CONNECT WITH US
Friday 9 May 2025
Cutting power and costs: DEEPX outperforms even free chips in TCO
In today's AI economy, reliability is not optional—it's essential. AI now runs factory lines, city cameras, and delivery robots, where even a one-second pause can trigger costly failures or safety risks. Any AI system that can't operate 24/7 without human intervention is simply not viable.To succeed at the edge, AI must meet four strict demands: sub-100 ms latency, 99.999% uptime, a power budget under 20 W, and junction temperatures below 85 °C. Without these, systems overheat, slow down, or fail in the field.Credit: DEEPXArchitected for reliability: DX-M1's thermal and performance breakthroughsThe GPGPU-based AI systems fall short of these requirements. They consume over 40 W—far beyond what low-power infrastructure and mobile robot batteries can support. They also require fans, heat sinks, and vents, which add noise, cost, and new points of failure. Moreover, their dependence on remote servers introduces cloud latency and ongoing bandwidth expenses.DEEPX overturns these hurdles. The DX‑M1 chip delivers GPGPU-class accuracy while consuming less than 3 W of power. In thermal testing with YOLOv7 at 33 FPS under identical conditions, DX‑M1 maintained a stable 61.9 °C, while a leading competitor overheated to 113.5 °C—enough to trigger thermal throttling. Under maximum load, DX‑M1 sustained 75.4 °C while achieving 59 FPS, whereas the competitor reached only 32 FPS at 114.3 °C. This demonstrates that DX‑M1 delivers 84 % better performance while running 38.9 °C cooler.A key strength of the DX-M1's architecture is its balance of speed and stability. Unlike some DRAM-less NPUs that rely on bulky on-chip SRAM—often leading to overheating, slowdowns, and low manufacturing yield—DX‑M1 combines compact SRAM with high-speed LPDDR5 DRAM positioned close to the chip. This results in smoother, cooler, and more reliable AI performance, even in compact, power-constrained environments. As a result, DX-M1 reduces hardware and energy costs by up to 90 %, making it one of the most cost-effective AI chips available.Credit: DEEPXThe true cost of AI hardware: More than the price tagDEEPX recently supported two customers building AI systems for factory robots and on-site servers. At first, both companies planned to use 40 W GPGPUs. But during testing, they realized the hidden costs:Running a 40 W GPGPU nonstop for five years uses twice as much money in electricity as it costs to buy one DX-M1 chip. The heat from GPGPUs requires fans and cooling systems, which consume extra power and increase maintenance needs. Even if the GPGPU hardware were free, the total cost of operation would still be more than double compared to using DX-M1.When the companies tested multiple NPU vendors for power efficiency, heat, and accuracy, they found that DEEPX's DX-M1 was the best fit for their real-world use. Over five years, DX-M1 cuts electricity and cooling costs by about 94% compared to GPGPU-based systems. This huge saving gives companies using AI at scale a major business advantage.In short, the most cost-effective AI hardware is not the one with the lowest price tag—but the one that delivers high performance with low power, stable heat, and reliable results over time.The future of AI will be built not just on speed or model size—but on reliability. Without stable, predictable performance, AI cannot scale into the real world. In factories, cities, and autonomous machines, even a momentary delay can lead to failure, risk, or lost trust. That's why reliability isn't just important—it's foundational. DEEPX is leading this transformation by reducing risk, lowering long-term costs, and delivering AI that operates independently, safely, and without interruption.Credit: DEEPX
Friday 9 May 2025
Is your AI system built to last—or bound to fail? Only DEEPX has the answer
DEEPX ensures unmatched AI reliability with lower power, lower heat, and a total cost of ownership lower than even "free" chips.For Lokwon Kim, founder and CEO of DEEPX, this isn't just an ambition—it's a foundational requirement for the AI era. A veteran chip engineer who once led advanced silicon development at Apple, Broadcom, and Cisco, Kim sees the coming decade as a defining moment to push the boundaries of technology and shape the future of AI. While others play pricing games, Kim is focused on building what the next era demands: AI systems that are truly reliable."This white paper," Kim says, holding up a recently published technology report, "isn't about bragging rights. It's about proving that what we're building actually solves the real-world challenges faced by factories, cities, and robots—right now."Credit: DEEPXA new class of reliability for AI systemsWhile GPGPUs continue to dominate cloud-based AI training, Kim argues that the true era of AI begins not in server racks, but in the everyday devices people actually use. From smart cameras and robots to kiosks and industrial sensors, AI must be embedded where life happens—close to the user, and always on.And because these devices operate in power-constrained, fanless, and sometimes battery-driven environments, low power isn't a preference—it's a hard requirement. Cloud-bound GPUs are too big, too hot, and too power-hungry to meet this reality. On-device AI demands silicon that is lean, efficient, and reliable enough to run continuously—without overheating, without delay, and without failure."You can't afford to lose a single frame in a smart camera, miss a barcode in a warehouse, or stall a robot on an assembly line," Kim explains. "These moments define success or failure."GPGPU-based and many NPU competitor systems fail this test. With high power draw, significant heat generation, the need for active cooling, and cloud latency issues, they are fundamentally ill-suited for the always-on, low-power edge. In contrast, DEEPX's DX-M1 runs under 3W, stays below 80°C with no fan, and delivers GPU-class inference accuracy with zero latency dependency.Under identical test conditions, the DX-M1 outperformed competing NPUs by up to 84%, while maintaining 38.9°C lower operating temperatures, and being 4.3× smaller in die size.This is made possible by rejecting the brute-force SRAM-heavy approach and instead using a lean, on-chip SRAM + LPDDR5 DRAM architecture that enables:• Higher manufacturing yield• Lower field failure rates• Elimination of PCIe bottlenecks• 100+ FPS inference even on small embedded boardsDEEPX also developed its own quantization pipeline, IQ8™, preserving <1% accuracy loss across 170+ models."We've proven you can dramatically reduce power and memory without sacrificing output quality," Kim says.Credit: DEEPXReal customers. Real deployments. Real impact.Kim uses a powerful metaphor to describe the company's strategic position."If cloud AI is a deep ocean ruled by GPGPU-powered ships, then on-device AI is the shallow sea—close to land, full of opportunities, and hard to navigate with heavy hardware."GPGPU, he argues, is structurally unsuited to play in this space. Their business model and product architecture are simply too heavy to pivot to low-power, high-flexibility edge scenarios."They're like battleships," Kim says. "We're speedboats—faster, more agile, and able to handle 50 design changes while they do one."DEEPX isn't building in a vacuum. The DX-M1 is already being validated by major companies like Hyundai Robotics Lab, POSCO DX and LG Uplus, which rejected GPGPU-based designs due to energy, cost, and cooling concerns. The companies found that even "free" chips resulted in a higher total cost of ownership (TCO) than the DX-M1—once you add electricity bills, cooling systems, and field failure risks.According to Kim, "Some of our collaborations realized that switching to DX-M1 saves up to 94% in power and cooling costs over five years. And that savings scales exponentially when you deploy millions of devices."Building on this momentum, DEEPX is now entering full-scale mass production of the DX-M1, its first-generation NPU built on a cutting-edge 5nm process. Unlike many competitors still relying on 10–20nm nodes, DEEPX has achieved an industry-leading 90% yield at 5nm, setting the stage for dominant performance, efficiency, and scalability in the edge AI market.Looking beyond current deployments, DEEPX is now developing its next-generation chip, the DX-M2—a groundbreaking on-device AI processor designed to run LLMs under 5W. As large language model technology evolves, the field is beginning to split in two directions: one track continues to scale up LLMs in cloud data centers in pursuit of AGI; the other, more practical path focuses on lightweight, efficient models optimized for local inference—such as DeepSeek and Meta's LLaMA 4. DEEPX's DX-M2 is purpose-built for this second future.With ultra-low power consumption, high performance, and a silicon architecture tailored for real-world deployment, the DX-M2 will support LLMs like DeepSeek and LLaMA 4 directly at the edge—no cloud dependency required. Most notably, DX-M2 is being developed to become the first AI inference chip built on the leading-edge 2nm process—marking a new era of performance-per-watt leadership. In short, DX-M2 isn't just about running LLMs efficiently—it's about unlocking the next stage of intelligent devices, fully autonomous and truly local.Credit: DEEPXIf ARM defined the mobile era, DEEPX will define the AI EraLooking ahead, Kim positions DEEPX not as a challenger to cloud chip giants, but as the foundational platform for the AI edge—just as ARM once was for mobile."We're not chasing the cloud," he says. "We're building the stack that powers AI where it actually interacts with the real world—at the edge."In the 1990s, ARM changed the trajectory of computing by creating power-efficient, always-on architectures for mobile devices. That shift didn't just enable smartphones—it redefined how and where computing happens."History repeats itself," Kim says. "Just as ARM silently powered the mobile revolution, DEEPX is quietly laying the groundwork for the AI revolution—starting from the edge."His 10-year vision is bold: to make DEEPX the "next ARM" of AI systems, enabling AI to live in the real world—not the cloud. From autonomous robots and smart city kiosks to factory lines and security systems, DEEPX aims to become the default infrastructure where AI must run reliably, locally, and on minimal power.Everyone keeps asking about the IPO. Here's what Kim really thinks.With DEEPX gaining attention as South Korea's most promising AI semiconductor company, one question keeps coming up: When's the IPO? But for founder and CEO Lokwon Kim, the answer is clear—and measured."Going public isn't the objective itself—it's a strategic step we'll take when it aligns with our vision for sustainable success." Kim says. "Our real focus is building proof—reliable products, real deployments, actual revenue. A unicorn company is one that earns its valuation through execution—especially in semiconductors, where expectations are unforgiving. The bar is high, and we intend to meet it."That milestone, Kim asserts, is no longer far away. In other words, DEEPX isn't rushing for headlines—it's building for history. DEEPX isn't just designing chips—it's designing trust.In an AI-powered world where milliseconds can mean millions, reliability is everything. As AI moves from cloud to edge, from theory to infrastructure, the companies that will define the next decade aren't those chasing faster clocks—but those building systems that never fail."We're not here to ride a trend," Kim concludes. "We're here to solve the hardest problems—the ones that actually matter."Credit: DEEPXWhen Reliability Matters Most—Industry Leaders Choose DEEPXVisit DEEPX at booth 4F, L0409 from May 20-23 at Taipei Nangang Exhibition Center to witness firsthand how we're setting new standards for reliable on-device AI.For more information, you can follow DEEPX on social media or visit their official website.
Tuesday 6 May 2025
Best and brightest AI minds celebrated at Taiwan's inaugural Best AI Awards
To cultivate a stronger foundation of artificial intelligence (AI) talent and encourage greater investment in AI research and development, Taiwan's Ministry of Economic Affairs (MOEA) hosted the inaugural 'Best AI Award Competition,' culminating in the finals held on May 3, 2025, at the Taipei International Trade Center's Exhibition Hall 1.This year's competition attracted participation from 1,253 teams spanning 36 countries. From this pool, 233 teams advanced to the finals, where they competed for gold, silver, and bronze awards, as well as honorable mentions, across the 'AI Application' and 'IC Design' categories. These categories were further divided into groups: Public Corporations, SMEs and Startups, Students, and International Teams.The Gold Award was bestowed upon eight winners: HiTRUSTpay, EYS3D Microelectronics, Daya Yoo, Jmem Tek, National Central University, National Taiwan University, Touch Lab (Philippines), and Arba Lab (UK).According to an MOEA press release, the 'Best AI Awards' aspires to be Taiwan's equivalent of the "Oscars for AI", embodying diversity, global reach, and a forward-looking vision. Its core objectives are to ignite the creativity of the next generation, foster stronger ties between academia and industry, and nurture a deeper pool of AI-savvy talent and innovative enterprises. Ultimately, the competition aims to drive the industrialization of AI and the adoption of AI across industries, thereby solidifying Taiwan's position in the AI landscape. The competition offers substantial prizes, with student group winners vying for up to NT$300,000 and winners in the enterprise open, startup and SME, and international groups competing for up to NT$1 million.Credit: MOEAAt the awards ceremony, Deputy Minister Ho Chin-tsang emphasized AI's accelerating transformation of the global industrial structure, impacting sectors from manufacturing and healthcare to finance and everyday services. He stressed the 'Best AI Awards' competition platform's crucial role in forging stronger links between talent development, technological applications, and industry demands. In response to the widespread anticipation for AI technologies, the MOEA seeks to foster a practical, application-oriented approach, encouraging innovative concepts to address real-world industry challenges. This strategy aims to continuously cultivate new talent and generate cutting-edge solutions within Taiwan's evolving AI ecosystem. Deputy Minister Ho also highlighted that the competition entries exemplify the fusion of AI technology with tangible needs and creative execution, showcasing the immense potential of translating innovative ideas into viable solutions. Looking ahead, the MOEA will sustain its commitment to facilitating the adoption of innovative solutions and maximizing their value through strategic policy initiatives and industry partnerships.Credit: MOEADirector-General Kuo Chao-chung of the MOEA's Department of Industrial Technology underscored the impressive international engagement of the inaugural Best AI Awards, attracting both enthusiastic domestic participation from enterprises and academic institutions and the involvement of 353 international professionals from 36 countries, including India, the Philippines, the United States, and the United Kingdom. This global participation establishes the competition as a vital platform for the international exchange of AI innovation. To expedite the industry's capacity building in AI talent and applications, the MOEA will not only continue to host the Best AI Awards but also leverage the pilot production capabilities of schools and research institutions to support businesses in design, new product development, and prototyping. Furthermore, the MOEA will collaborate with agencies such as the Small and Medium Enterprise Administration and the Industrial Development Administration to facilitate industry-wide AI transformation and develop practical AI expertise.Chiu Chui-hui, Director-General of the Industrial Development Administration, cited a report by the Artificial Intelligence Technology Foundation, which identifies the 'shortage of relevant technical talent' as Taiwan's primary obstacle to AI advancement. The Global Artificial Intelligence Index corroborates this, revealing that while Taiwan excels in infrastructure (ranking 4th), it faces challenges in talent (38th), research (27th), and commercialization (39th). Consequently, the Best AI Awards are designed to expedite the real-world application of AI and the cultivation of skilled professionals. Chiu emphasized the competition's significance as Taiwan's premier AI competition, characterized by its scale, prestigious awards, and high standards (with a highly selective 7.4% award rate). He expressed his hope for collaborative efforts across all sectors to broaden the adoption of AI and harness its power to drive industrial innovation.The 'Best AI Awards' entries spanned a diverse range of application areas, including information and communication technology (18.4%), manufacturing (16.2%), healthcare (15.9%), wholesale and retail (10.2%), education (8.6%), and finance (7.8%). This diversity not only underscores the application of cutting-edge technologies but also highlights the immense potential of AI to be successfully integrated into various industries.The gold medal winners' covered areas and their attempted solutions to problems are summarized as follows:CategoryTeamWork TitleDomainPain Points to SolveAIStudentNational Central UniversityRealization of Highly Flexible Production LinesManufacturing AutomationTraditional production lines are inflexible and struggle to adapt to rapidly changing and varied production needs.AI SME and StartupData YooFarmiSpaceFarmingEnhance the efficiency and management of agricultural production, and optimize agricultural resource allocationAI Public CompanyHiTRUSTpayVeri-id equipment id verification and real-time AI anti-scam/fraud serviceCybersecurity, FintechSolve device identity verification issues, and provide real-time prevention of financial fraud, thereby enhancing transaction security.AI_INTLTouchLab (Philippines)AI driven TOUCH System – Digitizing TouchDigitalization, human-machine interactionDigitize touch, enabling machines to understand and simulate tactile sensations, thereby enhancing the precision and scope of human-computer interaction.IC StudentNational Taiwan UniversityTernary-weight Transformer model software-hardware synergetic design and neural network accelerator IC design and implementationIC design, AI hardware accelerationDesign higher-performance, lower-power AI chips to accelerate the computation of Transformer models and improve the efficiency of AI applications.IC SME & StartupJmem TekArgusNPU – PQC security edge AI inference systemCybersecurity, Edge AI, AI IC designProvide AI inference capabilities with post-quantum security on edge devices to protect sensitive data and enable high-performance AI applicationsIC Public CompanyeYs3D MicroelectronAI Edge Computing Car Parking Management SystemSmart City Transportation ManagementUse edge AI to improve parking management efficiency, optimize the allocation of parking resources, and alleviate traffic congestion.IC_INTLArbaLab(UK)ArbaEdgeAI Edge ComputingRealizing high-performance AI computing on edge devices to reduce reliance on cloud computing, enhance response speed, and improve privacy protection.Source: MOEAThe winning entries of the 'Best AI Awards' showcased the dynamic development and diverse applications of AI technology, with particularly strong innovation evident in sectors such as healthcare and manufacturing, as well as in the burgeoning fields of AIoT and edge AI integration.The Ministry of Economic Affairs pointed out that looking ahead, it will continue to collaborate with industry, academia, and research institutions, hoping that the "Best AI Awards" can become an important annual platform to promote Taiwan's AI technology development, talent cultivation, and innovative applications, and to continuously discover more promising new talents. In addition, matchmaking events will also be held concurrently during the "COMPUTEX" exhibition every May, with plans to invite more than 20 domestic and foreign venture capitalists and buyers to participate, fostering in-depth exchanges and cooperation between participating teams and the industry, and further expanding the commercialization opportunities for AI innovative applications.Through this comprehensive strategy, the Ministry of Economic Affairs aims to expedite the creation of groundbreaking AI applications and the cultivation of interdisciplinary AI expertise, ultimately steering Taiwan toward its ambitious vision of becoming a globally recognized "AI Island". For the latest updates, follow the official LinkedIn page of the Best AI Awards.