coding gains are substantial, but not universal

Google’s benchmarks show a large generational improvement in several software engineering tests.

On FrontierCode 1.1 Main, which measures production code quality, Gemini 3.7 Flash scores 43.6%, up from 34.4% for Gemini 3.6 Flash. That also narrowly exceeds the 42.7% Google reports for Claude Sonnet 5 and 41.3% for GPT-5.6 Terra.

On DeepSWE v1.1, a long-horizon software engineering evaluation, 3.7 Flash reaches 65.3%, compared with 49.0% for its predecessor. GPT-5.6 Terra remains ahead at 69.6% in Google’s table.

Web development shows another notable gain. Gemini 3.7 Flash receives an Elo score of 1588 on Code Arena, versus 1538 for 3.6 Flash, 1541 for Claude Sonnet 5 and 1523 for GPT-5.6 Terra. Google says the new model can produce more functional layouts and feature-complete applications in fewer prompts while more closely following reference screenshots, images and design systems.

The broader benchmark table is more mixed, which is important for enterprises evaluating the model against particular workloads rather than looking for a single “best” model.

Gemini 3.7 Flash scores 85.8% on Terminal-bench 2.1, compared with 87.4% for GPT-5.6 Terra. Terra also leads Google’s comparisons on Terminal-bench 3.0 and OSWorld-2.0. Claude Sonnet 5 leads the Agent’s Last Exam multimodal desktop and operating-system tasks with a 33.3% pass rate, versus 26.3% for Gemini 3.7 Flash.

In other words, Google’s own results do not show 3.7 Flash universally displacing higher-priced competitors. They instead suggest a model that has become substantially more competitive in coding and agent workloads while occupying a lower price tier.

Enterprise workflows may be the more important test

The gains extend beyond software development.

On AutomationBench, which Google describes as measuring enterprise workflow automation, Gemini 3.7 Flash scores 30.4%, up sharply from 17.0% for 3.6 Flash. Google’s table lists Claude Sonnet 5 at 10.7% and GPT-5.6 Terra at 23.6%.

The model also reaches 34.0% on GDP.PDF, an evaluation of complex PDF comprehension, compared with 22.0% for 3.6 Flash, 28.0% for Claude Sonnet 5 and 24.7% for GPT-5.6 Terra.

That combination is relevant for enterprise agents because many practical deployments require more than generating text or code. An agent may need to interpret a long report, identify relevant information, decide which tool to invoke, update another system and produce a document for a human reviewer. Reliability across that chain can matter more than performance on an isolated reasoning benchmark.

Google is putting that thesis into practice with Gemini Spark. Google AI Pro and Ultra subscribers can use 3.7 Flash in Spark, the company’s personal AI agent. Google says the upgrade improves Spark’s knowledge work and tool use across Google Workspace applications, including workflows that consolidate files, draft emails and update status documents.

For enterprises, 3.7 Flash is also available through the Gemini Enterprise Agent Platform and Gemini Enterprise app.

Price becomes part of the model competition

Gemini 3.7 Flash’s introductory pricing is a notable bid to embed the model into enterprise workflows.

Until Dec. 31, developers pay $0.75 per million input tokens and $3.75 per million output tokens. Context caching costs $0.075 per million tokens during the introductory period. Google says standard prices will double on Jan. 1, 2027, to $1.50 for input and $7.50 for output, with context caching rising to $0.15.

For comparison, Gemini 3.6 Flash’s standard API pricing is $1.50 per million input tokens and $7.50 per million output tokens. Google’s benchmark table lists Claude Sonnet 5 at $2 and $10, respectively, while GPT-5.6 Terra is listed at $2 and $12.

The economics become more pronounced for autonomous agents because a single user request can produce a long sequence of model calls, reasoning tokens and tool interactions. A model that costs less per token but requires substantially more retries may not ultimately be cheaper.

Conversely, Google’s combination of lower introductory token pricing and claimed improvements in first-pass accuracy could materially change the cost of running high-volume coding or document-processing agents if those gains carry over to production.

That is the metric enterprise teams will ultimately need to test: not price per million tokens in isolation, but cost per successfully completed task.

Google’s AI shake-up raises the stakes for Gemini

Gemini 3.7 Flash arrives amid a broader debate over whether Google is losing ground at the AI frontier. The company has not released Gemini 3.5 Pro, despite saying in May that the flagship model would arrive the following month.

By July, Google said it remained in partner testing and would become broadly available when ready; Thursday’s announcement offered no further timetable. Google’s latest released general-purpose Pro model therefore remains Gemini 3.1 Pro, introduced in February.

Reuters reported in July that Gemini 3.5 Pro missed its original target after falling short of internal goals, particularly in coding, even as Google began training what it calls its most ambitious model yet, Gemini 4.

The delay coincides with a major overhaul of Google’s AI leadership announced last week.

Google DeepMind co-founder and Nobel Prize Winner Demis Hassabis has relinquished day-to-day control of the company’s famed DeepMind AI division to become its chair and, simultaneously, to take on the role of Alphabet’s chief scientist.

Meanwhile, former DeepMind CTO Koray Kavukcuoglu now runs the unit as a senior vice president reporting directly to CEO Sundar Pichai.

Kavukcuoglu controls Gemini model development, frontier research, the Gemini app and developer teams—effectively consolidating the full Gemini chain under a more product-focused operator.

Chief scientist Jeff Dean, Gemini co-lead Oriol Vinyals, Quoc Le and Sanjay Ghemawat left to establish the research startup Discovery Loop.

Those exits followed Gemini co-lead Noam Shazeer’s move to OpenAI and Nobel Prize-winning AlphaFold scientist John Jumper’s departure for Anthropic. Reuters reported that internal disagreements, constrained compute allocation and Google’s bureaucracy contributed to slower releases and weaknesses in coding.

Outside interpretations range from organizational repair to a more fundamental retreat.

SemiAnalysis has argued that Google is increasingly prioritizing the highly profitable business of supplying cloud infrastructure to AI companies—including Gemini competitors—over keeping its own models at the absolute frontier. That analysis also claimed Google had effectively canceled 3.5 Pro, although Google has not confirmed that and continues to describe the model as delayed.

The Verge offered a more measured assessment: the departures and model delays are serious, but Google retains enormous advantages through Search, Workspace, Android, Cloud, custom AI chips and consumer distribution. Google says the Gemini app has surpassed 950 million monthly users, giving it a reach that does not depend entirely on owning the highest-scoring model.

Current benchmarks similarly depict a company behind the overall leaders but still firmly competitive. Artificial Analysis places Claude Opus 5 at 63 on its overall model Intelligence Index, while Google reports a score of 56 for Gemini 3.7 Flash—an improvement from 52 for 3.6 Flash but not a return to the top.

Arena’s early human-preference results are more favorable, provisionally ranking 3.7 Flash ninth overall and eighth for web development.

The resulting picture is not that Google has abandoned advanced AI, but that it has become stronger at rapidly shipping efficient Flash models while struggling to deliver the premium flagship required to reclaim broad leadership. Gemini 4 will now serve as the clearest test of whether the leadership reorganization fixes that execution gap.

Available now across Google’s developer stack

Developers can access Gemini 3.7 Flash through the Gemini API in Google AI Studio and Android Studio, as well as Google’s Antigravity environment. Enterprises can deploy it through Gemini Enterprise Agent Platform and Gemini Enterprise, while consumers with Google AI Pro or Ultra subscriptions can access the model through Spark in supported countries.

Google is also shipping updated safeguards covering chemical, biological, radiological and nuclear risks and cyber-offense misuse, according to the company.

The unusually fast jump from Gemini 3.6 Flash to 3.7 Flash points toward a model development cycle in which algorithmic improvements can reach production products without waiting for a new flagship generation. Ars Technica also highlighted the three-week interval between the two releases, while Google says the techniques behind the update will inform future models.

For developers, that faster cadence creates its own operational question. Models can improve quickly, but production teams still have to benchmark new releases against their own repositories, prompts, tool schemas and failure modes before changing a deployment.

Gemini 3.7 Flash gives those teams a particularly strong incentive to run that evaluation. Google’s own numbers show major improvements in production coding, web development, document comprehension and workflow automation without claiming leadership everywhere. At its introductory price, Google is effectively betting that developers will value a model that is competitive enough with more expensive systems while being cheap enough to run repeatedly inside agents.

Whether that advantage survives the return to full pricing in January will depend less on leaderboard positions than on how reliably 3.7 Flash completes real work.

Databricks has closed a $5bn round at a $190bn valuation, led by Coatue, with revenue run-rate past $7bn and growth above 80% year on year. That is a 42% valuation increase in six months, and it comes after chief executive Ali Ghodsi called 2026 a bad year to go public.

Databricks has closed $5bn at a $190bn valuation. Coatue led, joined by Blackstone, MGX, T. Rowe Price and new investor Sixth Street Growth. It is the company’s second round this year.

The growth is the part worth pausing on. Revenue run-rate has passed $7bn, up more than 80% year on year in the second quarter, against 65% growth at a $5.4bn run-rate back in February. Companies of this size do not usually accelerate.

The valuation has moved with it, up 42% from $134bn in February. TNW reported the round at $188bn last month, and it has closed $2bn above that.

Databricks also disclosed figures that rarely accompany a private raise. It says it has been adjusted free cash flow positive over the last twelve months, its data warehousing business is past a $1.5bn run-rate and growing over 100%, and its Lakebase database has passed $100mn.

The customer concentration is heavy at the top. More than 1,000 accounts now spend at a $1mn run-rate, and more than 100 at $10mn.

Set against the public market, the price is less extravagant than it sounds. At $190bn on a $7bn run-rate, Databricks is valued at roughly 27 times revenue, while Snowflake trades near 23 times on $5.03bn of trailing revenue.

The growth rates are nothing like each other. Snowflake grew around 30% last year, Databricks says more than 80%, so a four-point multiple premium is a modest reward for nearly three times the pace.

Read those numbers with care, because they are not the same measure. Run-rate annualises current revenue and flatters anything growing quickly, while Snowflake’s multiple rests on twelve trailing months.

Why this is private money rather than a listing is already on the record. Ghodsi has called 2026 a terrible year to go public, with SpaceX, OpenAI and Anthropic lined up to absorb roughly $200bn of listing capital.

So the company raises at public scale and stays private. The money goes to three products aimed at enterprise AI agents, and Ghodsi’s pitch is that buyers want agents that hold context, stay accurate and respect a budget, rather than another chatbot.

Evergreen Candy Crush is one of the most successful and enduring mobile game franchises ever created. Launched by Swedish video game developer and publisher King in 2012, the original Candy Crush Saga quickly climbed the popularity charts on both Android and iOS. Nearly a decade and a half after its launch, it still boasts millions of users and generates billions of dollars in revenue.

In FY 2025, Candy Crush Saga reportedly generated $876.5 million in revenue, with most of it coming from in-app purchases. While that sounds impressive, it marks a slight dip from the previous six years, when the game generated more than $1 billion in annual revenue.

Candy Crush Saga follows a freemium model, meaning it’s free to play, but players can purchase extra lives, boosters, and special passes to make it easier to progress through its more than 23,000 levels.

While King hasn’t disclosed detailed revenue and profit figures in recent years, it reported surpassing $20 billion in revenue in 2023. That same year, the company also claimed to have reached 5 billion installs across all Candy Crush titles, including the original Candy Crush Saga and its spinoffs Soda Saga, Jelly Saga, and Friends Saga.

Players at the Candy Crush Saga All Stars esports event in Los Angeles in 2025.

Research by David Curry, data editor at app analytics platform Business of Apps, suggests that the original game currently has around 81.5 million monthly active users and was installed more than 190 million times across Android and iOS in 2025.

In its 2014 IPO filing, King claimed that Candy Crush Saga had 93 million daily active users worldwide in December 2013. If Curry’s estimates are accurate, this would suggest that the game has not experienced a massive decline in its user base over the past 12 years.

Candy Crush’s enduring popularity remains something of a mystery among the mobile gaming community. However, some believe its success can partly be attributed to its “turn-based” gameplay, which allows players to put the game down at any time and pick it up where they left off without penalty.

Paula Ingvar, general manager of Candy Crush Saga, attributes the game’s success to giving players “the satisfaction of making progress with small wins” in 10- to 20-minute sessions, rather than forcing them to finish a level in one sitting.

David Nieborg, a professor of media and platform studies at the University of Toronto and a video game researcher, believes the game’s popularity also has a lot to do with the fact that it is “non-place-based, nongendered, nonracialized,” helping it appeal to players across geographical, linguistic, cultural, and socioeconomic boundaries.

With the revelation that ICE will spend up to $20 million by March on gloves that can deliver painful electric shocks to subdue an individual, according to a notice published earlier this week by the Department of Homeland Security, it’s worth looking at how this apparel, designated as a nonlethal tool for law enforcement, actually works.

These shock gloves go by the unsubtle acronym of GLOVE (Generated Low Output Voltage Emitter) and are manufactured by Compliant Technologies, based in Lexington, Kentucky. The GLOVE, previously used in jails and police departments in the US, looks and functions like a normal pair of patrol gloves until officers press a switch to activate an electrical mode.

Classified as a CD3 (conductive distraction and de-escalation device), the gloves don’t work like stun guns or Tasers, which shoot out probes or use high voltages to cause neuromuscular incapacitation that overrides the central nervous system. Instead, the GLOVE works using neuro-peripheral interference, essentially shocking the skin to bombard the sensory nerves and temporarily short-circuit the brain’s focus.

Compliant Technologies did not respond to WIRED’s request for comment.

When activated via a one-second press on a wrist switch, conductive pads embedded on the palm and undersides of the fingers of the gloves can transmit high-frequency electrical pulses directly to the wearer’s point of contact on a subject. A rechargeable battery and microprocessor module tucked inside a zippered pouch on the cuff powers the entire system.

The electrical pulse overstimulates nerve endings in the subject’s peripheral nervous system, resulting in pain that inhibits coordinated muscle movement, but without causing systemic muscle lockup. The pain induced is localized to just the area touched by the glove’s conductive pads. The manufacturer claims the gloves result in a subject’s compliance in less than three seconds and leave no burns, marks, or scars.

Because the current flows only between the contact pads on the device and across the subject’s skin, the officer wearing the GLOVE, as well as any assisting officers holding the subject, should not feel electrical shocks. The gloves, capable of operating between 14 and 122 degrees Fahrenheit, come in three versions: Gen 3, 4, and 5. Maximum voltage is capped at 380 V for all, and the minimum voltage for Gen 3 and 4 is 210 V, which rises to 324 V for Gen 5.

Maximum stimulation time on Gen 3 and 4 models is a sobering two hours, though that drops to 90 minutes for Gen 5. The battery is apparently good for 2,000 charge/discharge cycles, and it can recharge to full in two hours. Four LED lights on the gloves display the current charge level, each noting 25 percent battery increments.

Gen 4 and 5 of GLOVE both include mechanisms that record “activation events,” tracking the date, hour, minute, and skin-contact time for audit trails and after-action reporting. Gen 4 stores this data on a removable microSD card, while Gen 5 does away with this in favor of built-in internal memory, accessed via a USB-C cable.

There’s a test function on the gloves, too, where operators can verify they are working by tapping an active pad against their own forearm to feel a minor electrical pulse. According to the manuals, GLOVE users must be certified and then recertified every other year to continue using the device.

GLOVE, which can only be sold to verified professional entities such as the military, correctional facilities, and law enforcement agencies—has some clear usage recommendations in Compliant Technologies’ instruction manuals:

• No more than two GLOVE units (or, in other words, one pair) should be used simultaneously on a single subject.
• Continuous shocks should not exceed 15 seconds.
• Operators must target arms and legs, while strictly avoiding the head, face, throat, chest, and groin.
• The gloves should not be used on the elderly, young children, pregnant individuals, or people with severe disabilities.
• The technology must not be used to counter verbal defiance, as punishment, or for torture.