Edifier has expanded its MR Studio Monitor Series for 2026 with the new MR4.5, a near-field active loudspeaker designed for content creators, musicians, producers, and desktop listeners who want greater flexibility without moving into the larger and more expensive MR5.

Positioned between the MR4 and flagship MR5, the Edifier MR4.5 combines studio-focused sound reproduction with wireless connectivity, customizable tuning, and a compact design suitable for home studios, editing suites, gaming systems, and desktop audio setups.

The new model joins the MR3, MR4, and MR5 to complete Edifier’s current studio monitor lineup, offering buyers another option in a category where accuracy increasingly has to share desk space with convenience.

Related Reviews:

Features and Connectivity

Two Speakers: The Edifier MR4.5 includes two speakers. The active speaker serves as the system hub, handling connectivity, amplification, and signal distribution for both channels. No external amplifier, stereo receiver, or home theater receiver is required. The passive speaker connects to the active speaker using the supplied cable, which carries both power and audio signals.

Adaptable Setup: The MR4.5 includes acoustic tuning and desktop compensation controls that allow users to adjust its performance for different rooms, speaker positions, and listening distances. These controls can be useful when the speakers are placed on a desk, close to a wall, or in a small home studio where placement options may be limited.

Wired and Wireless Connectivity: The MR4.5 supports both wired connections and wireless streaming. Bluetooth 6.0 includes LDAC support for higher-resolution wireless audio, while multipoint connectivity allows the speakers to remain connected to more than one compatible device. This makes it easier to switch between a computer, smartphone, tablet, or other source without repeatedly disconnecting and pairing devices.

Bi-Amplified Speaker Design: The MR4.5 uses a bi-amplified design with an active crossover that divides the audio signal between the mid-bass driver and tweeter. Edifier designed the system to provide a balanced transition between the drivers and accurate sound reproduction for music production, video editing, gaming, and everyday desktop listening.

Extended Frequency Response: The MR4.5 has a specified frequency response of 50 Hz to 40 kHz. Edifier has tuned the system for a neutral, transparent presentation intended to reproduce recordings without excessive tonal coloration.

Dimpled Tweeter Waveguide: The MR4.5 uses a dimpled tweeter waveguide designed to improve high-frequency extension and clarity while creating a wider listening area. This should provide more consistent performance when the listener moves away from the central listening position.

80 Watts of Total Output: The MR4.5 delivers 80 watts RMS of total system output and a maximum sound pressure level of 98 dB at one meter. Although primarily intended for near-field listening, it should also provide sufficient output for some mid-field setups and medium-sized rooms.

Hi-Res Audio Support: The MR4.5 uses an analog front end and Class D amplifiers and supports digital audio signals up to 24-bit/96 kHz. Wireless playback is available through Bluetooth with LDAC support. Under compatible conditions, LDAC can transmit audio at up to 990 kbps and support signals up to 24-bit/96 kHz.

Wired Connection Options: Balanced XLR and TRS inputs allow the MR4.5 to connect to professional audio interfaces, mixers, and other studio equipment. RCA and 3.5 mm auxiliary inputs accommodate common consumer audio sources, while a headphone output supports private listening.

Bluetooth Multipoint: Bluetooth 6.0 provides a stable wireless connection, while Bluetooth Multipoint allows two devices to remain connected simultaneously. Users can switch between sources without disconnecting and re-pairing each time.

Room Acoustic Compensation: Rear knobs and the EDIFIER ConneX app provide multiple settings for room acoustic compensation, allowing you to adapt to different room modes and achieve the optimal listening experience.

Edifier MR Studio Monitors Comparison

The Bottom Line 

Although Edifier has enjoyed success with its MR Studio Monitor Series, a substantial gap remained between the MR4 and MR5 in both price and features. The MR4.5 fills that space by adding Bluetooth, LDAC, balanced XLR connectivity, room compensation, app-based adjustments, and an active bi-amplified crossover—features not offered by the more basic MR4.

The MR4.5 also provides a logical step down from the MR5, which uses a larger three-way driver configuration, a tri-amplified active crossover, and 110 watts of total output. Those additional capabilities may make the MR5 better suited to larger rooms and more demanding studio applications, but they may be unnecessary for desktop users and smaller production spaces.

The MR4.5 does not include HDMI ARC or eARC, USB audio, optical connectivity, or a phono input. Buyers who require HDMI eARC, USB-C, and optical inputs should consider Edifier’s M90 active speakers. The Onkyo Creator Series GX-30ARC goes further by combining HDMI ARC, USB-C, optical, and a line/phono input for direct connection to a turntable with a moving-magnet cartridge.

Price & Availability

Related Reading

For years, musicians have said AI song generators fed on their work without asking. A hacker just opened the black box and showed them exactly how.

Leaked source code from Suno, one of the biggest AI music tools, tells the story. It trained its model by scraping millions of songs and lyrics from across the internet. 404 Media got the data from a hacker who breached the company.

The haul is enormous. One file, labelled “youtube_music,” had logged more than 2 million clips. Others list tens of thousands of hours pulled from Deezer, Genius, and the stock library Pond5. In all, it adds up to decades of recorded music.

Ripping the vocals

The code is specific about what it wanted. To capture clean voices, it hunted for a cappella versions of songs on YouTube. To slip past YouTube’s defences, Suno routed its scraping through a proxy firm called Bright Data. It also trawled 420,000 podcasts, chasing roughly a million hours of speech.

The 💜 of EU tech

The latest rumblings from the EU tech scene, a story from our wise ol’ founder Boris, and some questionable AI art. It’s free, every week, in your inbox. Sign up now!

None of this is entirely new. In court, Suno has already admitted training on “essentially all music files of reasonable quality” on the open web. But the leak shows the machinery behind that line.

The record labels have long alleged as much. Suing Suno, the RIAA said the company copied “decades worth of the world’s most popular sound recordings,” and did it by “stream ripping” from YouTube, dodging the platform’s copy protections.

‘Fair use,’ Suno says

Suno’s defence, like most AI firms, is fair use. It says it trains on “publicly available music files” and builds its models for “original creation.” It even leaves artist names out of its training data, to discourage copycats.

On the breach, the company played it down. It called the November 2025 incident “limited” and “quickly contained,” said the exposed code was outdated, and insisted no sensitive data leaked. It does not hold customers’ full card numbers, it added, and it decided it did not need to tell users at all.

The hacker, who goes by ellie.191, tells a different story. They said they got in through the Shai-Hulud worm, grabbed an employee’s credentials, and pulled customer emails, phone numbers, and Stripe records. Suno never warned those users, some of them told 404 Media. The motive? “I like to hack anything and everything.”

The bigger fight

The timing matters. Some labels have already made peace, striking licensing deals with AI firms rather than fighting on. Sony is still in court, with a pivotal fair-use ruling due this summer.

Meanwhile artists keep saying the deals do little for them. Suno’s chief executive, Mikey Shulman, once said most people “don’t enjoy the majority of the time they spend making music.” The people whose music trained his model might disagree.

A team of ten domino builders spent many days organizing over 30,000 tiles into a gigantic three-dimensional pyramid. Then, on one fateful day and time, all of those tiles came tumbling down in an incredible display. The FALLDOWN Domino Team, led by Steven Price of Sprice Machines, completed the feat on June 20th, 2026, in a gym in Garden City, Michigan. Guinness World Records has now taken a look at the achievement and made it official: the 3D pyramid is the new record holder after 29,193 dominoes came crashing down in one spectacular chain reaction.


The pyramid was the centerpiece of a much larger installation that included no fewer than 123,456 dominoes in total. So how did the team bring this mammoth of an installation to life? They created a variety of incredibly huge ordinary goods, like giant paperclips, enormous friendship bracelets, massive bonsai trees, school supplies, and even a rubber ducky that appears out of nowhere. This all led to a series of chain reactions that snaked through the entire device and fed into the pyramid, giving the record attempt one more push. Because it was so clean and white, a burst of multi-colored dominoes pouring up the side of a building was rather beautiful, while yet meeting all of the criteria for a successful 3D pyramid.

Sale

LEGO Architecture Trevi Fountain Building Set – Collectible Model of Italian Landmark for Display for…

• ICONIC LANDMARK MODEL – Embrace every step in the architectural recreation of an iconic Rome baroque monument with this 1,880-piece Trevi Fountain…
• HISTORY & TRAVEL-LOVERS MEMORABILIA – Your building journey lets you recreate the iconic Rome fountain in all its glory, while enjoying updated…
• AUTHENTIC DETAILS – Admire the facade’s backdrop, the details of the statue figures, mini hippocampi and small ’Lovers’ Fountain’, plus…

Building one of these things (a 3D pyramid, as the foundation must be just right or you’ll end up with a pile of dominoes on the floor) needs near-perfectionism. If you get it wrong, you’ll lose everything before you’ve even dropped the last domino. This one’s base was a 35×35 grid of dominoes stacked in layers to form a pointed apex. To get here, they approximately doubled the number of dominoes Andreas Zauner managed in 2020 in a similar challenge. This required a lot more tiles and a lot of rigorous engineering to ensure that everything stayed up during the construction.

Steven Price has set multiple records with his FALLDOWN crew, including the longest domino wall and the largest 2D pyramid, to name a few. So this time, they decided to try with 3D pyramids. Oh, and they were going to use a special mechanism to bring the whole thing down, but Guinness eventually convinced them to just let the dominoes do their thing. That’s right, in order for it to be officially recorded, the entire process must begin with a traditional domino-to-domino touch.
[Source]

What a difference eight days makes.

On July 7, Dynaudio announced that it would cease commercial operations in North America and permanently close its U.S. subsidiary in the fall of 2026. The Danish loudspeaker manufacturer said it was redirecting its market development efforts toward Europe and Asia, despite acknowledging that North American sales had grown in recent years.

The announcement sent shockwaves through the high-end audio industry. Dynaudio had just shown major new products at AXPONA 2026 and HIGH END Vienna, including the $7,000 Legend bookshelf loudspeaker and Symphony Opus One immersive audio system. eCoustics had to cancel multiple forthcoming Dynaudio reviews, including our planned evaluation of the Legend.

Now Dynaudio’s parent company, Goer Dynamics, has announced a strategic partnership with IMAX and IMAX China to create what the companies describe as the world’s first IMAX branded in-vehicle entertainment system.

The system is expected to enter commercial production by the end of 2026 and will initially be marketed to premium automakers in China, although none have been named at this time.

North America may no longer justify the cost of maintaining a traditional high-end loudspeaker business. China’s enormous electric-vehicle market, apparently, is another matter.

What Are IMAX and Goer Dynamics Building?

The proposed system combines an IMAX certified 4K HDR flip-down display with an IMAX developed multidimensional acoustic architecture.

The large display will use custom image processing and ambient-light adaptation intended to preserve picture quality as conditions inside and outside the vehicle change. That matters in a car, where sunlight can turn an otherwise respectable display into an expensive black mirror before you have reached the end of the driveway.

Goer Dynamics and IMAX also promise high dynamic range audio, substantial low-frequency output and controlled bass distortion. Modular configurations will allow automakers to adapt the system to different vehicle platforms, cabin layouts, speaker counts and trim levels.

The companies are positioning the system primarily for autonomous vehicles, where the cabin can become what the automotive industry insists on calling a “third living space.” That phrase sounds considerably more appealing than “the place where you spend three hours moving six miles on the Garden State Parkway.” As someone who has spent enough time trapped on New Jersey highways to watch the extended edition of Lawrence of Arabia, I understand the opportunity.

The Dynaudio Connection

The partnership is between IMAX and Goer Dynamics, not directly between IMAX and Dynaudio.

Goer Dynamics was founded in 2020 under China’s Goer Group and owns and operates Dynaudio, XEO and Libratone, and also holds a minority stake in fellow Danish high-end manufacturer Gryphon Audio Designs. Its businesses cover home audio, professional studio monitoring, automotive systems, consumer electronics and audio-visual technology. The company claims to have supplied in-vehicle entertainment systems for nearly three million new-energy vehicles.

Dynaudio became part of the wider Goertek organization when the Chinese electronics manufacturer acquired a majority interest in the Danish company in 2014. Dynaudio said at the time that the acquisition would give it access to additional engineering expertise in electronics, wireless technology and manufacturing.

Automotive audio is not a side project that appeared after someone discovered a large dashboard and an empty licensing agreement. Dynaudio has worked with Volvo, Volkswagen and Bugatti, and more recently developed premium systems for Chinese automaker BYD. The BYD Seal, for example, offers a 12-speaker Dynaudio system rated at 775 watts, while the Yangwang U8 luxury electric SUV uses a 22-speaker Dynaudio Evidence system.

That history gives Goer Dynamics considerable automotive acoustic experience, even though the announcement does not confirm whether Dynaudio engineers, drivers, DSP technology or branding will appear in the finished IMAX system.

Why China Comes First

China is the logical starting point.

More than 13 million electric cars were sold there in 2025, representing approximately six out of every ten EVs sold worldwide. The International Energy Agency expects electric vehicles to approach 60% of Chinese new-car sales during 2026.

Chinese manufacturers are also competing aggressively over cabin technology. Large displays, premium audio, karaoke, gaming, streaming video, reclining seats and smart-cockpit features have become important differentiators, especially in higher-priced electric vehicles.

Neither company has identified a launch partner, although Goer Dynamics’ existing relationships offer some obvious possibilities. Dynaudio currently supplies systems for BYD vehicles including the Seal and the premium Yangwang U8, while Goer Dynamics has also announced cooperation with Hongqi and Voyah. Those brands would be logical candidates for the first IMAX-equipped vehicles, but there is no confirmation that any has signed on.

For Dynaudio’s parent company, the potential scale is vastly different from selling $7,000 bookshelf loudspeakers through a shrinking network of specialist North American dealers.

That does not make Dynaudio’s withdrawal from North America any less disappointing. It does help explain where the company’s owners believe the larger opportunities now exist.

IMAX Has Already Been Inside a Car

The claim that this is the first IMAX branded in-vehicle entertainment system requires some qualification.

IMAX Enhanced content has already been offered in select Mercedes-Benz vehicles through Sony Pictures Entertainment’s RIDEVU service. The platform delivers selected films with IMAX’s expanded aspect ratio and remastering process, accompanied by DTS audio. It can distribute content across as many as six built-in or connected screens, although the driver can watch only while the vehicle is parked.

We experienced the Mercedes-Benz demonstration at CES 2025 and gave the IMAX Enhanced DTS system a Best in Show award. The sound was far more convincing than anyone had a right to expect inside an E-Class sitting in a convention center.

The new Goer Dynamics project appears to go further by offering automakers a complete IMAX branded hardware and acoustic platform rather than adding IMAX Enhanced content to an existing infotainment system.

That is a meaningful distinction, but IMAX is not entering the automotive market from a standing start.

What We Still Do Not Know

The announcement contains a substantial amount of language about immersion, dynamic range and cinematic detail, but very few specifications.

IMAX and Goer Dynamics have not disclosed:

• The size, resolution beyond 4K, brightness or display technology of the screen
• The number or type of loudspeakers
• Amplifier power or system frequency response
• Whether Dynaudio drivers or acoustic technologies are involved
• Whether the finished product will carry Dynaudio branding
• Support for DTS, Dolby Atmos or another immersive audio format
• Compatibility with IMAX Enhanced movies
• Streaming, rental or download partners
• Available content libraries
• Connectivity requirements
• Subscription costs
• Which automaker will become the first confirmed customer
• Which markets will follow China
• Whether the display can be used only by rear passengers or while the vehicle is parked

We also do not know what IMAX certification means in this specific environment. A cinema auditorium is built around controlled light, fixed seating positions and carefully placed loudspeakers. A vehicle cabin contains glass, reflective surfaces, road noise, moving passengers and seats that may recline, rotate or slide.

Producing consistent immersive sound inside that environment is not impossible, but it requires far more than placing a logo on a screen and adding enough bass to shake loose the toll receipts.

The Bottom Line

The IMAX and Goer Dynamics partnership is more significant than another oversized rear-seat display. It combines IMAX’s entertainment brand and image technology with a company that has already supplied systems for nearly three million new-energy vehicles and owns one of the most respected names in loudspeaker engineering.

It also arrives at an uncomfortable moment.

Dynaudio is preparing to leave North America, close the subsidiary it built to support dealers and customers, and focus on Europe and Asia. Multiple eCoustics reviews disappeared from our schedule almost overnight. One week later, its parent company unveiled a potentially large automotive partnership aimed squarely at China.

The message is difficult to miss. Traditional high-end audio remains culturally important, but automotive entertainment may offer the scale, recurring technology partnerships and overseas growth that selling passive loudspeakers through North American dealers no longer provides.

Whether any recognizable Dynaudio technology reaches the finished IMAX system remains unknown. So do the screen size, speaker arrangement, audio format, content services, automakers and price.

Watching Oppenheimer on the New Jersey Turnpike may soon be technically possible. Missing your exit near Secaucus because the Trinity test sequence was getting interesting will remain entirely your fault.

[Source: businesswire.com]

Related Articles:

Xgimi doesn’t seem to tire of launching more projectors and we bring news of of another two to keep track of.

The Flip series is being expanded with the Elfin Flip Laser and Elfin Flip 4K joining the range, as Xgimi looks to widen the range of portable home entertainment experiences its offers, while also upgrading both “optical performance and audio experience”, with RGB Triple Laser tech and Hardon Kardon tuned sound.

Xgimi states that the Elfin Flip 4K is meant for travel or intended to be a “casual streaming device”, like its MoGo series. It describes it as a “performance-class” home projector, weighing 1.55kg and measures about 25cm wide, ensuring it can be transported between the rooms of your home, producing a “full cinematic” experience wherever you choose to place it.

The Flip 4K is powered by a 4K RGB Triple Laser engine that can deliver 1600 ISO lumens of brightness, 110% of the BT.2020 colour gamut and a Delta E of less than 1 colour accuracy, the kind of specs you don’t often see for home projectors such as this.

Advertisement

For those who want a simple installation, Xgimi’s Intelligent Screen Adapt (ISA) can get the picture ready automatically, with smart features such as Uninterrupted Auto Keystone, Auto Focus, Intelligent Screen Alignment, Intelligent Obstacle Avoidance, Wall Colour Adaptation, and Intelligent Eye Protection ensuring you get the best image for the space the projector is in.

Advertisement

The Flip 4K is also made with gamers in mind, as Xgimi says it can deliver input lag at less than 1ms at 1080p/120Hz, and it packs in VRR and ALLM support. The Elfin Flip Laser doesn’t feature the same level of gaming support, and the resolution takes a hit, dropping down to 1080p.

Both projectors feature a 7W Harman Kardon speaker that Xgimi says can deliver a room-filling sound experience that “eliminates the need for external audio equipment” (that’s a big claim). There’s also Google TV for entertainment, offering nearly all the apps (for the UK, we’ll assume that iPlayer and Channel 4 are still missing).

Availability for both projectors starts July 15th on the Xgimi website, with availability on Amazon starting July 22nd. Prices for the Elfin Flip 4K is $999 / £869, with the Elfin Flip Laser at $799 / £689.

Advertisement

Advertisement

Automation and [SvHCI’s VMware] VM Import Utility were absolutely vital to scaling this migration. Operating in a 24/7/365 retail environment meant that minimizing business disruption was critical. It required meticulous planning and heavy automation to ensure our store operations ran as smoothly as possible throughout the entire transition.

SvHCI product maturity in relation to APIs meant it was a small amount of extra work, and the main challenge of the migration was [finding] the time available to do it, for the scale of the environment. They were having to simultaneously plan, develop, and implement.

For many companies, the idea of moving off VMware is daunting due to the money, time, and staff that it may require. Some also report challenges in finding alternatives with the same capabilities and compatibility as VMware. Total or even partial migrations can seem particularly implausible for organizations that depend on VMware technology.

As a result, there are many VMware customers interested in quitting or reducing their use of VMware products, but have yet to make the move or are still in the planning phases. In September, Gartner estimated that 35 percent of VMware workloads would migrate elsewhere by 2028.

StorMagic targets VMware’s larger customers

StorMagic has a reputation for serving small-to-medium-sized businesses (SMBs), but today’s announcement highlights its interest in winning over the enterprise-sized firms that Broadcom’s VMware strategy targets, especially enterprises with numerous SMB-sized locations.

“In reality, we have always focused heavily on two distinct markets: SMB/mid-market datacenters and the ‘edge’ environments of large, highly distributed enterprises, like Sheetz. A distributed enterprise with hundreds or thousands of retail, grocery, or branch locations actually faces similar IT challenges at each site as a local SMB,” Scott Mann, StorMagic’s SVP of global sales, told Ars via email, pointing to these organizations having limited physical space, power, on-site technical staff, and budget.

The executive sees further opportunity among VMware’s current enterprise clients.

“Historically, large enterprises tolerated the ‘VMware tax’ at their edge locations because it was the status quo. However, with recent massive industry shifts, specifically Broadcom’s acquisition of VMware, enterprises are facing massive budget increases just to keep their remote sites running,” Mann said.

Other enterprises recently revealed to be migrating off of VMware include Allstate, T-Mobile, and UK grocery chain Tesco.

For its part, Broadcom has argued that changes to VMware’s licensing model are in line with the rest of the industry, and its acquisition of VMware is considered financially successful.

Beta Pictoris d is estimated to be around two times the mass of Jupiter.

NASA scientists have discovered a new planet orbiting a neighbouring star located 63 light years away from us. The new exoplanet, named ‘Beta Pictoris d’, is the third to be found contained within the Beta Pictoris planetary system.

The 23m-year-old star Beta Pictoris offers scientists a rare glimpse into how newborn planetary systems form, and how its young planets interact with the dust and residual material left behind from their formation.

The sun, in comparison, is around 4.5bn years old.

The new discovery makes Beta Pictoris the second planetary system ever known to contain at least three planets that have been imaged, NASA said today (15 July).

According to the team behind the discovery, Beta Pictoris d is estimated to be around two times the mass of Jupiter, while orbiting its star at around 30 astronomical units – which is comparable to the region Neptune occupies in our solar system. It’s the smallest of the three exoplanets orbiting this star, and takes the widest orbit of the known three.

Beta Pictoris d remained hidden under one of the brightest debris disks known to us, concealing it from traditional discovery techniques. It was discovered rather unexpectedly using the James Webb Space Telescope’s (JWST) Near-Infrared Spectrograph.

“There was an unexpected bright source of light within the Integral Field Unit imaging, but we’ve learned not to trust bright blobs in images,” said Jean-Baptiste Ruffio, a research scientist at University of California, San Diego and principal investigator of the first Webb observations where the discovery was made.

“They can be instrumental artifacts or other structures in the debris disk. By obtaining a spectrum at the same time as the image, we were able to quickly confirm our suspicions.”

The new spectroscopy technique also revealed the object’s motion, allowing scientists confirm that the exoplanet is indeed orbiting Beta Pictoris, rather than a behaving like a background star or a brown dwarf with carbon monoxide in its atmosphere.

This is one of the first times researchers have discovered new planets mainly using moderate-resolution spectroscopy.

Scientists say this new discovery could help explain some of the puzzling structures of the Beta Pictoris debris disk.

“This discovery adds another piece to an already fascinating planetary system,” said Aidan Gibbs, the lead author of the new study published in the Astrophysical Journal Letters.

“Beta Pictoris has long served as a laboratory for understanding how planetary systems form and evolve, and now we have another planet helping us tell that story.”

Last year, astronomers witnessed the very early stages of a new solar system being created around a baby star roughly 1,300 light years away, while earlier this year, 25-year-old University of Galway scientist Chloe Lawler discovered a 5m-year-old exoplanet some 437 light years away.

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

Disclaimer: Unless otherwise stated, any opinions expressed below belong solely to the author.

Just last month, Apple, the last holdout in the personal computing market, was forced to hike prices of its computers and tablets within the range of 10 to 30%, depending on the type and model.

The giant from Cupertino was able to wait out the AI-induced inflation thanks to its long-term contracts on memory chips and the TSMC manufacturing capacity it had booked for its Apple silicon processors well in advance.

The fact that it has long enjoyed some of the highest margins in the industry must have also helped, providing a buffer that allowed it to absorb some of the costs seeping through in other areas.

Windows PC buyers have had a much worse time for the past year, as the AI revolution hit their devices first. Laptops may still be attainable, but building a new desktop PC is currently nearly impossible for regular consumers after RAM prices exploded by several hundred per cent in late 2025.

This came on top of inflated prices of graphics cards, which AI came for first, as hyperscalers like OpenAI, Anthropic or Google needed to secure millions of them to train their artificial intelligence models.

That said, amid the surge hitting Nvidia and AMD cards, RAM sticks and SSD storage, one component remained unaffected: the central processing unit (CPU).

Unfortunately, it is about to change.

AI does not run on GPUs alone

GPUs are excellent at performing large numbers of relatively similar calculations simultaneously. This makes them indispensable for training artificial intelligence models, which is why they were essential in the early years of the AI boom.

To build your own AI model, you need GPUs—and A LOT of them.

But they do not operate independently.

CPUs still have to prepare and feed data to accelerators, manage memory, handle networking, launch tasks and coordinate all the other processes taking place around the model.

This becomes particularly important during inference—the stage when a trained AI model responds to users.

Training may take place once or periodically, but inference occurs every time somebody asks ChatGPT a question, generates an image, writes code with Claude or tells an AI agent to complete a task.

As the number of AI users and applications grows, inference demand grows with it.

On top of that, the emerging generation of AI agents is particularly hungry for general-purpose computing.

Unlike a chatbot that produces one answer and stops, an agent may browse files, call external tools, execute code, check results and repeat the process multiple times. Each of these actions creates work that CPUs are much better suited to handle.

Until recently, AI companies needed roughly one CPU for eight GPUs, but that ratio is expected to shrink rapidly and may approach 1:1 parity by 2029.

Consider the millions of GPUs that were sold in the past two to three years. Now, their deployments may need three, four, maybe even eight times as many CPUs. And there are new data centres being built as we speak.

So, not only is the new approach going to have to fill existing gaps, but also respond to the future demand.

Intel is already running short

This would not be a problem if chipmakers had large amounts of spare capacity. But they don’t.

Intel acknowledged that demand was already outpacing supply in late 2025 and warned that shortages would persist into 2026. Its data-centre business was unable to fully meet customer demand because of limited wafer capacity at its own factories.

The company is now prioritising the production of server chips, including its more lucrative Xeon processors, as AI demand grows.

This makes commercial sense. One high-end server processor can cost thousands of dollars, while the CPU in an ordinary laptop may cost the manufacturer a fraction of that.

But Intel cannot simply create more factory capacity overnight. If it produces more Xeons using constrained manufacturing lines, something else may have to give.

And that something could be consumer processors.

Industry reports suggest server CPU prices have already risen by as much as 20% since Mar, while consumer models have reportedly become 5 to 10% more expensive in some channels.

Additional increases may follow later this year.

AMD is benefiting too. Its data-centre revenue rose 57% year-on-year to US$5.8 billion in the first quarter of 2026, driven partly by strong demand for its EPYC server processors.

Unlike Intel, however, AMD does not own its leading-edge factories. It relies primarily on TSMC, which is also producing chips for Nvidia, Apple and numerous other companies competing for limited advanced capacity.

So, while Intel has to choose what to produce in its own factories, AMD has to compete for space at somebody else’s, which is the same Taiwanese company everybody already relies on.

The bottleneck is getting tighter.

Ordinary buyers will end up paying too

While server and consumer CPUs are not always manufactured on the same processes, and chip companies cannot freely convert every production line from one product to another, there are several ways the pressure can still reach consumers.

Manufacturers may prioritise their limited capacity, engineering resources and components for more lucrative enterprise products. Computer makers may pay more for chips under their supply agreements. Shortages of resources and input components can also raise the cost of the entire system.

And many high-end consumer processors can be used directly in enterprise settings. While they may not perform the most important and valuable tasks, they can serve in support roles to help save the precious server models for where they are needed most.

That, in turn, would hoover them up from the consumer market and drag the prices of all processors with them, as consumers turn to the next best option.

There is always another bottleneck

The AI boom began with the impression that the industry merely needed more GPUs. It quickly became clear that it also needed memory, storage and things outside of technical components, like electricity, building materials or qualified construction labour.

Now, CPUs are joining the list.

This is what happens when hundreds of billions of dollars are invested in one industry at the same time. Solving one shortage merely exposes the next bottleneck after that.

For consumers, the frustrating part is that they are competing with some of the richest companies in history, many backed by tacit or direct government support, as entire nations see harnessing AI as a strategic interest.

Hardware manufacturers will naturally sell their limited capacity where it generates the highest returns. At the moment, it is increasingly inside AI data centres rather than the computers sitting on our desks.

That’s why, unfortunately, if you were waiting for GPUs and RAM to become cheaper before buying your next PC, there may soon be another item to worry about. And there is no end in sight.

• Read other articles we’ve written on the artificial intelligence boom here.

Featured Image Credit: Shutterstock

OpenAI has trained an elite hacker, then locked it in a cage. Its whole job is to break OpenAI’s own AI. The company says it is too dangerous to let anyone else near it.

The model is called GPT-Red, and OpenAI detailed it this week. It is an automated red-teamer: software that hunts for ways to hijack or sabotage other AI systems, so the holes can be patched before release. Humans have long done this work by hand. It is OpenAI’s deepest push yet into automating its own AI security, and GPT-Red does it at machine speed.

OpenAI aimed it at prompt injection, where hidden instructions, buried in an email, a web page, or a file, trick a model into doing something it should not. Then it set the hacker loose on real targets.

The training dojo

GPT-Red learns by fighting. OpenAI put it in a self-play loop against a squad of defender models. GPT-Red is rewarded for landing an attack; the defenders for fending one off. As the defenders wise up, GPT-Red must invent nastier tricks. OpenAI says it poured some of its largest ever compute runs into the model, an amount it calls unprecedented for safety work.

It got good. Speaking to MIT Technology Review, the team said GPT-Red found a whole new class of attack they had never seen, which they call a “fake chain of thought.” It plants a false note in a model’s private working memory, tricking it into trusting something that is not true.

“It’s like if I told you that 1+1=3 and that you have verified this already,” said OpenAI researcher Chris Choquette-Choo. “The model’s like, ‘Oh, okay, of course,’ and it just spits out 3.”

Hacking the vending machine

The tests got physical. In one, GPT-Red attacked Vendy, an AI agent that runs a real vending machine in OpenAI’s office, built by Andon Labs. It changed the prices, marked a pricey item down to the 50-cent minimum, and cancelled a customer’s order. OpenAI says it has disclosed the flaws.

The scores are striking. Against an older GPT-5, more than 90% of GPT-Red’s strongest attacks worked. Against the new GPT-5.6, fewer than 23% did. In a rerun of a 2025 test, GPT-Red beat human red-teamers hands down, cracking 84% of scenarios to their 13%.

Kept in a cage

OpenAI trained GPT-5.6 against GPT-Red, and calls it its most robust model yet against prompt injection. But it will not hand out the attacker itself, so its skills stay clear of real agent hijackers. It is not the first lab to build something and decide against releasing it.

“It’s not a trivial thing that someone could easily do,” Choquette-Choo said, “just go and train a super-attacker using this idea.”

GPT-Red still has blind spots. It is weak at drawn-out, back-and-forth attacks, and at hiding instructions inside images. And human testers keep catching things it misses. “I think human expertise will still be very important,” said Jessica Ji, an AI security analyst at Georgetown’s CSET.

The bigger idea is a flywheel: use today’s models to harden tomorrow’s. OpenAI already does this to make its AI smarter. Now it wants safety to scale just as fast. A full paper is due later this week.