Technology

• Google’s AI Overview is now powered by Gemini 2.0
• Google’s new experimental AI Mode offers reasoning and follow-up responses in search
• Unlike traditional search, AI Mode actively synthesizes information from multiple sources

Google is doubling down on bring AI to Search with an expansion to its AI Overviews and new experimental AI Mode. In Google's ideal world, you would basically never leave the search engine and it would handle a lot of that pesky thinking on your behalf. On the other hand, there would be fewer open tabs giving you anxiety.

Overviewing AI

AI Overviews have already become a familiar sight for many users, appearing at the top of search results with neatly summarized answers to big questions. Now, with Gemini 2.0 behind the wheel, Google claims these overviews will be faster, more detailed, and capable of handling trickier queries in areas like coding, advanced math, and multimodal reasoning. In other words, if you were relying on Reddit and Stack Overflow to explain why your Python script keeps throwing an error, Google wants you to ask it instead.

Google claims there are a lot of fans who want to see AI even more embedded in their online searches. That's what led to the creation of AI Mode. Currently in an experimental mode in Labs. AI Mode aims to bring better reasoning and more immediate analysis to your online time. AI Mode won’t just spit out a quick answer to your query. It will actively break down complex topics, compare multiple options, and pull from online sources to provide more nuanced responses. AI Mode should help prevent you from going from a simple search for a new toothpaste to spending 30 minutes on Wikipedia learning about the history of medieval dentistry.

Let’s say you’re trying to figure out the best way to track your eating habits. A regular Google search might give you a mix of ads, product reviews, and a few scientific studies buried somewhere on page two. AI Mode, on the other hand, can analyze the pros and cons of different apps, smart wearables, and AI-enhanced scales in a single response, summarizing the key features and limitations; even pulling in fresh user reviews. If you need more details, you can ask a follow-up question like, “How does eating late at night impact heart rate?” and get an instant, well-researched answer without having to piece it together from five different websites.

AI internet

This shift isn’t happening in a vacuum. AI-powered search has been heating up, with OpenAI’s ChatGPT Search and Perplexity AI both offering their own takes on AI-assisted browsing. ChatGPT Search integrates OpenAI’s conversational abilities with web access, but it can still struggle with sourcing real-time information. Perplexity AI, meanwhile, prides itself on transparency, always citing sources alongside its answers. Google, being Google, wants to combine deep AI reasoning with its long history of crawling and ranking the web.

Of course, any time AI takes a bigger role in search, there are questions. Will AI Mode and AI Overviews make it harder for small content creators to get traffic if Google keeps more users within its own ecosystem? What biases will even more complex synthesized answers introduce? Google insists that human oversight and ranking systems are still in play, but the company also acknowledges that AI is a work in progress. There’s a fine line between an AI-powered search assistant and an all-knowing gatekeeper that decides what’s “true” before you even click a link.

Beyond the potential pitfalls, AI Mode represents a shift in online research from passive to dynamic. It moves the burden of crafting the right query, clicking through results, and assembling information from different sources from you to the AI. Faster answers, smarter suggestions, and less time wasted on irrelevant links is a tempting proposition. But how good can it actually be compared to you putting little thought into it? AI Mode may as well be called Lazy Mode. Will you still bother clicking through to the long-form articles, deep-dive blog posts, or thoughtful articles on tech news websites for context and nuance beyond an AI summary at this level?

You might also like

• Yes, Google's new Gemini 2.0 Flash is much better than the old 1.5 model – here's what I found
• Google is bragging about Gemini 2.0 Flash, but how does it compare to ChatGPT for casual AI chatbot users?
• Google Gemini's new model is the brainstorming AI partner you've been looking for

Beneath the ES90's class-bending proportions is a new 800-volt EV architecture, enabling faster charging and improved efficiency.

Full spoilers follow for Invincible season 3 episode 7.

Well, I warned you that Invincible season 3 episode 6's mid-credits scene confirmed that what would follow was an absolute bloodbath.

To be frank, I'm a long-time Invincible fan, so I knew what was coming long before season 3's penultimate episode, titled 'What I Have Done?', which arrived today (March 6). Nevertheless, the show's depiction of The Invincible War, aka the near-world-ending event that's been teased since Invincible season 3 started on February 6, is even more soul-shattering than its comic-book counterpart. And, as we sift through the wreckage of this destruction-laden entry, I suspect you've got questions about what transpired.

Want to know who voices Conquest? Need more information about those enigmatic figures working with Angstrom Levy? Or are you hoping against hope that those two characters didn't actually die? I'll answer those queries and more in this article, so make sure you're caught up as full spoilers immediately follow for episode 7.

How do Rex Splode and Darkwing die in Invincible season 3 episode 7?

Redemption for Darkwing! (Image credit: Prime Video)

There are plenty of heroes who die in this episode. After all, Angstrom Levy unleashes 18 multiversal variants of Mark Grayson/Invincible on the Earth that the titular character – from now on, I'll call him Invincible Prime to spare any confusion – inhabits. They're all as powerful, not to mention more cruel, than Invincible Prime, too, so they quickly lay waste to his home world. Remember, Angstrom promised that, in return for them decimating Invincible Prime's planet and ruining Mark's life, he'd help these variants to conquer other universes, which is why they agree to aid Mark's arch-nemesis.

There are two major character deaths that matter more than most in 'What Have I Done?', though. Darkwing is the first of those, with the Batman-inspired hero sacrificing himself to save his fellow Guardians of the Globe by stranding one of Invincible's variants – and himself – in the Shadow-Verse. This realm is not only devoid of light, but also inhabited by terrifying creatures who prey on anything that enters their domain. It's safe to assume, then, that Darkwing and this version of Invincible are no more.

I'm not crying, you are... *sniff* (Image credit: Prime Video)

Episode 7's most soul-destroying demise, however, is reserved for one of the best Prime Video shows' most developed characters: Rex Splode.

I already knew he wasn't long for this world. Indeed, Invincible season 3 episode 5's sweet scene between Rex and Rae foreshadowed this (nobody in Invincible can have nice things!). At least Rex's sacrifice was a noble one, with the kinetic energy-based superhero using his own skeleton in a Hail Mary move to kill the Invincible variant who attacks the Teen Team's base and almost murders Rudy/Robot, Monster Girl, and Bulletproof.

It's the same way that Rex bows out in the source material, but Amazon's adaptation makes his death even more heart-breaking. He's enjoyed more airtime and character growth in Invincible's TV adaptation than he does in the graphic novels. Rex's character evolution from a wisecracking womanizer to heroic martyr in the hit animated series, then, has been one of its biggest triumphs. It's just incredibly sad that we'll never see him, nor hear Jason Mantzoukas' terrific take on the character, again. *sniff*

Who is Conquest? And who voices him in Invincible season 3?

Wait, is that Jeffrey Dean Morgan's voice!? (Image credit: Prime Video)

I'll answer the latter first as it's the easiest question to, well, answer: it's Jeffrey Dean Morgan.

The Walking Dead alumnus was first linked with a role in Invincible in early 2023. However, ahead of Invincible season 2 part 1's release, Robert Kirkman, co-creator of Invincible's comic book series and its TV namesake, told me that he was "going to disappoint people" by saying Morgan wouldn't voice someone in season 2. It's now clear that he was going to play Conquest in season 3, so I guess Kirkman can say he didn't actually lie to our faces.

Anyway, who is Conquest? In the comics, he's not only a high-ranking Viltrum Empire warrior, but also the second-most powerful of them all. The superpowered humanoid race's autocratic leader Grand Regent Thragg is the only one who's more formidable than Conquest, which means the latter is even scarier than Nolan Grayson/Omni-Man. Oh, and he's been sent to Earth to see if Mark/Invincible Prime is ready to perform his Viltrumite duties and prepare his planet to be taken over by the highly-advanced, dictatorial species.

I won't say much else about Conquest's role in Invincible because, you know, spoilers for future seasons. You'll definitely want to prepare yourself for a cataclysmic and incredibly tense season 3 finale, though!

What happens to the remaining multiversal Invincible variants?

"Angstroooooom!" (Image credit: Prime Video)

After their three-day-long, catastrophic attack on Invincible Prime's homeworld, the remaining eight multiversal variants – the other 10 have been killed by Mark and his fellow heroes – regroup at Mark's home at Angstrom's behest. However, when Angstrom says he wants them to bring Invincible Prime to him, they refuse because that wasn't part of the original deal. Cue the octet turning on Angstrom.

Knowing that he's physically outmatched by one Invincible, let alone eight, Angstrom reacts by activating his plan's fail-safe. Opening a portal behind each variant, he uses their spherical drones to push them through said multiversal doorways and maroons them on the desolate version of planet Earth that Angstrom and Invincible Prime fought on in Invincible season 2 part 2's finale. Hey, he says he was going to betray them anyway, so he might as well do so now...

Invincible season 3 episode 7's other heroes explained: Wolf-Man, Brit, Best Tiger, Tech Jacket, and more

'What Have I Done?' is full of Image Comic superhero cameos, such as Wolf-Man (Image credit: Prime Video)

I'll readily admit that I didn't recognize every hero that Global Defence Agency (GDA) chief Cecil Stedman calls on to defend Earth from the 18-strong group of multiversal Invincibles in Invincible season 3's penultimate episode. There are a few, though, that I recognize from the source material, and I'm sure that my fellow Invincible devotees will be delighted by their inclusion in Prime Video's adaptation.

The first – and perhaps most crowd-pleasing – is Wolf Man, who battles one of the variants in what appears to be New York City and winds up winning the contest. Wolf Man is the star of Invincible spin-off comic series The Astounding Wolf-Man, which is another of Image Comics series, and the eponymous hero's real name is Gary Hampton, who was the wealthy CEO of a record company until he was attacked by a werewolf during a family vacation. Now, he uses his abilities to do good, all the while trying to suppress his new-found animalistic instincts.

Jonathan Banks voices Brit in Invincible's TV adaptation (Image credit: Prime Video)

The next most recognizable face is Brit, who's voiced by one of nine new, big-name cast additions for season 3 in Jonathan Banks, who famously portrayed Mike Ehrmantraut in Breaking Bad and Better Call Saul.

Another Image Comics creation, Brit is a long-time friend of Cecil's who, per the source material, was born sometime before World War One. His abilities include superhuman strength, agility, and reflexes, and decelerated aging. He's also virtually indestructible – indeed, as noted in Invincible's graphic novels and other Image supplementary materials, Brit is immune to all forms of damage and disease. You'd wonder why he's not been used as one of the GDA's first lines of defence before now, then...

Well, this is a deep-cut Image Comics reference! Say hello to Best Tiger, everyone (Image credit: Prime Video)

Three other heroes – two of whom we first saw in Invincible season 2 episode 8 – also appear in 'What Have I Done?'. Kid Thor and Knock-Out are the aforementioned duo, and they're joined by Bolt, another Batman-esque individual who acquired his superhuman abilities after being hit by a lightning bolt, hence his superhero name. The trio appear at the prison where Invincible Prime is fighting Mohawk-vincible, mistake Invincible Prime for one of his multiversal variants, and attack him before they realize the errors of their ways.

But wait, there's more. The biggest deep-cut character in 'What Have I Done?' is arguably Best Tiger. A China-born superhuman, he's essentially Daredevil with guns (he's also not blind – he just covers his eyes with a bandana for the challenge). He possesses superhuman agility, enhanced agility, and is an expert martial artist. He's also an elite marksman so, really, he's a cross between Daredevil and one of his most notorious foes in Bullseye.

Holy crap, it's Tech Jacket! (Image credit: Prime Video)

Last but not least, this episode marks the debut of another key Invincible hero in Tech Jacket. I believe he's the individual who fights one of the Invincibles in space. If I'm right, that means Tech Jacket has actually appeared in Invincible's TV show earlier than he does in the comics.

Anyway, for the uninitiated: Tech Jacket is another Image Comics brand that stars Zack Thompson. One day, the ordinary high-school kid comes across a dying member of the highly-intelligent but physically weak alien race called the Geldarians. To make up for their lack of physicality, each Geldarian is given a Tech Jacket, a technologically-advanced vest that not only turns into a suit of armor when its user is attacked, but also grants the wearer near-invulnerability, super-strength, and the ability to fly and fire lasers.

Knowing its crashed spaceship will kill them both, the dying Geldarian gives its Tech Jacket to Thompson to save his life. Now permanently bound to one of the most powerful devices in the known universe, Thompson becomes a hero in his own right and is soon recruited by The Coalition of Planets to help them in their quest to bring down the Viltrum Empire.

Who are the Technicians in Invincible season 3?

Well, aren't you guys the definition of creepy... (Image credit: Prime Video)

The Technicians are introduced in the first scene of 'What Have I Done?', which takes place in the immediate aftermath of Mark's near-fatal beatdown of Angstrom from season 2 episode 8.

As he does in the comics, Angstrom somehow summons the strength to open one of his portals to the Technicians' laboratory and, after getting their attention, is pulled into their dimension. Despite his absolutely mushed-up face, he tells the Technicians to "fix him", which they duly do.

Alright, but who are they? Again, I'm not going to reveal too much because they continue to play a role in the graphic novels and, by extension, will appear in the show again at some point. All you need to know for now is that the Technicians are cybernetically-enhanced humanoids who hail from another dimension and specialize in the art of body modifications.

After patching up Angstrom, they make a deal with him for reasons I won't ruin here. As Angstrom finds out, it's also best to stay on their good side, too. Indeed, when Angstrom returns after his latest bout with Mark and tells the Technicians to replace the arm he lost – the one that was severed by one of Angstrom's portals when Mark grabbed it as the gateway closed – the trio initially refuse to aid him. He hasn't fulfilled his side of the bargain so, in light of needing their services again, they tell him that he works for them now. I guess Angstrom's one-man crusade against Mark will have to wait!

Invincible's long-running Ka-Hor joke explained

Maybe you'll escape one day, Ka-Hor... (Image credit: Prime Video)

A new season of Invincible wouldn't be complete without a check-in with Ka-Hor, and the R-rated series' latest chapter obliges. Wait, you don't remember who Ka-Hor is? Alright, allow me to explain.

A completely original character for Invincible's animated adaptation, Ka-Hor is a Egyptian mummy who's permanently bound to the tomb he was laid to rest in because of an unknown curse. The only way he can escape is if he binds his soul to a living, male host.

So close, yet so far (Image credit: Prime Video)

Three seasons in and Ka-Hor hasn't been able to leave his crypt. He was almost freed by a present-day worshipper called Zaiem in season 1, but Mark/Invincible accidentally prevented Ka-Hor's escape when he inadvertently caused a sandstorm near Ka-Hor's tomb and trapped Zaiem inside.

In season 2, Ka-Hor's attempts to break out are dealt a blow when the next two individuals to find his tomb are women. Once again, Mark traps them all inside Ka-Hor's burial chamber with another accidental sandstorm.

That brings us onto season 3. During a battle between Immortal and one of the multiversal Invincibles, Ka-Hor's tomb is dealt significant damage when one of Mark's variants crashes into it. Sensing an opportunity to escape, Ka-Hor tries to bind to said Invincible.

Unfortunately for him, he's not strong enough to overpower his potential new host. To make things worse, Immortal and his adversary take their fight elsewhere soon after, which humorously leaves Ka-Hor without another host body. Oh, and the two women he encountered in season 2 finally escape his crypt, too. At least somebody got what they wanted.

You might also like

• Invincible season 3 is another soaring entry of the popular Prime Video series that packs a real punch in more ways than one
• Invincible season 3 changes Oliver Grayson's shocking coming-of-age moment from the comics for the better – and for the worse
• Find out everything we know so far about The Boys season 5

Commentary: QR codes printed directly onto the machine make it easy for someone like me to upgrade my components with confidence.

• UK CMA was worried about Microsoft’s involvement in OpenAI
• “OpenAI does not qualify for investigation,” says Authority
• Critics worried about government pressure on CMA

The UK’s Competition and Markets Authority (CMA) has ended its investigation into Microsoft’s partnership with OpenAI and its multibillion-dollar investment into the company behind ChatGPT.

Despite Microsoft investing heavily in the AI startup and having exclusive use of some of its AI products, the CMA has now concluded the partnership has not been problematic.

The watchdog also considered how the companies’ relationship changed during the short period that CEO Sam Altman was fired, then rehired.

UK CMA deems Microsoft-OpenAI partnership healthy

The CMA said yesterday: “The CMA has decided that Microsoft’s partnership with OpenAI does not qualify for investigation under the merger provisions of the Enterprise Act 2002.”

Expanding on its conclusion in a phase one summary, the CMA noted that “no relevant merger situation ha[d] been created,” therefore it did “not ha[ve] to conclude on whether the other criteria for establishing a relevant merger situation [were] met.”

The Authority initiated the probe back in December 2023 because of Microsoft’s involvement in bringing Sam Altman back to OpenAI, but despite Microsoft’s “high level of material influence” over OpenAI, it doesn’t have full control over the company.

The official decision came just one day after the merger inquiry was launched, and critics have argued that political changes and government pressure on regulators to focus on economic growth may have guided the CMA’s conclusion.

Foxglove co-executive director Rosa Curling told the BBC: “This is a bad sign that Big Tech has successfully convinced the prime minister to defang our competition regulator and let Big Tech gobble up the current generation of cutting-edge tech – just like they did the last one.”

Microsoft welcomed the outcome: “Our OpenAI partnership and its continued evolution promote competition, innovation, and responsible AI development, and we welcome the CMA's conclusion, after careful and prudent consideration of the commercial realities, to close its investigation.”

You might also like

• We’ve listed the best AI tools and best AI writers
• Microsoft’s exclusive cloud deal with OpenAI is coming to an end
• Access the most powerful chips with the best cloud computing services

The Digital Operational Resilience Act (DORA) came into effect on January 17, 2025. Financial services institutions (FSIs) across the EU must now fully comply with its stringent cybersecurity and operational resilience requirements. But achieving compliance is not just about meeting regulatory expectations. DORA represents a fundamental shift in how financial institutions approach digital security, ensuring they can withstand cyber threats, operational disruptions, and third-party vulnerabilities.

For firms that have already established a compliance framework, the focus now moves to long-term resilience and continuous improvement. For those still catching up, the urgency to close security gaps has never been greater. Failing to meet DORA’s requirements carries not only financial penalties but also the risk of operational restrictions and reputational damage. In this new era of cybersecurity regulation, FSIs must go beyond basic compliance measures and embed resilience into their core strategies.

A shift in cyber resilience thinking

For years, financial institutions have relied on traditional cybersecurity approaches, primarily focused on perimeter security to keep external attackers at bay. However, recent cyber incidents have made it clear that threats do not always come from outside an organization. Many damaging breaches have originated from within digital supply chains, through third-party vulnerabilities, or from internal weaknesses. In 2023, third-party attacks led to 29% of breaches with 75% of third-party breaches targeting the software and technology supply chain. This evolving threat landscape has forced financial institutions to rethink their approach. The future of cyber resilience isn’t about building higher walls - it’s about securing every layer, inside and out.

DORA mandates a resilience-first mindset, shifting the focus from prevention alone to a more comprehensive strategy that includes rapid response and recovery. It is no longer enough to defend against cyber threats; organizations must assume that breaches and disruptions will happen and ensure they can respond swiftly and effectively. This change means cybersecurity is no longer just the responsibility of IT management. It is now a board-level priority, requiring CFOs, CIOs, and risk officers to play a direct role in overseeing governance structures, risk assessments, incident response planning, and ongoing security monitoring.

The growing role of automation in compliance

With DORA now in full effect, financial institutions are also navigating additional regulatory frameworks such as the NIS2 Directive and the Cyber Resilience Act (CRA), both of which introduce further security and operational resilience requirements. The increasing complexity of compliance is prompting many organizations to turn to automation to streamline regulatory processes.

Okta’s 2024 Businesses at Work report found that data compliance tools were the fastest growing applications with 120% year-on-year growth. As firms seek to reduce the burden on their security teams while ensuring continuous adherence to evolving regulations, the rising popularity of these tools is unsurprising.

Automating security audits, compliance validation, and real-time threat detection allows financial institutions to maintain compliance efficiently while also enhancing their ability to identify and mitigate risks before they escalate into major incidents. In a landscape where regulatory expectations will only become stricter, automation is important for maintaining both security and operational efficiency.

Addressing digital supply chain risks

One of the most pressing concerns for financial institutions under DORA is the security of their digital supply chains. High-profile cyberattacks in recent years have demonstrated that vulnerabilities often originate not from within an organization's own IT infrastructure, but through weaknesses in third-party service providers, cloud platforms, and outsourced IT partners. DORA places a strong emphasis on third-party risk management, making it clear that security responsibility extends beyond a firm’s immediate network.

Ensuring supply chain resilience requires a proactive and continuous approach. FSIs must conduct regular security assessments of all external vendors, ensuring that partners adhere to the same high standards of cybersecurity and risk management. It is no longer sufficient to perform security checks only at the beginning of a partnership; ongoing monitoring and real-world scenario testing are essential to ensure that contingency plans hold up under real conditions. The ability to anticipate and respond to emerging threats within the supply chain is critical to maintaining operational stability and regulatory compliance.

Navigating post-implementation compliance challenges

While many FSIs had operational resilience frameworks in place before DORA’s enforcement date, aligning these existing efforts with the regulation’s EU-wide supervisory structure presents new challenges. Firms that have not been closely following the consultation process may struggle to adapt to these additional requirements.

At this stage, financial institutions must prioritize regular compliance evaluations to ensure that their security frameworks remain aligned with DORA’s evolving mandates. Conducting a gap analysis is critical to identifying areas where improvements are needed. Engaging with regulators, industry bodies, and technology partners can provide valuable insights into best practices and common pitfalls. Additionally, collaboration within the financial sector will be essential, as firms can learn from each other’s experiences and share strategies for maintaining long-term compliance.

The cost of non-compliance

The consequences of failing to comply with DORA are severe. Regulators now have the authority to suspend business operations, issue cease-and-desist orders, and demand access to sensitive data for compliance reviews. For critical third-party service providers, non-compliance could result in financial penalties of up to 1% of their global daily turnover for up to six months - a staggering cost that could significantly impact business operations.

Beyond regulatory penalties, the reputational damage of non-compliance may be even more devastating. The financial sector operates on trust, and any failure to meet cybersecurity standards can lead to a rapid loss of confidence from both consumers and investors. A single security lapse or compliance failure can undermine an institution’s credibility, and once trust is lost, rebuilding it can take years. FSIs must recognize that compliance is not just about avoiding fines - it is about preserving their reputation and long-term viability in an increasingly digital financial ecosystem.

The role of identity security

One of the most effective ways to strengthen cybersecurity resilience under DORA is through identity management (IAM). Research indicates that 80% of cyberattacks originate from compromised credentials, making authentication and access control a top priority for financial institutions.

A robust IAM strategy involves implementing multi-factor authentication (MFA), enforcing least-privilege access policies, and continuous monitoring for credential-based threats. The adoption of a zero-trust security model, where no user or system is automatically trusted, further enhances security by ensuring that every access request is verified before granting permissions. As cybercriminals continue to develop more sophisticated attack methods, securing user identities will remain a cornerstone of both regulatory compliance and overall cyber resilience.

An opportunity for long-term resilience

DORA has transformed the cybersecurity landscape for financial services firms. Compliance is no longer a one-time activity - it is an ongoing effort that requires constant adaptation to emerging threats and regulatory updates. Organizations that approach DORA as an opportunity to strengthen their overall cybersecurity posture will be best positioned for success.

FSIs that invest in proactive security strategies today will not only protect themselves from regulatory penalties but will also build stronger, more resilient digital ecosystems. Cyber resilience is now a business imperative, and those that take it seriously will emerge as leaders in the evolving financial landscape. Compliance in itself should not be the security strategy of any organization, but it is a rising tide that raises all ships to a better security foundation to the benefit of all.

We've set up a list of the best network monitoring tools.

This article was produced as part of TechRadarPro's Expert Insights channel where we feature the best and brightest minds in the technology industry today. The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/news/submit-your-story-to-techradar-pro

The four non-3D Zen 5-based AMD Ryzen processors top our leaderboard when it comes to price per core.

Data collated at the beginning of March 2025 shows that the 9900X, the 9950X, the 9700X and the 9600X are the most competitive price wise.

The Ryzen 9900X is by far, the most balanced offer of the quartet, with a low TDP per core (just 10W), a high base speed (4.4GHz) and a very reasonable price at $387.75 (or $31.56/core) at the time of writing, almost a quarter cheaper than its suggested retail price.

This is the second of several articles based on data I’ve compiled on 41 AMD Zen 4 and Zen 5 CPUs (socketed, OEM). In the rest of the series, I will be looking at the cost per core, performance per core, AMD CPUs that are getting more expensive, all this with the new Ryzen 9 9900/9950 X3D CPUs in the backdrop.

Not bad for a near-flagship CPU launched less than one year ago. The 9950X has a cost per core slightly higher, at $34.05, but is the fastest consumer CPU that AMD has to offer (until the launch of the 9950X3D).

The table of all the CPUs I have analyzed can be found at the end of this article. They have been sorted by cost per core. Some CPUs are not yet on sale at the time of writing.

A ‘value’ 192-core CPU?

At just over $10,000 from a reputable retailer (Wiredzone), the EYPC 9965 is AMD’s most expensive CPU ever launched and one that I covered extensively in a recent article.

It has 192 cores, which translates into a per-core cost of $52.26; far more than any consumer Ryzen CPUs but still a third of the cost of the most expensive AMD CPU (per core).

It delivers one of the lowest TDP per core (at just 2.6W) and the lowest TDP per GHz* (1.16W), thanks to its Zen 5c architecture, a more compact (but compatible) version of the Zen 5.

Its smaller sibling, the 96-core AMD EPYC 9655, has the largest discount I’ve seen across the 41 CPUs I’ve tracked, with a staggering 56.8% reduction from the sticker price.

It is a full Zen 5 part and as such gets a much higher TDP per core, twice the amount of cache and a faster base speed.

* Lowest TDP per GHz is calculated by taking the CPU TDP and dividng it by the number of cores x the base speed in GHz. It delivers a very rough composite efficiency metric.

The mystery of the ThreadRipper Pro 7945WX

At the other end of the spectrum, the EPYC 9175F is the most expensive AMD CPU per core costing of just under $160, that’s almost 5x that of the 9950X, which shares the same number of cores (16).

The reason why it is so expensive is that it has 32x more cache per core than an average consumer CPU (512MB) and cache is a very, very expensive commodity.

Other F-labelled EPYC CPUs trawl the bottom of my cost per core leaderboard; F stands for Fast and these CPUs are high frequency optimized parts with big cache memory.

One more thing. I’d like to draw your attention to the existence of the Ryzen Threadripper PRO 7945WX.

It is the only AMD CPU from this list that you cannot buy as it is available exclusively in workstations from Lenovo, HP and Dell.

What makes it so special for me is its high TDP per core, the highest of all the CPUs I’ve analysed.

At 6.21W, this Zen 4 part is 5.5X more power hungry than the EPYC 9965 (or 3x that of the 9900X, a similar 12-core CPU).

Maybe that’s because it has the highest base speed of any CPU in the list (jointly with the 9800X3D) and is built on an older manufacturing technology.

You might also like

• Future exascale supercomputers will be built using these HPE Lego-like cabinets able to hold nearly 100,000 AMD EPYC Cores
• First AMD EPYC 9965 benchmarks are in; Turin server CPU sets flurry of new world records across wide range of software, but will that be enough to stop Xeon and Arm's forthcoming onslaught?
• Want to put 24TB of RAM in your PC? This motherboard can, and will even accommodate 384 AMD EPYC Turin cores and almost 1 petabyte storage

In a world where Generative AI (GenAI) is reshaping the global business landscape, mastering faster decision-making at the right time at the right pace has become a crucial competency for leaders seeking to maintain a competitive edge. GenAI is rapidly becoming an integral part in business operations, driven by its ability to streamline processes, enhance effectiveness, improve competitiveness, and lead to significant cost reductions and revenue enhancements.

According to a recent IDC report, enterprise spending on GenAI globally will grow by 30% in 2024 – $40 billion from an estimated US$16 billion in 2023. Spending is also expected to swell to more than $150 billion by 2027, with the banking, retail, and professional services industries being the top spenders. This spending is bound to increase significantly beyond the estimates in coming years.

The latest developments in the field include the broad adoption from open public models to private models to pre-trained open source AI models to untrained open source and custom models. The significant investments are in both infrastructure and AI-enhanced products and services.

Enterprises are moving beyond initial experiments with GenAI and towards aggressive infrastructure and trained data model building, aiming for a transformation that integrates GenAI at the core of digital business activities. This strategic integration is expected to provide a competitive edge and catalyze a shift towards more dynamic, efficient, and innovative business environments.

For business leaders, the challenge now lies not just in adopting these technologies but in integrating GenAI with human intuition to optimize faster and right decision-making processes. This nuanced approach ensures that businesses keep pace with technological changes and stay ahead of their competitors. Understanding and harnessing the power of GenAI is essential for any organization aiming to thrive in this transformative era.

Strategic Implications of GenAI

GenAI stands distinct with its ability to create new content, ranging from text to images, from existing data sets, a stark contrast to other AI technologies that primarily analyze or interpret data. This capability positions GenAI as a revolutionary technology in enhancing business strategy, across global tech giants. 

In marketing, for instance, GenAI enables the creation of highly personalized content that resonates with diverse customer segments, dramatically improving engagement rates. Product development also benefits from GenAI as it can suggest innovative product features or designs by analysing current market trends and consumer feedback. Moreover, in customer service, it enhances responsiveness and personalization, as seen in AI chatbots that provide real-time, context-aware solutions to customer queries. These examples illustrate significant efficiency gains and competitive advantages, marking GenAI as a transformative force across business functions. 

One significant example of strategic implication of GenAI is in the telecommunications industry where it helps to optimize network performance and management. GenAI models analyze vast amounts of data from network operations, including traffic patterns, equipment health, and historical performance metrics. These models can simulate various scenarios to predict potential network failures or degradations before they occur.

GenAI plays a crucial role by creating synthetic datasets and simulations that mirror real-world network conditions. This allows telecom operators to test and validate maintenance strategies, capacity planning, and network upgrades without disrupting actual service. By simulating different traffic loads and failure conditions, AI can recommend optimal configurations and preemptive actions, leading to reduced downtime, improved service quality, and cost savings on emergency repairs. The predictive insights generated by AI ensure that the network remains resilient and capable of handling increasing data demands.

Challenges in AI-driven Decision-Making

Integrating GenAI into business decision-making processes presents nuanced challenges, necessitating a balanced approach to utilizing AI outputs. In complex scenarios, the efficacy of GenAI hinges on the model’s training adequacy. If a model is not sufficiently trained for a specific task, human intervention becomes crucial, as human expertise can surpass undertrained AI models in navigating intricate decisions. Conversely, when models are well-trained, they can outperform humans by delivering consistent and data-driven insights at scale.

Therefore, the integration of GenAI requires astute judgment to discern when to rely on AI and when to defer to human judgment. This balance ensures that decision-making processes harness the strengths of both AI and human intelligence, leveraging AI for efficiency and precision, while capitalizing on human intuition and experience in areas where AI’s training may fall short. This nuanced approach is essential for maximizing the potential of GenAI in business contexts.

Another notable challenge associated with integrating GenAI into business decision-making processes is data privacy. GenAI systems require vast amounts of data to train and operate effectively. This reliance on large data sets raises concerns about compliance with global data protection regulations such as GDPR in Europe or CCPA in California, which mandate strict guidelines on data usage, storage, and privacy. In the telecommunications industry, GenAI could be used to analyze customer call data to improve service offerings or personalize marketing strategies.

However, this data often contains sensitive personal information. Ensuring that GenAI applications comply with data protection laws requires robust anonymization techniques and secure data handling practices. Failing to adhere to these regulations can result in substantial fines and damage to the organization's reputation, illustrating the complexity and risk associated with deploying GenAI in sectors with stringent privacy requirements. There are multiple data masking and data anonymization solutions available in the market which needs to be applied if there are PII in the data before using them for training the models.

To navigate these complexities, leaders must stay abreast of the latest developments in GenAI by engaging with ongoing education, participating in industry forums, and fostering partnerships with AI ethics boards. By doing so, they can implement GenAI's capabilities responsibly and effectively, ensuring that their strategic decisions are both innovative and ethically sound. 

Embracing the Epoch of GenAI: A Strategic Imperative 

In the vanguard of technological evolution, the strategic integration of GenAI stands as a linchpin for redefining decision-making and operational efficiency within forward-looking businesses. This leap towards GenAI adoption is not merely an enhancement but a transformative shift that sets enterprises apart in today's competitive landscape. 

To harness GenAI’s full capabilities, establishing a solid business case is crucial before implementation. This involves identifying key objectives and anticipated benefits aligned with business goals. Conducting a Proof of Concept (PoC) or pilot project helps validate GenAI’s potential, demonstrating tangible results and addressing any challenges. By doing so, businesses can ensure a strategic, well-informed adoption of GenAI, optimizing its impact and value.

In a nutshell, diving into GenAI isn't just about keeping up with the latest tech trends. It's about seizing an opportunity to redefine how your business operates, making decisions smarter and faster than ever before. The future is about those who adapt, and with GenAI, that future is bright.

We've compiled an extensive list of the best AI tools.

This article was produced as part of TechRadarPro's Expert Insights channel where we feature the best and brightest minds in the technology industry today. The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/news/submit-your-story-to-techradar-pro

We’ve reached a point of obfuscated redundancy in the auto space. Many manufacturers pushing out new models incorporate aspects calling on Tesla’s minimalism – but that means the axing of many important things, like the instrument cluster, tactile buttons and dials, and a focus on customization in the infotainment system, with this trend more noticeable in the EV space. Among the axing of supposed non-essentials is the humble full-size spare tyre – which is why it’s cause for celebration whenever it returns.

The Chery Omoda E5, sold in the UK as simply the Omoda E5 and originating from China, is the car we’re celebrating today. I was diplomatic in my initial impressions article on the E5, noting that it’s likely a better fit for value-savvy Aussies than Britons, who have a greater variety at the car’s price point. It lacks a competitive angle on driving comfort, boot capacity, and DC recharge time, though its price to range ratio is considerable and the features offered in the slightly more expensive trim are notable.

Across both trims in Australia and the UK, there is one standout feature – the full-sized spare tyre. There’s also the expected internal space for it to be stored under the boot mat in the back.

A spare tyre is a rarity among new cars, especially for EVs. It’s a feature only typically found in cars built for off-roading lifestyles, such as heavy-duty SUVs and utes, but you’ll find exceptions here and there, like the 2022 Subaru BRZ.

But it should be a greater consideration for drivers and manufacturers alike. Roadside assistance, though often reliable, especially for drivers who may not be comfortable repairing a spare on the roadside, can often be time consuming when your car is capable of carrying a spare. So let’s talk about why it’s great that the Omoda E5 has included it.

Spare a spare?

(Image credit: Behold! The spare tyre found in the Chery Omoda E5. Zachariah Kelly / TechRadar)

The spare tyre is an inclusion that has largely been lost over time, with the argument often going that the space is better used to improve volume capacity, that it adds room for a larger battery or fuel tank, or that it adds unnecessary weight. Of the more than 30 EVs I’ve reviewed, only two have featured a full-sized spare tyre – the Omoda E5 and the 2024 Hyundai Kona EV (though there are a handful of other EVs offering spares).

I’m not going to fly in the face of removing redundancy, but a spare tyre is the furthest thing from it. This is the kind of thing where the term ‘have it and not need it, rather than need it and not need it’ applies.

The most common argument against the provisioned spare tyre is that it’s unnecessary. In many cases, a space–saver tyre is enough to get the job done, or a puncture repair kit would suffice.

A puncture repair kit will only work for holes smaller than 3mm (per Drive, who interviewed a former roadside assistance mechanic on the topic), and even then, the tyre may have been further damaged by driving at low pressure. Meanwhile, a space saver limits your speed to 80km/h (49mph) and can only be used for a short distance. These are solutions that could work in the city, but are unideal if you live far out from town.

Within reason, tyres can be repaired for general road use after a tread puncture, but a puncture to the tyre wall is often unlikely to be repairable. A full-size spare tyre gives the owner greater agency over their car. As it’s specced in-line with the rest of the car’s tyres, it can be fitted and used ongoing as if things were normal – though it would be best to replace (or repair the spare) at your earliest convenience.

But obviously roadside assistance can be preferred. People who are less nimble might not want to bend their back to get the wheel out of the back of the car, or might not want to kneel down to replace the tyre if it’s too heavy.

Getting by sparingly

(Image credit: Zachariah Kelly / TechRadar)

It’s hard to think of a spare tyre as anything less than necessary in rural and regional areas. You can get by in cities where a puncture repair kit or space-saver could be relied on for a short distance, but places like regional New South Wales in Australia may not have a tyre shop or mechanic for a great distance.

And to give yourself peace of mind on the road, especially if you live far out of town or are roadtripping, it’s probably not a bad idea to pick up a compatible full-size spare tyre, to save yourself from calling up roadside assistance.

You might also like…

• Chery's Omoda E5 is a brilliant EV in Australia, but a poor purchase in the UK
• I've driven Hyundai's enormous Ioniq 9 – and it's the most luxurious EV the brand has ever made
• Road trips ahoy! Volvo says its upcoming ES90 will offer up to 435 miles of range

Here are the answers for The New York Times Mini Crossword for March 6.

Here are hints and answers for the NYT Strands puzzle No. 368 for March 6.

New tech at Mobile World Congress could let you charge your phone when it's out in the sun or even under room lights.

Hints and answers for the NYT Connections: Sports Edition puzzle, No. 164, for Thursday, March 6.

Hints and answers for Connections for Thursday, March 6.

The low-cost mission may be in danger, and it's not NASA's first such failure.

Landing on the moon is hard, but Intuitive Machines will try for a gentle touchdown of its IM-2 Athena lander packed with rovers, a drone and a NASA moon drill.

I love diving into learning about new things and falling down research rabbit holes, but sometimes I just need a quick, efficient answer to a question or a concise guide to a task. If I’m trying to figure out how long to roast chicken or whether Pluto has been reinstated as a planet, I want a short list of bullet points and a simple yes or no.

So, while ChatGPT's Deep Research feature has proven to be an amazing researcher that is great when I want to immerse myself in a topic, I haven't made it my default tool with the AI chatbot. The AI model's database, as well as its search tool, resolve pretty much any day-to-day question or issue I might ask it. I don't need a formal report on how to make a meal that takes 10 minutes to compile. But, I do find the comprehensive answers from Deep Research viscerally appealing, so I decided it was worth comparing it to the standard (GPT-4o) ChatGPT model and giving it a few prompts that I could imagine submitting on a whim or with little long-term need.

Beef Wellington

(Image credit: ChatGPT Screenshots)

For the first test, I wanted to see how both models would handle a classic, somewhat intimidating recipe: Beef Wellington. This isn’t the kind of dish you can just throw together on a weeknight. It’s a time-consuming, multi-step process that requires patience and precision. If there was ever a meal where Deep Research might prove useful, this was it. I asked both models: “Can you give me a simple recipe for kosher Beef Wellington?”

Regular ChatGPT responded almost instantly with a straightforward, well-structured recipe. It listed ingredients in clear measurements, broke the process down into manageable steps, and offered a few helpful tips to avoid common pitfalls. It was exactly what I needed in a recipe. Deep Research took a full ten minutes and had a very long, complex mini-cookbook centered on the dish. I had multiple versions of Beef Wellington, which did all adhere to my specific requests, but ranged from a Jamie Geller-inspired method to a 19th-century traditional preparation with some substitutions. That's not counting the extra suggestions about decorations and an analysis of various types of puff pastry and their butter-to-flour ratios. If I'm honest, I loved it as a piece of trivia obsession. But, if I wanted to actually just make the dish, it was a bit too much like those recipe blogs where you have to scroll past someone's entire life story just to get to the ingredients list.

TV time

(Image credit: ChatGPT Screenshots)

For the second test, I wanted to see if Deep Research could help me buy a TV so I kept it simple with: “What should I consider when buying a new TV?”

Regular ChatGPT gave me a quick and clear answer. It broke things down into screen size, resolution, display type, smart features, and ports. It told me that 4K is standard, 8K is overkill, OLED has better contrast, HDMI 2.1 is great for gaming, and budget matters. I felt like I had a decent grasp of what to look for, and I could have easily walked into a store with that information.

Deep Research had its usual extra questions about what's important to me, but it was faster this time, only six minutes before delivering a full report on several TVs. Except rather than a simple pros and cons list, I got a lot of unnecessary detail on things like OLED vs. QLED panels, the reason TV refresh rates affect video games, and the impact of compression algorithms on streaming quality. Again, this was all incredibly informative, but entirely unnecessary for my purposes. And unlike Beef Wellington, I'm not going to keep coming back to the TV buying guide on a semi-regular basis.

Telescope look

(Image credit: ChatGPT Screenshots)

For the final test, I decided to get a little more academic in light of my recent decision to pursue astronomy more seriously as a hobby. Still keeping it brief, I asked, “How does a telescope work?”

Regular ChatGPT responded instantly with a simple, digestible answer. Telescopes gather and magnify light using either lenses (refracting telescopes) or mirrors (reflecting telescopes). It briefly touched on magnification, resolution, and light-gathering power, making it easy to understand without getting too technical.

Deep Research gave me a report of a kind I might have written in high school. After asking how technical I wanted my answer, and me responding that I didn't want it to be technical, I waited about eight minutes for a lengthy discussion of optics, the development of different kinds of telescopes, including radio telescopes, and the mechanisms behind how they all work. The report even included a guide on buying your first telescope and a discussion on atmospheric distortion in ground-based observations. It was answering questions I hadn't asked. Admittedly, I might do so at some point so the anticipation of follow-up queries wasn't a huge negative in this instance. Still, a couple of sentences about mirrors would have been just fine in the moment.

Deep thoughts

After running these tests, my opinion of Deep Research as a powerful AI tool with impressive results remains, but I feel much more aware of its excesses in the context of regular ChatGPT use. The reports it generates are detailed, well-organized, and surprisingly well-written. For a random tour of interesting information, it's pretty great, but I much more often just need an answer, not a thesis. Sometimes a shallow dip is preferable to a deep dive.

If the regular ChatGPT approach is accurate and does in seconds what takes Deep Research several minutes and a lot of unnecessary context to provide, that's going to be my preference 99 times out of a hundred. Sometimes, less is more. That being said, Deep Research's shopping advice would be great for a much bigger purchase than a TV, like a car, or even when looking for a house. But for everyday things, Deep Research is just doing too much. I don't need a jet engine for an electric scooter, but, for a transcontinental flight, that jet engine is good to have on-hand.

You might also like

• I tried Deep Research on ChatGPT, and it’s like a super smart but slightly absent-minded librarian from a children’s book
• I tried Perplexity’s Deep Research and it doesn't quite live up to ChatGPT's research potential
• I replaced my to-do list with ChatGPT’s Tasks feature and it completely changed the way I plan my life

• The latest audacious display launch from Thanko certainly pushes boundaries
• Perfect for working on the move - if you have the space
• New double 24-inch monitor can be extended up to 270 degrees

Doubling up on monitors is a surefire way to help drive productivity, but Japanese brand Thanko now seems to have taken things to the next level.

The firm has released a new double 24-inch monitor, and it’s quite the site to behold, as users can expand the monitor into two screens, connected in one single unit.

For those that are short on desk space, it’s a very handy piece of equipment and can be extended up to 270 degrees. The monitor’s measurements come in at 542 x 17 x 650mm (when unfolded) and 542 x 25 x 323mm (when folded).

Getting flexible

Admittedly, it doesn’t quite meet the mark with port options, featuring just a single HDMI slot and two USB-C ports - and one of these is for power supply.

It also boasts an array of ports and features, including a single HDMI port and two USB-C ports, although admittedly one of these is for power supply. These are complemented by a 3.5mm headphone jack as well as two 2W speakers.

From a performance perspective, it also gives users a maximum refresh rate of 100Hz alongside a response speed of 14ms.

The monitor has been touted as a ‘portable’ monitor. Given it weighs some 5kg, or roughly 11lbs, it could make for a great piece of equipment if you're on the move.

You can get your hands on the dual monitor for around ¥62,800 ($420).

Pushing boundaries

This isn’t the first Thanko product to push boundaries, as in February 2022, the gadget maker unveiled an audacious vertical display which allowed users to keep tabs on social media feeds.

The Thanko TL Portrait Display was designed to complement a laptop or desktop display - boasting a display size of 7.9 inches, the compact monitor fitted neatly alongside a laptop, according to reports at the time from Tom’s Hardware.

You might also like

• Looking for a gaming monitor? Check out our favorites
• All the best monitors for remote working right now
• We've also rounded up the best monitors for a dual screen setup

Google's also working on a new AI Mode in Search.

The General Services Administration plans to sell hundreds of government buildings, including FBI headquarters. Use our interactive tools to see where they are—and the congressional districts they're in.