Hardware News

The emergence of the new Siri will trigger an update of Apple devices for the smart home Apple's smart home products, such as the HomePod mini and Apple TV, have been among the most discussed over the past couple of years. The reason for this is that the current versions of devices in this category were released 4-6 years ago. The update was hindered by issues with the development of the AI assistant Siri. Apple managed to resolve these issues, allowing the company to continue working on new smart home devices. According to sources, Apple will release updated HomePod mini and Apple TV 4K this fall. Developers will replace the S5 chip used in the current version of the HomePod mini with the S9 or a more powerful one. New color options for the casing may also appear. The device will receive a new N1 chip responsible for Wi-Fi wireless communication, which debuted in the iPhone last year. From a hardware perspective, the improvements are quite modest, but importantly, they will ensure support for the updated Siri on the HomePod mini, making the device significantly more useful. As for the Apple TV 4K, the updated set-top box will replace the A15 Bionic chip with the A17 Pro. This doesn't seem too impressive since the A17 Pro chip is already three years old, but apparently, it is quite sufficient for the set-top box. This update will also enable support for the Siri AI assistant and some Apple Intelligence AI features. A redesign of the remote control is also expected. Currently, the HomePod mini can be purchased for $129, and the price of the Apple TV 4K starts at $199. These are the prices of the devices after the retail price increase that Apple implemented in June. The cost of the next-generation models is still unknown. The company has been developing a central component for the smart home for quite some time. Such a hub may come to market in two versions: a model in the form of a screen mounted on the wall using a magnetic system, and a version with a screen attached to a base with a speaker, similar to the Google Nest Hub. This product is expected to debut "between October and early next year". This means it could be presented simultaneously with the HomePod mini and Apple TV 4K. The hub will feature a 7-inch square display, and its software foundation will be the new homeOS. The device will allow for FaceTime video calls, interaction with the Siri AI assistant, control of smart home devices, and more. As for the hardware, the new product will be equipped with an A18 chip. In the near future, a new version of the full-sized HomePod is also expected. The device should receive a more modern microprocessor to ensure support for the Siri AI assistant. Exact launch dates have not been announced. In the more distant future, Apple plans to release more smart home devices. In early 2027, a security camera and a video doorbell for the smart home are expected to appear. Thus, instead of just developing software for smart home devices, Apple will produce them itself. If you value what we do, you can support our projects on Boosty. With love, NSWTL.

Nvidia accelerates preparations for the launch of chips with Feynman architecture, which will require TSMC's A16 process technology In the second half of 2028, Nvidia is expected to bring Feynman architecture computing accelerators to the market, and Taiwanese sources report that the preparation process for this debut is accelerating, placing increased demands on the company's contractors in terms of readiness for new technologies. In addition to TSMC's A16 process technology, this involves new chip packaging methods. These include three-dimensional SoIC layout and the co-packaging of optical connections (CPO) method, as explained by DigiTimes. For Taiwan's TSMC, this means it will have to accelerate the construction of the AP7 and AP8 factories, which will process chips using the SoIC method. Initially, it was expected that they would be able to process 20,000 silicon wafers per month by the end of 2026, but now the plan has been revised to accelerate, and by the end of 2027, this number will increase to 50,000 wafers per month. The installation of engineering networks, construction of "clean rooms," and installation of equipment at these TSMC facilities must be carried out ahead of the original schedule. Feynman accelerators, in addition to using the A16 process technology, imply the use of three-dimensional SoIC integration technology, the application of customized HBM memory, and the co-packaging (CPO) of optical connections. Nvidia's needs have long determined the pace of technological development for TSMC and its closest partners. At the same time, the three-dimensional SoIC packaging technology will also be in demand by other TSMC customers, so accelerating expansion in this area will benefit the company to some extent. In the case of the 2-nm technology, everything is going smoothly for TSMC, and A20 Pro processors for Apple are being manufactured in sufficient quantities and with an adequate defect level, but there are delays with DRAM memory integration due to its shortage. For this reason, TSMC has accumulated about $1 billion worth of A20 Pro processors awaiting memory chip installation. The packaging technology was specifically adapted to Apple's needs, and it is compared to the mobile version of CoWoS. For Microsoft, TSMC plans to produce more than 300,000 Maia 300 chips by the end of next year, indicating growing demand for specialized AI chips from cloud giants. From the second half of this year, TSMC is also beginning to expand its chip packaging capabilities using the CPO method, which allows for the integration of optical connections with data transfer speeds of up to 400 TB/s. If you value what we do, you can support our projects on Boosty. With love, NSWTL.

US Authorities Warn Allies Against Collaborating with Chinese AI Developers The current US administration's confidence in the need to maintain technological leadership on the global stage is leading to initiatives that further polarize the market and geographically isolate it. Now, US authorities are advising their allies to refrain from collaborating with China in the fields of AI, semiconductor components, and critical minerals. According to Reuters, citing its own sources, the US government is preparing to send an official letter related to the "Statement on AI Opportunities" signed in June, which will be sent to 35 recipients. These recipients will represent a proportional number of countries ready to commit to the US not to participate in initiatives promoted by China in the relevant areas of activity. This is not the first coalition the current US administration is trying to form in its confrontation with China. Last year, the Pax Silica alliance was established to ensure stable supply chains for the development of the American AI industry, semiconductor industry, and critical minerals. More than two dozen countries joined this alliance, with Kazakhstan managing to be both among the allies of neighboring China, which is unlikely to please US authorities in the long term. In addition to Pax Silica participants, the new US initiative may strengthen the coalition with new countries willing to move in the same direction in terms of AI development. China is also observing this market polarization and is taking proportional measures. In July, Xi Jinping founded the World AI Cooperation Organization, which will promote Chinese open-weight AI models beyond the country's borders. Kazakhstan is a participant in both initiatives, and this duality in political positioning has begun to irritate American officials. In the upcoming letter, US allies are explained that it is not possible to sit on two chairs in this field simultaneously, and Pax Silica members should not support competing alliances aimed against US interests. This appeal currently exists as a draft document, and it is unclear when it will be sent to recipients. The Chinese embassy in the US condemned the politicization of the technological sphere and trade, calling such practices destructive. However, it cannot be claimed that China's policy in this area is entirely open. It is believed that Chinese authorities intend to restrict foreigners' access to advanced locally developed AI models. Kazakhstan is of interest to American politicians due to its reserves of rare earth minerals, which are needed for the development of the US semiconductor and defense industries. The former country could become an alternative to Chinese supplies, which are limited due to political reasons. If you value what we do, you can support our projects on Boosty. With love, NSWTL.

Taiwanese Engineer Accused of Industrial Espionage Slipped Up by Leaving Hidden Camera Transmitters On According to Tom’s Hardware, citing Freedom Times, a former Nanya Technology engineer is accused of illegally copying 32 files containing information about the DRAM memory chip manufacturing process. He slipped up by bringing a camera hidden in a snack bag to work, forgetting to turn off the Wi-Fi and Bluetooth transmitters. According to the investigation, Yeh Yen-wei, who worked as a senior engineer in Nanya's advanced process development division since 2022, began negotiations with a Chinese recruitment agency last year to find a new job in China. By mid-February of this year, he had submitted his resignation from Nanya voluntarily, but was scheduled to leave the company only in mid-April. At the end of March and twice in April, he visited the secret part of Nanya's laboratories on weekends or after hours, hiding an Insta360 X4 Air camera in a snack bag to avoid arousing suspicion from security. Using his access rights at the time, the suspect opened documents related to Nanya's developments on the corporate network. He used the camera he brought to the site to take photos for saving the information. After leaving the company, he had the opportunity to use the saved information. However, he did not account for the fact that in the restricted access area, Nanya's security specialists monitor the activity of equipment emitting radio signals, and the presence of a Wi-Fi and Bluetooth signal source with the visitor was detected retrospectively by security, who then passed the information to Taiwanese law enforcement. He is now under investigation and faces up to 10 years in prison under local laws. The investigation believes the suspect intended to pass the Taiwanese company's technological secrets to his new employer in China. However, there is no evidence that the suspect managed to transfer valuable information to the Chinese side. If you value what we do, you can support our projects on Boosty. With love, NSWTL.

Sony explained how the new Enhanced PSSR upscaler improves graphics quality on the PS5 Pro At the SIGGRAPH 2026 conference, Daniel Craig, the lead software engineer at Sony Interactive Entertainment, detailed the updated Enhanced PSSR upscaler model for the PS5 Pro gaming console. One of the most significant architectural changes in the Enhanced PSSR upscaler is related to how the neural network's computational resources are utilized. According to Sony, in the original Color Predicting Network (CPN) neural network, about 30% of the parameters were used at a 540p resolution. In the new Kernel Predicting Network (KPN), as much as 90% of the parameters use a resolution of 540p and below, while in 4K, the neural network hardly uses any resources. The new KPN neural network also does not predict the final color directly. Instead, it generates blending weight coefficients, and the actual blending is performed using standard filtering. This avoids storing and blending HDR colors in an 8-bit format and reduces the load on the neural network when performing these operations, thereby increasing overall efficiency. Sony also notes improved temporal stability, sharper detail, reduced memory consumption, and GPU runtime. The company claims that image quality improves at any resolution, but the improvement is especially noticeable at lower resolutions. The company also reported that the neural network's training time has been reduced from approximately four GPU-months to less than one GPU-week. At the SIGGRAPH 2026 conference, Sony also confirmed that Enhanced PSSR replaces the version with which the PS5 Pro was launched on the market, rather than functioning as a separate mode. The new version was developed by Sony and AMD as part of the Project Amethyst research project, and its implementation is adapted to PlayStation hardware. Existing PS5 Pro games can use Enhanced PSSR through a system override without updating the game itself. Developers who integrate the new library directly will have full access to the improved algorithm. If you value what we do, you can support our projects on Boosty. With love, NSWTL.