GoKawiil
Tech News
SK Hynix CEO Kwak Noh-jung said he sees no signs of oversupply in the memory chip market and expects the current shortage to persist until 2030, driven largely by AI-related demand for advanced memory like HBM. He made the comments as the company broke ground on a new chip packaging facility in Indiana, part of its broader US expansion tied to growing customer needs.
Security researchers warn that enterprises deploying AI agents are prioritizing gateway controls before establishing the identity and attribution systems those gateways depend on. A real-world example cited is a LiteLLM flaw added to CISA's Known Exploited Vulnerabilities catalog in June, which let attackers run commands on the host without credentials, one of seven vulnerabilities found in that gateway in a single month. Analysts argue gateways should be the fifth layer of defense, not the first, since without knowing which agent is acting and why, enforcement systems can't tell legitimate actions from technically permitted but inappropriate ones.
Google announced new Play Store requirements that will force Android app developers to operate within strict memory thresholds beginning in 2027, aiming to stop individual apps from hogging RAM and slowing down entire devices. Apps that fail to comply risk being throttled or removed from the store, and Google explicitly linked the policy to the ongoing memory supply shortage driven by AI demand.
In an Inside Java Podcast episode recorded at JavaOne 2026, Oracle Java Architect Ron Pressler discussed with host Nicolai Parlog why Java's often-criticized memory consumption is actually intentional. He argued that Java trades extra memory for reduced CPU cycles, a tradeoff enabled by its moving garbage collectors, which many leaner platforms cannot replicate.
XCENA partnered with Samsung to build the MX1, a CXL 3.2 device that can expand host memory by up to 2TB of DDR5 while also connecting SSDs and performing on-board computation. The chip, made on Samsung's 4nm process, packs 3072 small RISC-V cores organized into 24 independently schedulable subsystems, alongside two Arm Cortex A53 cores for control, and draws about 40W on the chip and 90W for the full board with DIMMs.
The GrapheneOS development team says it could not finish porting its security-focused Android build to Google's new Pixel 11 phones because the devices lack support for Arm's memory tagging extension (MTE), a hardware-level security feature. Developers say they built a partial port after a week of effort but had to halt the project, calling the omission an apparent cost-cutting move by Google. As a result, the team may abandon future support for Pixel devices and focus instead on Motorola hardware.
The GrapheneOS project has said that Google's upcoming Pixel 11 will no longer support hardware Memory Tagging Extension (MTE), a security feature present in recent Pixel devices. The announcement came via a social media post rather than a detailed report, so specifics on why the feature was removed remain unclear.
Japanese researcher @dydt_Nao successfully demonstrated a dinner-plate-sized USB flash drive built from magnetic core memory, transferring both a text file and a small monochrome bitmap image between two PCs. The device, which uses a Raspberry Pi Pico as its controller, stores only 128 bytes of data and displays activity through a 32x32 LED array that lights up as each magnetic core is accessed. The creator plans to showcase the device at Maker Faire Tokyo 2026 in early September.
Samsung presented its Processing-in-Memory approach for LPDDR5X-9600 chips, embedding a MAC computation block into each of the 16 DRAM banks while keeping compatibility with standard memory controllers. By computing directly within each bank, the design taps into 614 GB/s of internal bandwidth versus the 76.8 GB/s ceiling of conventional two-bank DRAM access. Each PIM block includes an instruction register file, a source register file for activation vectors, and a scale register, with model weights fetched directly from the attached DRAM bank.
A security researcher describes building a new memory architecture for LLM agents used in vulnerability research, after noticing that standard retrieval-based memory let models keep reasoning from facts that had since been disproven. Instead of just storing and retrieving past observations, the system tracks logical dependencies between facts, so that when one assumption is invalidated, everything derived from it can be automatically flagged or revised. The approach effectively borrows ideas from program analysis, treating an investigation's knowledge state like a dependency graph rather than a flat log of conversation history.
An eight-year study published in Neurology found that adults consuming roughly the amount of artificial sweetener contained in one daily diet soda experienced memory and mental sharpness decline 62 percent faster than those consuming far smaller amounts. The research tracked participants over time to compare cognitive changes based on sweetener consumption levels.
A preliminary survey found that 80% of Micron Taiwan employees support a possible strike, as the labor union demands the company replace its current bonus structure with a profit-sharing model similar to those at SK hynix and Samsung. Mandatory mediation talks are expected to start by mid-September, as required under Taiwan law before any formal strike vote can occur. The push comes after Micron posted $41.46 billion in revenue and $28.24 billion in net income, fueled largely by AI-driven demand for high-bandwidth memory chips.