South Korea-based semiconductor startup XCENA announced it has raised $135 million in a Series B funding round. Atinum Investment and IMM Investment co-led the investment, with participation from Corstone Asia, SBI Investment, and Mirae Asset Capital. Founded in 2022, the company has now raised a total of $185 million and reached a valuation of $570 million.

XCENA develops semiconductor technologies aimed at solving memory and data transfer bottlenecks in AI data centers. The company argues that constant data movement between CPUs, GPUs, and memory in AI systems creates significant cost and performance inefficiencies. To address this, XCENA is working on a memory-centric architecture that brings processing power closer to where data resides.

The startup's MX1 chip is designed to handle tasks such as data preprocessing, cache management, and data transfer directly on the memory module, reducing the workload on CPUs and GPUs. The chip focuses on inference processes for large language models and generative AI applications. XCENA aims to lower data center costs, improve energy efficiency, and enable AI infrastructure to operate with fewer servers.

XCENA was founded by Dohun Kim and Harry Juhyun Kim. The company plans to use the new funding to accelerate mass production of the MX1 chip, advance product development, and initiate commercial-scale deployment for AI data center customers. The Series B round brings the startup's total funding to $185 million since its inception in 2022.

The company's valuation now stands at $570 million, reflecting investor confidence in its memory-centric approach to AI hardware. XCENA's technology targets the growing demand for efficient AI inference in data centers, where power and cost constraints are becoming increasingly critical.

XCENA's MX1 chip is expected to enter mass production in the near term, with commercial deployments planned for AI data center clients. The company has not disclosed specific timelines or customer names but stated that the funding will support scaling operations and meeting market demand.

"This investment validates our vision of transforming AI data center architecture through memory-centric computing," said Dohun Kim, co-founder of XCENA, in a statement. The company aims to deliver tangible improvements in performance and energy efficiency for AI workloads.

NVIDIA took the stage at Computex to announce the RTX Spark, a new superchip built for Windows-based computers. The chip targets both laptops and compact desktop systems, aiming to accelerate artificial intelligence and graphics workloads. According to NVIDIA, the RTX Spark will power upcoming devices from major OEMs, including the Surface Laptop Ultra and Dell XPS 16.

The RTX Spark is positioned as a portable counterpart to the company's DGX Spark, a $3,999 AI mini desktop. While the DGX Spark runs on Linux and primarily serves large enterprises and AI developers, the RTX Spark is designed for Windows and aims to be more accessible to content creators and those new to AI development.

NVIDIA claims the RTX Spark delivers 1 petaflop of AI compute performance. The chip integrates 6,144 Blackwell RTX cores alongside 20 Mediatek Arm CPU cores. NVIDIA noted that this configuration resembles the RTX 5070 laptop GPU but with significantly lower power consumption.

The chip includes a neural processing unit (NPU) fast enough to meet Microsoft's Copilot+ requirements, which mandate 40 TOPS. For AI tasks, the RTX Spark leverages tensor cores within its Blackwell GPU architecture. The GPU can directly access a unified memory pool ranging from 16 GB to 128 GB, while the chip itself consumes between single-digit watts and 80 W.

NVIDIA CEO Jensen Huang described the RTX Spark as a complete reinvention of the PC. He stated that the chip transforms PCs into devices designed for AI agents. Huang emphasized that the RTX Spark represents a fundamental shift in how computers handle AI workloads.

NVIDIA collaborated with Microsoft over several years to design the RTX Spark. Pavan Davuluri, Microsoft's head of Windows and devices, confirmed that the company optimized Windows 11's workload profile scheduling for the new chip. Davuluri explained that whether users are checking emails or running a local agent for code debugging, the Windows scheduler on RTX Spark will ensure optimal performance and efficiency from the CPU.

The RTX Spark is expected to appear in systems from all major OEMs. Pricing and specific release dates for individual devices have not yet been announced. NVIDIA positioned the chip as a bridge between high-end AI development platforms and mainstream Windows computing.

NVIDIA's announcement at Computex underscores the company's push to embed AI capabilities into everyday computing devices. The RTX Spark aims to bring advanced AI processing to a broader audience without requiring specialized hardware or Linux-based systems.

Intel has officially introduced its next-generation Xeon 6+ processor family, code-named Clearwater Forest, designed for data centers and AI infrastructure. The new chips are built on the company's most advanced manufacturing process, the Intel 18A node, and deliver up to 288 cores per socket. The announcement comes as demand for computing power in AI and cloud services continues to surge.

The Xeon 6+ series represents a significant architectural leap from its predecessors. Intel 18A technology incorporates RibbonFET gate-all-around transistors and PowerVia backside power delivery, enabling higher performance and energy efficiency. The 288-core configuration targets massively parallel workloads such as large language model training, inference, and high-performance computing.

Intel positions the new processors as a direct response to the growing computational demands of modern data centers. The company emphasized that the Xeon 6+ will support advanced memory and I/O technologies, including DDR5 memory and PCIe 5.0 interfaces, to minimize bottlenecks in data-intensive applications. Specific clock speeds and thermal design power figures have not yet been disclosed.

The Clearwater Forest architecture introduces several microarchitectural improvements over the previous Granite Rapids and Sierra Forest designs. Intel claims the new cores deliver a substantial increase in instructions per clock, though exact performance metrics will be shared closer to launch. The processors also incorporate enhanced AI acceleration capabilities through integrated matrix engines.

Intel expects the Xeon 6+ to compete directly with AMD's EPYC processors and Arm-based server chips from companies like Ampere Computing. The company has not announced specific pricing or volume availability dates but indicated that sampling to key customers will begin in the first half of 2025. Production ramp is anticipated later that year.

Data center operators and cloud service providers are expected to be the primary customers for the Xeon 6+ series. Intel highlighted that the processors will be compatible with existing LGA 7529 socket platforms, easing upgrades for current Xeon users. The company also confirmed support for Intel's oneAPI software stack to simplify development across heterogeneous architectures.

Intel's announcement underscores its commitment to regaining leadership in the server processor market, where it has faced intense competition. The Xeon 6+ series is part of a broader roadmap that includes future generations built on Intel 14A and beyond. The company stated that the new chips will help customers achieve significant performance-per-watt improvements for AI and cloud workloads.

Intel plans to provide additional technical details at upcoming industry events, including the Hot Chips conference later this year. The Xeon 6+ processors are expected to enter volume production in the second half of 2025, with general availability following shortly after. Pricing will be announced at that time.

Nvidia is strengthening its position in the autonomous driving sector by expanding its DRIVE Hyperion platform through a series of new partnerships. The company announced collaborations with Foxconn, Uber, VinFast, and HUMAIN to develop robotaxi projects. These agreements mark a significant push by Nvidia to embed its technology across the global robotaxi ecosystem.

The DRIVE Hyperion platform serves as a complete reference architecture for autonomous vehicles, integrating Nvidia's DRIVE Orin system-on-a-chip, sensor suite, and software stack. By partnering with multiple companies, Nvidia aims to provide a standardized solution that can be adopted by various automakers and mobility service providers.

Foxconn, the Taiwanese electronics manufacturer, will leverage Nvidia's platform to develop electric vehicles and autonomous driving systems for its automotive clients. Uber, the ride-hailing giant, plans to integrate Nvidia's technology into its autonomous vehicle fleet, though specific details of the collaboration remain undisclosed.

Vietnamese automaker VinFast will use Nvidia's DRIVE Hyperion to accelerate the development of its self-driving cars. HUMAIN, a Chinese autonomous driving startup, will also adopt the platform for its robotaxi services. These partnerships highlight Nvidia's strategy to cover key markets, including North America, Asia, and Europe.

Nvidia's approach differs from competitors like Intel's Mobileye or Qualcomm, which offer more modular solutions. By providing a full-stack platform, Nvidia aims to reduce development time for automakers and mobility companies. The company claims its platform can handle Level 4 autonomous driving, meaning the vehicle can operate without human intervention in specific conditions.

The robotaxi market is projected to grow significantly in the coming years, with companies like Waymo, Cruise, and Baidu already testing services. Nvidia's partnerships could help it capture a larger share of this emerging market by supplying the underlying hardware and software.

Nvidia's DRIVE Hyperion platform is currently available for automakers and developers. The company has not disclosed specific financial terms of the partnerships or timelines for commercial deployment. However, Nvidia stated that the platform is production-ready and can be customized for different vehicle types.

With these new collaborations, Nvidia is positioning itself as a key enabler in the autonomous driving industry. The company's ability to secure partnerships with diverse players, from manufacturers to service providers, underscores its ambition to dominate the robotaxi ecosystem. Nvidia's CEO Jensen Huang emphasized that the company's platform offers a scalable and safe solution for autonomous mobility. The partnerships are expected to accelerate the deployment of robotaxis globally, though regulatory hurdles and public acceptance remain challenges. Nvidia continues to invest heavily in its automotive business, which it sees as a major growth driver beyond its core graphics and AI markets. The company's DRIVE platform is already used by over 370 partners worldwide, including automakers, trucking companies, and robotaxi developers. Nvidia's latest announcements reinforce its commitment to leading the autonomous driving revolution. The company plans to showcase its technology at upcoming industry events, including CES and the GPU Technology Conference. Nvidia's stock rose slightly following the news, reflecting investor optimism about its automotive prospects. The company faces competition from established players and new entrants, but its comprehensive platform approach may give it an edge in the race to deploy autonomous vehicles at scale.