Saturday’s technology news is led by a supply-chain attack that used a backup service to break into Microsoft 365 accounts, heavy demand for Apple’s first foldable iPhone, carrier pricing for iPhone Handoff, new App Store rules against nudify apps, Replit’s switch to OpenAI, Nvidia’s denial over the RTX 5090, Qualcomm’s rebuttal of overheating claims and ASUS’s first Intel Arc card.
Hackers use a ConnectWise backup credential to break into Microsoft 365 tenants
ConnectWise has cut off customer access to its Cloud Backup service, formerly known as SkyKick Cloud Backup, after attackers obtained a credential belonging to the application and used it to sign in to customers’ Microsoft 365 environments. The service is designed for managed service providers, the firms that run IT systems on behalf of other businesses, and it backs up and restores data such as emails and files stored in Microsoft 365 and Google Workspace. To do that job, the application is granted wide access in every tenant where it is installed, covering mailboxes, OneDrive, SharePoint, Teams and directory data. The security company Huntress said it had confirmed unauthorised sign-ins in 173 partner and customer tenants using the stolen vendor credential, and warned that the credential works against every organisation that has granted the application consent, making the potential exposure considerably wider.
Administrators began reporting suspicious activity on Friday, describing unauthorised access to client data that they linked to a possible breach at SkyKick or ConnectWise. Microsoft later confirmed malicious activity involving the backup service and said it had detected unauthorised authentication attempts from a single internet address on 15 October. Its threat intelligence team described the incident as an active supply-chain attack and assessed that the intruder had probably abused permissions the application already held, including broad rights to read and write directory information. Huntress also linked the same address to scripted sign-in attempts on 13 and 14 October. One administrator said on an online forum that on 15 October they had received an alert that a SkyKick service principal had given itself the Exchange Administrator role, even though their organisation had not used the app for years.
Because the attacker signed in as the backup application itself, using its secret, rather than as a person, protections such as multi-factor authentication offered no defence. The cybersecurity platform Todyl said the sign-ins were carried out by an automated script and meant the intruder could, in principle, read and alter email and attachments, OneDrive and SharePoint files, and Teams conversations. In the nine tenants it analysed, Todyl found only sign-in activity and no evidence that accounts had been created, passwords changed or a lasting foothold established. Huntress, however, saw ConnectWise management applications grant themselves directory roles in nearly 50 tenants, and in one case the attacker gained a Global Administrator role and added credentials to other tools within about five minutes. Microsoft also observed a backdoor being used for reconnaissance after a successful sign-in.
ConnectWise said in a security advisory that it had identified unauthorised access to application identities within the Cloud Backup environment and had shut down application access for partners and clients while it investigates and carries out remediation. The company said it had no indication so far that backup data held for its partners had been compromised, and that the existing backup archive was unaffected and would become accessible again once services are restored. It announced the interruption early on 17 October, coordinated universal time, but gave no timetable for reconnecting the service. Researchers at Huntress were more blunt about the scale of the risk, with one of them warning that the application’s credential was now in hostile hands. He said it had been used on 15 October to obtain Microsoft Graph tokens in customer tenants from the attacker’s address, underlining that the threat concerns live data in customers’ own Microsoft 365 environments rather than only the stored backup copies.
Researchers have urged administrators to act quickly, warning that the vendor’s shutdown does not remove any credentials already planted in customer tenants. ConnectWise recommends checking service principal sign-ins in Microsoft Entra, filtering by the Cloud Backup application identifier and looking for unexpected source addresses, while Microsoft advises reviewing changes to application credentials, OAuth permission grants and related activity. Huntress suggests disabling the affected ConnectWise applications and reviewing the secrets and certificates of all applications for unauthorised additions made since 13 October, noting that older tenants may list them under the SkyKick name. Todyl cautioned that disabling an application stops new access tokens being issued but does not invalidate existing ones until they expire, and that revoking user sessions does not end access gained through an application. The case highlights the risks of granting third-party tools broad standing permissions.
Apple’s iPhone Duo sells out early as delivery dates slip into 2027
Demand for Apple’s first foldable phone has pushed delivery estimates in the United States as far out as January 2027 within a day of orders opening. Pre-orders for the iPhone Duo began on Friday, and the launch-day pickup stock sold out quickly. Apple described the demand as nothing short of extraordinary and said it was working hard to get the device into customers’ hands as quickly as possible, adding that it would continue to accept orders online. Customers who ordered early have reported a wide range of outcomes, from being able to collect a phone from an Apple Store as soon as 23 October to delivery dates in November or December. Those ordering now are being offered January delivery windows on the online store, along with a pickup option that asks them to check back on 24 October.
The iPhone Duo is the most expensive iPhone Apple has ever sold. Prices start at $1,999 and reach $3,199 for the version with two terabytes of storage, creating a new tier above the Pro and Pro Max models. It also represents the most significant redesign of the iPhone’s hardware for several years. Apple’s chief executive, John Ternus, unveiled the device at an event at the company’s headquarters in Cupertino in September. Some analysts had expected the foldable to start at between $2,400 and $2,500, so the actual starting price was lower than many industry watchers had predicted. Apple has said customers will be able to buy the phone in stores or online, but it has not confirmed whether its shops or other authorised retailers will hold stock. Best Buy’s website was showing November delivery estimates.
Analysts say Apple is not betting that foldable phones will suddenly become a mass-market product. Instead, the bet is that a relatively small group of its highest-spending customers will become considerably more valuable. According to the research firm IDC, foldables are expected to make up just over two per cent of smartphone shipments worldwide this year but nearly seven per cent of industry revenue. By 2030, IDC expects the category to account for ten per cent of smartphone revenue while still representing only about three per cent of devices shipped. Analysts also expect the Duo to lift Apple’s average selling prices by encouraging customers to trade up, attracting people who currently use Android phones and creating a halo effect for the rest of the iPhone range, which could support the company’s revenue in the coming quarters.
An IDC research director told reporters that the Duo was fast becoming an aspirational foldable, comparing its appeal with Apple’s wider standing as an aspirational smartphone brand. She argued that the technology itself is not a breakthrough, since foldable Android phones have been available for years, but that what is new is Apple bringing the form factor to the very large number of people already invested in its ecosystem. Apple’s interest in folding designs goes back a long way: Ternus was one of six engineers at the company who applied for a patent covering foldable electronic devices as early as 2011. An analyst at Moor Insights who attended the launch event noted that many people had expected a considerably higher starting price, which suggests Apple has priced the device to reach beyond its most dedicated fans.
Strong early demand is not in itself a reliable guide to a device’s long-term success, and the real test will come in the months after launch, once the first wave of enthusiasts has been served. Pre-order numbers can be shaped by limited initial supply as much as by demand, and Apple has not disclosed how many units it has available. Nonetheless, the rush suggests that Apple’s customers are willing to pay a substantial premium for a new design. For anyone hoping to receive the phone in time for the end-of-year holidays, the lengthening delivery estimates are unwelcome news. Accessory makers have moved quickly to take advantage of the launch, releasing a range of cases and screen protectors for the new phone, and Apple’s own silicone case for the device has drawn attention because it has to be attached with adhesive.
Verizon will charge $10 a month for iPhone Handoff as EE sets UK pricing
Verizon has said it will support iPhone Handoff, a new feature in iOS 27 that lets one mobile number be used on two iPhones, but customers will have to pay $10 a month for it. The American carrier said in an announcement that the feature was coming soon to its network, in the same statement in which it revealed it would sell the iPhone Duo. Its fee is double the $5 a month charged by T-Mobile, which launched the feature in the United States, while Deutsche Telekom offers it in Germany. In the United Kingdom, EE has become the first network to confirm pricing. It will include the feature at no extra cost on its Full Works Plus plan, which also includes its Fast Lane service, and will charge £5 a month on a rolling contract for customers on its other plans.
Apple announced iPhone Handoff at its September keynote, saying it would rely on support from mobile carriers and naming T-Mobile, Verizon, EE and Deutsche Telekom as the first partners. The feature allows the same eSIM number to be used on two devices, provided that both are running iOS 27 and share the same Apple ID. The phones cannot use the number at the same time, however: only one device is active at any given moment, and users can switch between them whenever they wish. When they do, calls, messages and the Find My location service automatically move to the iPhone currently in use, and iCloud backups are synchronised across both devices. Apple has said the feature will be available by the end of the year, although several carriers still describe it as coming soon.
Setting up the feature is designed to be simple. On the newer phone, users go to the cellular section of the settings app, choose to set up cellular service or add an eSIM, and then select the option to transfer from an iPhone. Switching between the two handsets requires no further steps beyond unlocking the device the user wants to start using, at which point a message appears at the top of the screen confirming that the number has moved to that iPhone. Apple advises customers to protect both phones with passcodes so that they stay in control of which one is actively using their number. The design means that a second handset can sit unused in a drawer until it is needed, without the owner having to visit a shop or call the carrier.
EE has promoted the feature as a way for customers to choose the iPhone that suits the moment without worrying about missing a call, message or notification. It suggested that people might use a lighter phone when going to a festival, running or hiking, and a different one at other times, with contacts, calls and messages following them from device to device. The company said the flexibility on offer was not available anywhere else in the UK. The announcement formed part of EE’s pre-order campaign for the iPhone Duo, Apple’s new foldable phone, and Verizon likewise paired its Handoff news with confirmation that it would sell the device, linking the two launches in the minds of customers. Both companies are evidently hoping that the new feature will help persuade owners of an existing iPhone to add a second, more expensive handset to their account rather than replacing the phone they already have.
Carrier fees for features that require network support are not new, but the differences in pricing for iPhone Handoff have drawn attention because the feature itself is built by Apple and works in broadly the same way everywhere. Customers in the United States will pay twice as much on Verizon as on T-Mobile, while some EE customers in the UK will get it included at no extra cost on their plan. Other carriers around the world have yet to say whether, or on what terms, they will offer the feature. For people who own two iPhones, the monthly charge will be weighed against the convenience of not having to move a SIM card or maintain separate numbers for different phones and occasions. Whether rival networks follow EE in bundling the feature into premium plans, or follow Verizon in charging a higher standalone fee, will become clearer as more carriers announce their plans before the end of the year.
Apple tightens App Store rules against AI nudify apps
Apple has updated its App Store rules to ban more explicitly so-called nudify apps, which use artificial intelligence to turn ordinary images of people into sexually explicit pictures without their consent. The company has long prohibited pornographic material in its store, and apps built specifically to generate sexually explicit imagery have not been allowed for some time. The new rules go further by requiring developers to police how their apps are used. Developers whose apps host user-generated content must now proactively prevent and remove nonconsensual intimate images, including those generated with AI tools. They must stop users from uploading such material, take down anything that slips through their filters, give users a way to report offensive content and allow them to block abusive accounts. Apps that fail to put adequate safeguards in place risk removal.
The rules also deal with so-called bait-and-switch tactics, in which an app presents itself one way during Apple’s review process and then adds or changes features after approval. Apple has clarified that if it finds an app misrepresented its purpose during review and in fact has nudification capabilities, it will take the app down and expel its creator from the Apple Developer Program. Developers who do not remedy violations after being notified of them may also be removed. In addition, the updated rules spell out more clearly that apps that enable human trafficking or child exploitation will be removed from the App Store and blocked from alternative distribution methods, which Apple is now required to allow in some regions. Apple announced the changes on Friday as part of a wider update to its guidelines.
The changes follow significant criticism of Apple earlier this year, after users of the social network X used its built-in AI chatbot, Grok, to create nonconsensual sexualised deepfakes, including images of minors. Apple did not remove X or Grok from the App Store, although it told United States senators that it had threatened to do so. At Apple’s developer conference in June, protesters called on the company to remove Grok, X and all nudify technology from its store. In July, the San Francisco city attorney ordered Apple to remove 13 nudify apps, and the company said it had done so. Campaigners have argued for some time that app stores play a central role in distributing such tools and that the companies running them should take greater responsibility for the harm they cause.
Nonconsensual intimate imagery, sometimes described as image-based sexual abuse, has become a growing concern as AI image generators have become more capable and easier to use. Tools that can produce realistic fake nude pictures from a single ordinary photograph have been used to harass and humiliate women and girls, including schoolchildren, and lawmakers in several countries have moved to criminalise the creation or sharing of such images. Technology companies have faced pressure from regulators, campaigners and victims to stop hosting or promoting the apps that make them. Apple’s rule changes place much of the responsibility on developers to stop abuse within their own apps, rather than relying solely on the company’s review process to catch problems before an app is approved, which critics have argued was never sufficient on its own.
The new rules may affect a wide range of apps beyond those designed specifically for creating explicit images. Any app that lets users upload or share content, from social networks to messaging and photo-editing tools, will now need systems that can detect and remove nonconsensual intimate imagery. Smaller developers may find it costly to build such filters, though many larger platforms already have moderation teams and automated tools in place. How strictly Apple enforces the requirements, and whether it applies them to large and influential apps as firmly as to smaller ones, is likely to be watched closely by critics who said the company had been slow to act earlier this year. Campaigners are also likely to press other app store operators to adopt comparable requirements, so that developers cannot simply move abusive tools to rival platforms, and so that victims receive the same protection whichever type of phone the people targeting them happen to use.
Replit makes OpenAI its default AI provider instead of Anthropic
Replit, the AI coding platform popular with start-up founders and developers, has made OpenAI its default model provider in place of Anthropic, its chief executive said. Speaking on stage at the TechCrunch Disrupt conference, co-founder and chief executive Amjad Masad initially said the company was no longer working with Anthropic, citing a combination of reasons including cost. He said he believed ChatGPT was now better and that Replit used an internal benchmark, as well as a public one, to assess performance, adding that OpenAI’s models consistently performed better at building applications from scratch. He described the decision as one based on cost, partnership and performance, and said that for all those reasons Replit felt OpenAI was a better partner for now. Replit had announced a partnership with OpenAI in August, but had given no sign then of dropping Anthropic.
After the conference, Masad clarified that Anthropic had not been entirely cut out. He said customers could still choose Anthropic’s Claude models on the platform, which sells a cloud-based service that lets users build and deploy apps directly in their web browser, but that OpenAI’s ChatGPT was now the default. He said Replit remained model agnostic and that OpenAI was its preferred partner for powering Replit Agent, its main AI coding tool, based on performance, economics and the strength of the partnership. The change could still significantly reduce Replit’s spending with Anthropic, because most customers stick with the default model rather than selecting one themselves. People familiar with the relationship disputed Masad’s original remarks, and one, who asked not to be named, said the partnership between the two companies was still in place.
Before its deal with OpenAI, Replit relied on Anthropic’s Claude models to run its platform, and that relationship was said to have helped the company grow its revenue substantially. Masad said on stage that Replit had until recently been one of Anthropic’s five largest customers. The company’s website, blog posts and social media accounts promoted Anthropic’s products as recently as a few weeks ago: a blog post in late September named two of Anthropic’s Claude models as running in Replit Agent’s production environment, and around the same time a Replit product employee posted on social media praising Anthropic’s flagship model. The switch is therefore a notable shift for a company that had been closely associated with Anthropic’s technology and that had served as one of its most visible showcases among developers building with AI.
Masad also shared details about Replit’s growth, saying that business customers were providing a larger share of revenue. He said the company was not yet disclosing its annual recurring revenue, but that enterprise contracts had made up around two per cent of revenue last year and about 15 per cent this year, and that more than half of its revenue now came from business-to-business sales. The shift towards corporate customers reflects a wider trend in AI coding tools, which began as products aimed mainly at individual developers and hobbyists but are increasingly being sold to companies that want to speed up software development across their teams. Competition in the market has intensified as large AI developers have launched their own coding assistants and agents, putting pressure on platforms that build on their models.
The episode illustrates the shifting relationships between AI model developers and the companies that build products on top of their technology. Platforms such as Replit depend heavily on access to the most capable models, and the cost of using them can be one of their largest expenses, which gives them strong incentives to switch providers when prices or performance change. For model developers, large platform customers can account for a significant share of revenue, so losing default status on a popular service can have a noticeable financial effect. Many companies now design their products so that users can switch between models from different providers, which makes it easier to change suppliers but also means that loyalty to any single developer can be short-lived, as this change of default demonstrates.
Nvidia denies reports that the GeForce RTX 5090 has been discontinued
Nvidia has denied reports that it has stopped making the GeForce RTX 5090, its flagship consumer graphics card, while acknowledging that supply is tight. Jeff Fisher, a senior vice president at the company, told Taiwanese media at the launch party for its RTX Spark platform in Taipei that the rumours were untrue. According to his reported remarks, he said that the supply chain was indeed somewhat constrained but that Nvidia was doing everything it could to bring products to market, and that although there were other variants of the chip behind the RTX 5090, the card had definitely not been discontinued and the company would continue to ship it. He described the RTX 5090 as an extremely important core product. He also acknowledged that supply constraints were severe, which helps to explain why the card has become so difficult to buy.
The speculation began last week when leakers claimed that a factory notice showed Nvidia would no longer manufacture any RTX 5090 models, including the D variants made for the Chinese market. According to that claim, the RTX 5080 with 24 gigabytes of memory would become the most powerful GeForce card, while the 16-gigabyte version of the RTX 5080 would be discontinued. A second leaker said the GB202 chip used in the RTX 5090 would in future go only into data centre and professional cards, with the professional RTX PRO range taking priority. The RTX 5090 also appeared to vanish from Nvidia’s own online store and from the shelves of the American retailer Micro Center, which added weight to the reports. Nvidia had not commented until now.
Fisher’s comments leave some questions unanswered. He did not explain why the card disappeared from Nvidia’s own store and from retailers’ shelves, or whether production had been paused and then resumed, and it is not clear what he meant by other variants, although the same chip is also used in professional and data centre products. Some observers have asked whether the company changed course in response to the backlash, or whether there was a genuine disruption in the supply chain. He also stressed the importance of Nvidia’s gaming and GeForce business, saying the company would not be where it is today without it. That remark is notable because Nvidia now earns the great majority of its revenue from data centre products used to train and run artificial intelligence systems, leading some observers to question how much priority consumer graphics cards still receive within the company.
For buyers, the immediate problem is price and availability. The RTX 5090 has a recommended retail price of $1,999, but it is very hard to find at anything close to that figure. Prices have passed $5,000 in most regions, around two and a half times the official figure, and the card has disappeared from some major retailers altogether. Given the supply constraints that Fisher himself acknowledged, analysts expect the card to remain scarce and expensive for some time. A planned refresh of the RTX 50 series, rumoured to include a 24-gigabyte RTX 5080 Super, has reportedly been delayed by memory shortages, and the next generation of GeForce cards, the Rubin-based RTX 60 series, is not expected until late 2027 at the earliest, and possibly not until 2028, according to rumours.
The wider memory shortage has affected much of the technology industry, as manufacturers prioritise the high-bandwidth memory used in AI data centres over the chips found in consumer devices. Valve has said its products are limited by memory capacity, and Xbox has warned that memory costs will affect the pricing and availability of its next-generation hardware, codenamed Project Helix. There have also been reports that Sony could delay its next PlayStation until 2028 or 2029 because of the crisis. The combination of shortages and strong AI demand has pushed up prices for graphics cards, memory and storage, making it more expensive for consumers to build or upgrade gaming PCs. PC makers have warned that graphics card prices could rise significantly further, adding to the frustration of gamers who have faced shortages repeatedly over recent years.
Qualcomm rejects reports that its next flagship chip overheats
Qualcomm has dismissed reports that its next flagship smartphone processor, the Snapdragon 8 Elite Extreme Gen 6, suffers from overheating problems that could lead Samsung to drop it from the Galaxy S27 Ultra. A company spokesperson said that recent reporting about thermal issues related to the chip was inaccurate. The spokesperson added that Qualcomm was proud of the innovation and strong momentum behind its next-generation flagship platform and continued to work closely with phone makers around the world to deliver premium user experiences. Don McGuire, Qualcomm’s executive vice president and chief marketing officer, went further on social media, saying there were no systemic thermal issues with the chip and that the reports were completely inaccurate. He said phone makers consistently fine-tune their platforms as they launch products, balancing performance and power use for different purposes.
The reports originated in South Korea, where it was claimed that Samsung’s smartphone division had recently carried out internal performance verification of the chip for the Galaxy S27 Ultra and found that it exceeded its temperature limits during sustained high-load tasks, leading to performance throttling. Samsung reportedly raised its concerns with Qualcomm last month and set a deadline of mid-October for the problems to be resolved before it finalised its choice of processor. If Qualcomm failed to meet Samsung’s standards for performance, stability and heat, according to the claims, Samsung was prepared to use its own Exynos 2700 chip in the phone instead. The claims spread quickly across technology sites, because Samsung’s choice of processor for its most expensive phone is closely watched by both companies’ investors and by buyers.
For the last few generations, every Galaxy S Ultra phone has run on a Snapdragon processor, and the Galaxy S27 Ultra had been expected to follow suit. The Exynos 2700 is designed by Samsung’s System LSI division and built on the second generation of Samsung Foundry’s two-nanometre manufacturing process, with a ten-core processor. According to the reports, Samsung’s internal testing found its performance was on a par with, or slightly better than, Qualcomm’s chip. Choosing Exynos would reduce Samsung’s costs and increase the use of its own chip factories, strengthening that part of its business. The Snapdragon 8 Elite Extreme Gen 6 itself includes two cores running at five gigahertz and promises large improvements in processor and graphics performance over its predecessor, which was already regarded as one of the fastest mobile chips.
The Galaxy S27 Ultra is scheduled to enter mass production later this year and to launch in early 2027 alongside the Galaxy S27, S27+ and S27 Pro. According to the reports, the three smaller models could use the Exynos 2700 in most countries regardless of the Ultra’s processor. The question of which chip powers the Ultra matters to buyers because Samsung has in the past used different processors in different regions, and some customers have complained that Exynos versions performed worse or ran hotter than Snapdragon models. Samsung’s recent phones have used Snapdragon chips worldwide in the Ultra, partly in response to those concerns, which is why a switch back to Exynos would be seen as a significant change of direction for the company, and one that buyers would watch closely.
Heat management has become a growing challenge for smartphone makers as processors become more powerful and are asked to handle demanding tasks such as gaming and artificial intelligence features that run on the device itself. Some phone makers have even begun fitting cooling fans to gaming phones and tablets. Chips that run too hot reduce their speed to protect themselves, which can make phones feel slower during sustained use and drain batteries more quickly. Manufacturers therefore spend considerable effort tuning their devices, adjusting cooling systems and software to balance performance against temperature, a process McGuire described as normal. Whether Samsung ultimately chooses Snapdragon or Exynos for the Galaxy S27 Ultra is likely to become clear only nearer to the phone’s launch early next year, when the specifications are officially confirmed.
ASUS releases its first Intel Arc graphics card, aimed at AI workloads
ASUS has released its first graphics card based on Intel’s Arc technology, returning to Intel graphics after several years on the sidelines. The card, called the ASUS UGen Arc Pro B70, uses Intel’s largest Battlemage chip, known as BMG-G31, in its full configuration, with 32 gigabytes of memory. It is aimed mainly at artificial intelligence and workstation tasks rather than gaming, and ASUS places heavy emphasis on its AI performance on the product page. The card offers 367 trillion operations per second of AI computing power and supports multi-GPU set-ups, allowing four cards to be combined for up to 128 gigabytes of video memory. That capacity is valuable for running large AI models locally, which require large amounts of memory, and is one of the main reasons professional users have shown interest in the card.
The ASUS card takes up two expansion slots and uses a blower-style cooler, which pushes hot air out of the back of a computer case and is often preferred in workstations that hold several cards close together. It has four display outputs, one HDMI and three DisplayPort, and a maximum board power of 310 watts. The launch marks a notable change for ASUS. During the early days of Intel’s Xe graphics, ASUS was one of the first companies to release a card based on Intel’s DG1 chip, with passive cooling, but in the two generations of Arc products since then it, like other large board makers such as MSI and Gigabyte, did not release any Intel cards. Until now, ASRock and Acer have been the biggest names offering Intel Arc cards.
Smaller brands such as Sparkle, along with several new companies that make only Intel Arc products, have filled out the rest of the market. Intel’s dedicated graphics cards have struggled to win a large share against Nvidia and AMD, which dominate both gaming and professional graphics, and the absence of the biggest board partners has limited their reach in shops and in pre-built systems. Strong demand for AI hardware appears to have changed the calculation. The Arc Pro B70 has been seen as offering good value compared with rival professional cards, particularly in terms of memory capacity for its price, at a time when shortages have pushed up the cost of competing products and made high-end Nvidia cards very hard to obtain at their recommended prices.
The release comes as the wider graphics card market faces significant supply pressure. A global shortage of memory, driven by the demand for chips in AI data centres, has made graphics cards more expensive and harder to find, and Nvidia has acknowledged that supply of its flagship consumer card is constrained. For companies and researchers who want to run AI models on their own machines rather than in the cloud, cards with large amounts of memory are particularly sought after. By offering an Intel-based alternative, ASUS gives such buyers another option from a widely recognised brand, which may help Intel’s professional graphics products reach a wider audience among businesses that prefer established suppliers with strong support networks and distribution through major resellers. It also gives Intel a higher-profile partner at a time when buyers are actively looking for alternatives to scarce and expensive products from its larger rivals.
For Intel, winning the support of a major board partner such as ASUS is a welcome boost for its graphics business, which has faced difficulties since the first Arc cards launched. The company has continued to develop the product line despite wider financial pressures, and future generations are already in development. Whether other large manufacturers follow ASUS’s lead may depend on how well the Arc Pro B70 sells and on whether demand for AI hardware remains strong. The launch also underlines how AI has reshaped priorities across the hardware industry, with professional and data centre products increasingly receiving attention that once went mainly to consumer gaming hardware and enthusiast builders. For now, the ASUS card is aimed at professionals, but its arrival shows that even the largest board makers are willing to revisit decisions about which chip suppliers to support when market conditions change, and that Intel’s graphics products are gaining credibility among the industry’s established names.