News

Artificial Intelligence (AI) is rapidly evolving. Across industries, many organizations are increasingly deploying AI into systems that must run continuously, securely, and at scale.

As AI adoption accelerates, one thing is becoming clear: infrastructure planning cannot wait.

AI workloads are becoming more interconnected, distributed, and operationally integrated across cloud, data center, and edge environments. Infrastructure planning now requires organizations to align compute, networking, software, memory, and operational requirements across increasingly complex environments.

As a result, many enterprises are beginning infrastructure planning sooner rather than later.

As AI becomes more integrated into everyday business operations through continuous inference and agentic AI systems, infrastructure demands are evolving significantly.

Modern AI deployments increasingly require:

• Continuous inference running around the clock
• Multi-agent systems coordinating across applications and databases
• Real-time orchestration across cloud, data center, and edge environments
• Strong governance, security, and operational efficiency

These workloads require more than raw compute performance. They require balanced infrastructure where compute, networking, software, memory, and operational workflows work cohesively at scale.

Because of this, enterprises are beginning AI infrastructure planning earlier, recognizing that planning, testing, and Proof of Concepts (PoCs) for complex systems like this take time.

At the same time, the cost of delaying AI infrastructure planning is becoming more apparent. Delays can slow deployment readiness and postpone AI-driven benefits such as productivity gains and operational automation. As AI demand continues to rise, organizations are prioritizing earlier planning to secure the compute capacity needed to support long-term AI growth.

As AI infrastructure becomes more complex, infrastructure planning needs to begin earlier than traditional IT upgrade cycles. Evaluating workloads, validating deployment models, and ensuring scalability across environments takes time and time is of the essence if we want to be ahead of our competitors.

The conversation around AI infrastructure often begins with Graphics Processing Units (GPUs). But as deployments scale, AI performance depends not on individual components, but on how the entire system operates together.

Modern AI infrastructure relies on Central Processing Units (CPUs) for orchestration and data movement, GPUs for large-scale parallel compute, high-speed networking for low-latency communication across systems, and open software platforms for portability and scalability.

As AI systems become more distributed and inference-driven, orchestration and system balance become critical. CPUs play a pivotal role in managing workload coordination, memory access, and GPU utilization, ensuring infrastructure operates efficiently under sustained demand.

This shift reflects a broader industry reality: AI is no longer just a GPU problem. It is a full-stack infrastructure challenge that organization must tackle early on.

AI is also scaling in multiple directions at once.

Some workloads are expanding into large, centralized clusters, while others are moving closer to where data is generated – including edge deployments such as in factories or hospitals, and AI-enabled endpoints like the PCs.

For organizations, this creates unique infrastructure considerations around hybrid cloud, on-premises deployments, edge AI, compliance, and latency-sensitive applications.

This diversity underscores the importance of infrastructure strategies designed for modularity, portability, and adaptability that necessitates upfront planning.

As AI innovation accelerates, organizations are prioritizing infrastructure flexibility to support rapidly evolving models, frameworks, and deployment environments.

Open ecosystems can reduce integration complexity while supporting broader compatibility across software frameworks, cloud environments, and deployment architectures. They also provide greater flexibility to evolve infrastructure strategies over time while helping avoid the migration costs that can come with highly closed or single-vendor environments.

For many organizations, openness is no longer just a developer preference. It is becoming an important consideration for balancing performance, operational efficiency, cost optimization, and long-term infrastructure investment.

This is another reason infrastructure planning must happen early. Building AI environments that remain scalable, portable, and adaptable over time require long-term thinking around openness and interoperability from the beginning.

The next phase of AI growth will reward organizations that take a proactive approach to infrastructure planning.

Organizations that delay infrastructure planning may find it more challenging to deploy AI tools down the road, not only due to not having ample time to plan and test, but not securing the compute resources needed early on.

The cost of waiting is becoming ever clearer.

Ultimately, the companies that succeed in the next phase of AI will not necessarily be those with the largest clusters, but those that plan early and build balanced, scalable, and open infrastructure designed to support continuous innovation in an increasingly AI-driven economy.

We've featured the best AI chatbot for business.

This article was produced as part of TechRadar Pro Perspectives, our channel to 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/pro/perspectives-how-to-submit

• Swiss government confirms attackers breached BIT’s SharePoint servers
• Investigators suspect exploitation of recent SharePoint flaws
• No sensitive or confidential data is believed to have been stored on the platform

Cybercriminals broke into the IT network of the Swiss government and stole data from roughly 200 accounts. As a result, the Swiss government disconnected some of its servers from the wider internet and launched an investigation.

In an announcement, the Swiss government said that on July 28 2026 its security specialists noticed “abnormalities” in the Federal Office for Information Technology and Telecommunication’s (BIT) SharePoint servers.

Three days later, on July 31, the investigators determined that the attackers accessed data found in around 200 accounts, both user and technical.

The investigation is currently ongoing, the agency said, adding that it is getting support from Microsoft, as well. So far, the identity of the attackers is unknown, and the stolen data has not yet leaked to the dark web.

“No confidential information or particularly sensitive personal data may be stored on the SharePoint platform,” the announcement reads.

While BIT has not yet determined the initial access vector, it suspects it to be one of two flaws in SharePoint that Microsoft fixed last month:

“In mid-July, Microsoft announced several vulnerabilities in SharePoint,” it says in the announcement. “After the publication of the corresponding security updates, the FOITT immediately started work on importing them into its own systems.”

“The cyberattack was carried out by previously unknown actors, which was presumably made possible by exploiting these vulnerabilities in the SharePoint software.” It did not say which vulnerabilities those are, but in its report, BleepingComputer says that it could be one of these two: CVE-2026-56164 (an actively exploited privilege escalation vulnerability), or CVE-2026-50522 (a critical remote code execution flaw later exploited to steal SharePoint machine keys and maintain access after servers were patched).

Given its popularity among businesses of all sizes, SharePoint is a major target for cybercriminals. So far, no threat actors claimed responsibility for the attack, or demanded any ransom in exchange for the stolen data.

Via BleepingComputer

• Kit Connor has reportedly joined Marvel's new X-Men movie
• The Heartstopper star is said to be playing Scott Summers/ Cyclops
• He joins Sadie Sink and Samara Weaving as part of the hotly-anticipated MCU film

Marvel has reportedly found the star it wants to portray beloved X-Men member Scott Summers/ Cyclops in its cinematic universe.

Per Deadline, Heartstopper actor Kit Connor has been hired to play the mutant group's in-the-field leader. Typically, Marvel hasn't commented on the report, but Deadline is extremely reliable when it comes to casting updates. It stands to reason, then, that Connor is the latest star to land a role in the Marvel Cinematic Universe (MCU).

If true, Connor will be the third actor to sign on for the new X-Men movie. Sadie Sink, who was confirmed to be the MCU's version of Jean Grey in Spider-Man: Brand New Day, is all but locked in for a role in Marvel's mutant-starring flick. Meanwhile, Samara Weaving was reportedly cast as popular superhuman Emma Frost on July 31, with Deadline again being the first to break the news of her apparent involvement.

Little else is known about Marvel's X-Men film reboot. According to Variety, though, the comic giant has ambitions on releasing it as the first project of Marvel Phase 7. If true, the untitled flick will precede the Ryan Gosling-starring Ghost Rider and David Jonsson-led Black Panther III, both of which are primed to arrive in 2028.

Marvel may have locked in Sink, Weaving, and Connor for its forthcoming X-Men film, but fans are desperate to learn who else will feature alongside them.

Speculation has been rife about which other stars could appear, and you can catch up on the biggest rumors by reading my guide on everything we know so far about Marvel's new X-Men movie. Nonetheless, myriad new rumors have circulated online since news of Connor's involvement broke — and I wouldn't be surprised if there's some truth to all of them.

For starters, comic book creator Rob Liefeld took to X/ Twitter to write he'd heard that Inde Navarrette had officially signed on for the project. Liefeld's comments come weeks after the breakout star of Obsession confirmed she'd met with X-Men movie director Jake Schreier and, more recently, threw her hat in the ring to play Mystique.

Just gonna get it out of the way say I was also wrong about Pelphrey as Xavier. I heard a different name that’s not Skarsgard and that name is going around already so I’m sure it’ll leak soon but I’ll let it come out on its own. https://t.co/6bmdqgcD4yAugust 7, 2026

Elsewhere, trustworthy industry insider Apocalyptic Horseman has walked back on Tom Pelphrey (Task, Iron Fist) playing Charles Xavier/ Professor X, while another leaker in Caleb Williams has claimed that fan-favorite mutant Gambit won't be part of proceedings.

Meanwhile, on the latest edition of The Hot Mic podcast, the less-reliable Jeff Sneider has suggested Charles Melton (Beef season 2) is set to play Beast. Fellow podcast host John Rocha has also heard that Cailee Spaeny (Alien: Romulus, Beef season 2) has been tapped to portray Rogue, although Sneider believes she hasn't been.

Lastly, Sneider has indicated that Bill Skarsgard (It, Nosferatu), Ethan Hawke (Moon Knight, Blue Moon) and Jonathan Groff (Frozen, Wicked) are all in the running to play Professor X. Rocha, though, has heard Skarsgard isn't being looked at for that highly-coveted position.

As the X-Men casting rumor mill continues to capture everyone's attention, it might not be long until Marvel puts a pin in the speculation.

Indeed, with the 2026 edition of D23 Expo, aka the so-called ultimate fan event for Disney diehards that takes place every two years, being held between August 14 and 16, some fans believe Marvel will announce the film's full roster during said event. Let's hope that comes to pass because, I don't know about you, but I can't take many more months' of conjecture.

Want to see how the latest Spider-Man film tentatively sets up the next X-Men movie? Read my Spider-Man: Brand New Day ending explained piece. Once you're done, find out how to watch the X-Men movies in order.

Recruitment has long relied on signals that serve as proxies for candidate skillsets: degrees, previous job titles, years of experience, and employer names. Those indicators don’t always show whether someone actually has the skills needed for the role they’re applying for. As AI reshapes work, businesses need clearer evidence of what people can do.

This imperative has led to a drastic shift in hiring norms In the UK, 99 per cent of employers are using skills-based hiring in some capacity, according to research. At the same time, 58 per cent expect more than a third of core job skills to change by 2030. Recruiters now need to hire for current requirements while also thinking about how roles are likely to evolve.

For hiring teams, five changes stand out.

AI is accelerating the move away from degree-first evaluation. When tools, business needs, and ways of working change quickly, employers can’t just rely on static indicators of academic achievement. They need to know whether candidates have practical, current and job-relevant skills.

Degrees still retain significant signaling value. A degree offers an essential foundation, particularly for critical thinking, communication and domain knowledge. But employers increasingly want additional proof that a candidate can apply those strengths in workplace settings.

For recruiters, job descriptions and selection criteria need to become more skill-oriented. Instead of asking for broad experience in a field, businesses should define the specific capabilities needed for the role: data analysis, AI literacy, cloud computing, cybersecurity awareness, project management, or the ability to interpret AI-generated outputs. A clearer skills profile can also help organizations identify strong candidates who have not followed traditional pathways.

Experience still carries weight, but AI is changing how that experience is judged. In fast-moving areas such as generative AI, data, and cloud, a candidate with recent, verified learning may be better prepared than someone with greater experience but outdated skills.

That’s why 42 per cent of UK employers say they would choose a less experienced candidate with a GenAI credential over a more experienced candidate without one. It shows how quickly the value of verified AI capability is rising.

This has practical implications for recruiters. Years of experience shouldn’t be treated as a signifier of readiness. In some roles, recent evidence of applied learning may be more useful. The challenge is to distinguish between candidates who’ve completed primarily theoretical training and those who can show they’re ready to apply what they’ve learned.

AI has made it easier for candidates to produce polished CVs, cover letters and portfolios. It has also made it harder for employers to know which evidence to trust.

This is likely to increase the value of credentials that verify skills to employers. In the UK, 95 per cent of employers say micro-credentials help identify candidates with real-world, applied expertise in areas such as AI, data, and cloud. That matters because recruiters need evidence that can stand up to scrutiny.

The most useful credentials are those that assess applied skills, rather than only content completion. Employers will increasingly look for evidence that a candidate has built something, solved a practical problem or completed a project relevant to workplace needs.

Recruitment teams should reflect this in their processes. CV screening should be supported by practical assessments, structured interviews, and work-sample tasks. This blended approach will make decisions more accurate.

AI literacy is becoming a common requirement, but it is insufficient to simply possess technical knowledge, particularly for those deploying AI in non-technical roles. Businesses need people who can work effectively with AI, including knowing when to question it.

As AI becomes embedded in everyday work, candidates will need to show that they can evaluate outputs, check sources, spot weak reasoning, and apply context. A candidate who accepts AI-generated content uncritically will introduce risk, irrespective of technical proficiency.

This adds novel new stages to the assessment process. Recruiters may need to ask candidates to critique an AI-generated response, improve a flawed analysis, or explain what further evidence they’d need before making a decision. These exercises can reveal whether someone has the judgement and domain expertise required to use AI tools responsibly.

For many roles, the differentiator won’t be whether a candidate can prompt a system: it’ll be whether they can turn AI output into sound business action.

AI is compressing the shelf-life of skills. If, as expected, over a third of core skills will change by 2030 for many UK employers, hiring alone will not be enough to keep pace. In fact, 74 per cent of tech leaders acknowledge they cannot depend on new hires alone to fill AI skills gaps. That means skills-first hiring needs to become part of a wider talent development framework.

The same approach that helps recruiters identify the capabilities needed for a role can also help businesses map the skills they already have, spot gaps across the workforce, and create clearer routes for employees to build the capabilities the organization will need next.

Some capabilities will still need to be brought in from outside the organization, particularly in fast-moving areas such as AI, data and cloud. But many skills will also need to be developed internally as roles evolve. The strongest businesses will combine skills-based hiring with ongoing upskilling, giving employees clear routes to adapt as roles change.

AI is changing what employers need to know about candidates. Businesses that can identify real capability, verify applied skills, and recognize potential beyond traditional signals will be better placed to hire well. The value of this approach is already visible in workplace outcomes. 92 per cent of employers say entry-level hires with micro-credentials perform better in their first year on the job.

For candidates, the message is just as clear. In an AI-enabled labor market, employability will depend less on what someone once learned and more on what they can prove they can do now, and the evidence they’re still learning.

For employers, recruitment needs to become part of a broader skills strategy. AI may be changing the work, but the hiring challenge remains human: finding people with the skills, judgement and adaptability to help businesses compete and innovate.

We've featured the best website for hiring niche employees.

This article was produced as part of TechRadar Pro Perspectives, our channel to 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/pro/perspectives-how-to-submit

• Microsoft temporarily rolls back planned Teams tool concerning muted chats
• The update would have made it easier to organize (and ignore) unnecessary conversations
• It's not known when the update will get back in progress

Microsoft has pulled an update which would have made it much easier for Teams users to organize their chats - particularly those which have been muted.

The company had revealed it was working on new built-in sections within Teams to house all the muted chats in one area, meaning users shouldn't be distracted or annoyed by unnecessary alerts.

However the update has now been "temporarily rolled back", with Microsoft apologizing for the "inconvenience".

The move is a slightly odd one, as the description of the update makes it sound like something all Teams users would appreciate.

In its Microsoft 365 roadmap entry, the company said the new grouping will "help reduce clutter and keep conversations easy to find" - but also probably help lower stress levels for users plagued by alerts on chats they don't need to know about.

The change, which was in the works for users across Android, iOS, desktop and web, would have seen muted chats grouped together in a dedicated area, while meeting chats would have been able to be collected into a single section for quick access.

Microsoft had even added that "You’re always in control", as each section would have been able to be activated or turned off based on a user's specific preferences.

The new Meeting chats section would have replaced the Meeting chats filter, found at the top of the chat and channel list, giving users a much better insight to their conversations.

Microsoft hasn't given any details as to why the update is delayed, or when it will be coming back onto the workbench, so we'll just have to wait and see.

As mentioned, it's a shame, as Teams users no doubt would have welcomed any relief from the overload of chats. Slack already offers this feature, with muted channels and conversations hidden away if desired, so maybe Microsoft is taking extra time to finesse and develop the feature before releasing it to users.

Recently in Geneva, the world's largest forum on AI for Good, gathered around a simple question: how can technology best serve society? While the sentiment goes back to pre-industrial times, the conversation has firmly moved to algorithms. What about the machines that will come next?

Quantum technology has crossed a line. It is no longer a distant promise, but a growing portfolio of real applications. Quantum and quantum-inspired solutions are being applied to the simulations that underpin nuclear power, to protein design in drug discovery, and traffic forecasting in congested cities.

Quantum sensors are being tested for medical imaging, navigation where satellite signals fail, and monitoring carbon storage sites deep underground. These are no longer mere laboratory curiosities: they are increasingly emerging as working programs, with industrial partners and delivery dates.

For those of us in the industry, these developments are both timely and long anticipated. But they also force us to confront a less comfortable reality: the same machines that will one day simulate new molecules will also break the public-key cryptography that secures much of the modern internet.

By some estimates, up to 80% of the world's digital infrastructure is exposed. The world's digital economy - not to mention every AI model celebrated in Geneva, runs on encrypted data. Every health record or financial transaction owes its confidentiality to mathematical assumptions that a sufficiently powerful quantum computer will overturn.

The threat timeline is worse than most realize. Rather than waiting for when the machine arrives, the clock started the moment adversaries began harvesting encrypted traffic: the so-called “store now, decrypt later” approach.

Data that remains valuable for years, from medical records to state secrets, may already be beyond rescue. Nobody knows exactly when “Q-Day” comes. But those who harvest our data today have time on their side.

This is why quantum-safe security cannot be a footnote to quantum computing. It is the other half of the field, and it needs to mature on the same schedule.

Two defenses exist, and they are not interchangeable. Post-quantum cryptography replaces vulnerable algorithms with mathematics believed to resist quantum attacks; it is software, it scales, and it is the workhorse of the migration now beginning worldwide.

Quantum key distribution takes a different route: it distributes encryption keys using quantum states of light that cannot be intercepted without being disturbed. In this way, these schemes deploy a “quantum trip-wire", an eavesdropper reveals herself by the simple act of listening-in

Far from being in competition, both algorithmic and physics-based solutions have roles to play in tomorrow's networks. Serious infrastructure will layer both.

Neither of these solutions are theoretical. New post-quantum standards have been finalized, with contributions from research teams around the world. And quantum-secured networks are already running in the field: since 2022, Abu Dhabi has hosted entanglement-based metropolitan networks, like the ADGM Quantum Testbed announced in August 2025.

These are initiatives that bring quantum security to where data actually lives. Long-range terrestrial links, satellite connections and “last-mile” access networks are the next steps, extending that protection between cities and, eventually, across continents. If there's one lesson the field has learned from operating these systems, it's this: migration takes years, so the time to start is before the threat matures, not after.

There is a broader lesson in how this capability is being built. A growing number of nations have chosen to be builders of quantum technology rather than buyers, developing the full portfolio of sensing, communications and computing solutions. From processor fabrication to control software, to the hardware enabling quantum networks: the word “sovereignty” is often heard as a synonym for walls.

But our experience suggests otherwise. It is precisely the teams that build every layer themselves that can contribute most readily to the global commons, from working with international standards organizations like ITU and ETSI on quantum-safe networking, to releasing open-source middleware that researchers worldwide now use to program quantum hardware, wherever it was made.

It is this depth of capability that turns a nation from a spectator of the quantum era into an active participant.

That should also be the model for the decade ahead. No single country will own quantum technology, and no single vendor should be the sole authority on quantum-era trust. What the field needs now is what the AI community showcased in Geneva: open tools that lower the barrier to entry, interoperable standards so quantum-safe systems can talk to each other, and honest engagement with the risks alongside the promise.

For years, the defining question about quantum computing has been when the machines will arrive. Increasingly, I think that's the wrong question. The more important question is whether the digital world they inherit will still deserve their trust. The answer depends on what we choose to secure, build, and share today.

Quantum for good does not start with a breakthrough. It starts with trust.

We've featured the best desktop PC.

This article was produced as part of TechRadar Pro Perspectives, our channel to 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/pro/perspectives-how-to-submit