The Trump administration is preparing to roll back significant portions of the Biden-era rules that govern global chip exports and AI technology transfers. According to Bloomberg's sources familiar with the matter, the so-called "AI diffusion" framework, due to take effect on May 15, will be scrapped in favor of simpler, bilateral licensing agreements. Under the current plan, rather than sorting about 120 countries into three tiers with differing volume caps, the US will negotiate individual contracts with partners like the United Arab Emirates and Saudi Arabia. Even with these changes, restrictions on China's access to advanced chips will stay firmly in place, and may even be reinforced. The proposed regulations are expected to maintain the outright ban on shipments to China, Russia, Iran, and North Korea, while adding stricter oversight for nations that have previously rerouted US-origin semiconductors toward Beijing's AI and military programs. US officials are also considering lowering the notification threshold for smaller shipments, from 1,700 NVIDIA H100 equivalent units down to around 500, to close loopholes used by alleged smuggling networks.

Industry reaction has been mixed but largely positive. Chipmakers saw their share prices climb when news of the repeal broke, citing hopes for clearer rules and fewer compliance headaches. Governments in Southeast Asia and Eastern Europe are watching closely and urging Washington to provide detailed guidance during the transition to avoid market disruptions. The AI diffusion rule was introduced in January 2025 with the goal of drawing countries such as India, Malaysia, and Poland into a more stringent export regime. Critics have argued that its complex, tiered system stifled innovation and diplomatic flexibility. The incoming framework will instead rely on targeted, outcome-driven accords that tie access to strategic investments and broader trade incentives. An official announcement could come as soon as Thursday, just before President Trump's trip to the Middle East. Final details are expected to be released in the coming weeks, marking a new chapter in US semiconductor diplomacy.

The first couple of paragraphs within Anthropic's "Securing America's Compute Advantage: (Our) Position on the Diffusion Rule" article are standard fare. Roughly half-way through a read of this policy-related piece, the North American (Amazon-backed) AI startup makes some bizarre claims about the smuggling of AI-oriented products into China. Given ongoing global tensions and growing industry demands, these activities are somewhat expected—but Anthropic leadership described very specific methodologies. As stated within their "Chip Smuggling is a Major Threat" passage: "China has established sophisticated smuggling operations, with documented cases involving hundreds of millions of dollars worth of chips. In some cases, smugglers have employed creative methods to circumvent export controls, including hiding processors in prosthetic baby bumps and packing GPUs alongside live lobsters." Specific bits of hardware were not mentioned in this section, but the author later alludes to the frictionless transfer of thousands of "NVIDIA H100 advanced chips" into Chinese territories.

In a statement issued to CNBC, a Team Green spokesperson dismissed Anthropic's fanciful claims: "American firms should focus on innovation and rise to the challenge, rather than tell tall tales that large, heavy, and sensitive electronics are somehow smuggled in 'baby bumps' or 'alongside' live lobsters." This very public spat has received mainstream attention; with further coverage documenting additional "to and fro" barbs. NVIDIA criticized Anthropic's anti-foreign competition stance: "China, with half of the world's AI researchers, has highly capable AI experts at every layer of the AI stack. America cannot manipulate regulators to capture victory in AI." Amusingly, Anthropic's operations rely heavily on Team Green hardware—many online critics reckon that top US AI companies are jostling for priority access to cutting-edge GPUs/accelerators. In reaction to NVIDIA's dismissal of their report, a company spokesperson retorted with: "Anthropic stands by its recently filed public submission in support of strong and balanced export controls that help secure America's lead in infrastructure development and ensure that the values of freedom and democracy shape the future of AI."

The Trump administration is preparing substantial changes to the Biden-era Framework for AI Diffusion controlling advanced semiconductor exports. Sources close to the Reuters indicate officials will replace the current three-tier country classification with a unified government-to-government licensing system requiring bilateral approval for US chip acquisitions. The existing framework, implemented in January 2025, permits unrestricted exports to 17 allied nations plus Taiwan, imposes volume caps on roughly 120 countries and blocks shipments to China, Russia, Iran, and North Korea. Current regulations exempt orders below 1,700 NVIDIA H100 equivalent units from full licensing requirements, needing only a notification.

Former Commerce Secretary Wilbur Ross, acting as an informal adviser, verified that bilateral government agreements are under review. Officials are also considering reducing the notification threshold from 1,700 to approximately 500 H100 equivalents to address circumvention concerns. The proposal has drawn criticism from industry figures, including Oracle Executive VP Ken Glueck and a coalition of seven Republican senators who have urged Commerce Secretary Howard Lutnick to withdraw the existing framework entirely. The administration faces pressure to finalize regulations before the May 15 compliance deadline, balancing security objectives with trade considerations. An announcement is expected before the month's end.

MangoBoost, a provider of cutting-edge system solutions designed to maximize AI data center efficiency, has set a new industry benchmark with its latest MLPerf Inference v5.0 submission. The company's Mango LLMBoost AI Enterprise MLOps software has demonstrated unparalleled performance on AMD Instinct MI300X GPUs, delivering the highest-ever recorded results for Llama2-70B in the offline inference category. This milestone marks the first-ever multi-node MLPerf inference result on AMD Instinct MI300X GPUs. By harnessing the power of 32 MI300X GPUs across four server nodes, Mango LLMBoost has surpassed all previous MLPerf inference results, including those from competitors using NVIDIA H100 GPUs.

Unmatched Performance and Cost Efficiency
MangoBoost's MLPerf submission demonstrates a 24% performance advantage over the best-published MLPerf result from Juniper Networks utilizing 32 NVIDIA H100 GPUs. Mango LLMBoost achieved 103,182 tokens per second (TPS) in the offline scenario and 93,039 TPS in the server scenario on AMD MI300X GPUs, outperforming the previous best result of 82,749 TPS on NVIDIA H100 GPUs. In addition to superior performance, Mango LLMBoost + MI300X offers significant cost advantages. With AMD MI300X GPUs priced between $15,000 and $17,000—compared to the $32,000-$40,000 cost of NVIDIA H100 GPUs (source: Tom's Hardware—H100 vs. MI300X Pricing)—Mango LLMBoost delivers up to 62% cost savings while maintaining industry-leading inference throughput.

Scientists and engineers of all kinds are equipped to solve tough problems a lot faster with NVIDIA CUDA-X libraries powered by NVIDIA GB200 and GH200 superchips. Announced today at the NVIDIA GTC global AI conference, developers can now take advantage of tighter automatic integration and coordination between CPU and GPU resources - enabled by CUDA-X working with these latest superchip architectures - resulting in up to 11x speedups for computational engineering tools and 5x larger calculations compared with using traditional accelerated computing architectures.

This greatly accelerates and improves workflows in engineering simulation, design optimization and more, helping scientists and researchers reach groundbreaking results faster. NVIDIA released CUDA in 2006, opening up a world of applications to the power of accelerated computing. Since then, NVIDIA has built more than 900 domain-specific NVIDIA CUDA-X libraries and AI models, making it easier to adopt accelerated computing and driving incredible scientific breakthroughs. Now, CUDA-X brings accelerated computing to a broad new set of engineering disciplines, including astronomy, particle physics, quantum physics, automotive, aerospace and semiconductor design.

According to a recent Center for Strategic and International Studies report, Huawei got its hand on approximately two million Ascend 910B logic dies through shell companies that misled TSMC. This acquisition violates US export controls designed to restrict China's access to advanced semiconductor technology. The report details how Huawei leveraged intermediaries to procure chiplets for its AI accelerators before TSMC discovered the deception and halted shipments. These components are critical for Huawei's AI hardware roadmap, which progressed from the original Ascend 910 (manufactured by TSMC on N7+ until 2020) to the domestically produced Ascend 910B and 910C chips fabricated at SMIC using first and second-generation 7 nm-class technologies, respectively. Huawei reportedly wanted TSMC-made dies because of manufacturing challenges in domestic chip production. The Ascend 910B and 910C reportedly suffer from poor yields, with approximately 25% of units failing during the advanced packaging process that combines compute dies with HBM memory.

Despite these challenges, the performance gap with market-leading solutions still remains but has narrowed considerably, with the Ascend 910C reportedly delivering 60% of NVIDIA H100's performance. Huawei has executed a strategic stockpiling initiative, particularly for high-bandwidth memory components. The company likely acquired substantial HBM inventory between August and December 2024, when restrictions on advanced memory sales to China were announced but not yet implemented. The semiconductor supply chain breach shows that enforcing technology export controls is challenging, and third parties can still purchase silicon for restricted companies. While Huawei continues building AI infrastructure for both internal projects and external customers, manufacturing constraints may limit its ability to scale deployments against competitors with access to more advanced manufacturing processes. Perhaps a future domestic EUV-based silicon manufacturing flow will allow Huawei to gain access to more advanced domestic production, completely circumventing US-imposed restrictions.

Cadence today announced that MediaTek has adopted the AI-driven Cadence Virtuoso Studio and Spectre X Simulator on the NVIDIA accelerated computing platform for its 2 nm development. As design size and complexity continue to escalate, advanced-node technology development has become increasingly challenging for SoC providers. To meet the aggressive performance and turnaround time (TAT) requirements for its 2 nm high-speed analog IP, MediaTek is leveraging Cadence's proven custom/analog design solutions, enhanced by AI, to achieve a 30% productivity gain.

"As MediaTek continues to push technology boundaries for 2 nm development, we need a trusted design solution with strong AI-powered tools to achieve our goals," said Ching San Wu, corporate vice president at MediaTek. "Closely collaborating with Cadence, we have adopted the Cadence Virtuoso Studio and Spectre X Simulator, which deliver the performance and accuracy necessary to achieve our tight design turnaround time requirements. Cadence's comprehensive automation features enhance our throughput and efficiency, enabling our designers to be 30% more productive."

The battle of AI acceleration in the data center is, as most readers are aware, insanely competitive, with NVIDIA offering a top-tier software stack. However, AMD has tried in recent years to capture a part of the revenue that hyperscalers and OEMs are willing to spend with its Instinct MI300X accelerator lineup for AI and HPC. Despite having decent hardware, the company is not close to bridging the gap software-wise with its competitor, NVIDIA. According to the latest report from SemiAnalysis, a research and consultancy firm, they have run a five-month experiment using Instinct MI300X for training and benchmark runs. And the findings were surprising: even with better hardware, AMD's software stack, including ROCm, has massively degraded AMD's performance.

"When comparing NVIDIA's GPUs to AMD's MI300X, we found that the potential on paper advantage of the MI300X was not realized due to a lack within AMD public release software stack and the lack of testing from AMD," noted SemiAnalysis, breaking down arguments in the report further, adding that "AMD's software experience is riddled with bugs rendering out of the box training with AMD is impossible. We were hopeful that AMD could emerge as a strong competitor to NVIDIA in training workloads, but, as of today, this is unfortunately not the case. The CUDA moat has yet to be crossed by AMD due to AMD's weaker-than-expected software Quality Assurance (QA) culture and its challenging out-of-the-box experience."

TSMC, the world leader in semiconductor manufacturing, is moving to production with NVIDIA's computational lithography platform, called cuLitho, to accelerate manufacturing and push the limits of physics for the next generation of advanced semiconductor chips. A critical step in the manufacture of computer chips, computational lithography is involved in the transfer of circuitry onto silicon. It requires complex computation - involving electromagnetic physics, photochemistry, computational geometry, iterative optimization and distributed computing. A typical foundry dedicates massive data centers for this computation, and yet this step has traditionally been a bottleneck in bringing new technology nodes and computer architectures to market.

Computational lithography is also the most compute-intensive workload in the entire semiconductor design and manufacturing process. It consumes tens of billions of hours per year on CPUs in the leading-edge foundries. A typical mask set for a chip can take 30 million or more hours of CPU compute time, necessitating large data centers within semiconductor foundries. With accelerated computing, 350 NVIDIA H100 Tensor Core GPU-based systems can now replace 40,000 CPU systems, accelerating production time, while reducing costs, space and power.

Giga Computing, a subsidiary of GIGABYTE and an industry leader in generative AI servers and advanced cooling technologies, today announced new flagship GIGABYTE G593 series servers supporting direct liquid cooling (DLC) technology to advance green data centers using NVIDIA HGX H200 GPU. As DLC technology is becoming a necessity for many data centers, GIGABYTE continues to increase its product portfolio with new DLC solutions for GPU and CPU technologies, and for these new G593 servers the cold plates are made by CoolIT Systems.

G593 Series - Tailored Cooling
The GPU-centric G593 series is custom engineered to house an 8-GPU baseboard, and its design had foresight for both air and liquid cooling. The compact 5U chassis leads the industry in its readily scalable nature, fitting up to sixty-four GPUs in a single rack and supporting 100kW of IT hardware. This helps to consolidate the IT hardware, and in turn, decrease the data center footprint. The G593 series servers for DLC are in response to the rising customer demand for greater energy efficiency. Liquids have a higher thermal conductivity than air, so they can rapidly and effectively remove heat from hot components to maintain lower operating temperatures. And by relying on water and heat exchangers, the overall energy consumption of the data center is reduced.

The MLCommons consortium on Wednesday posted MLPerf Inference v4.1 benchmark results for popular AI inferencing accelerators available in the market, across brands that include NVIDIA, AMD, and Intel. AMD's Instinct MI300X accelerators emerged competitive to NVIDIA's "Hopper" H100 series AI GPUs. AMD also used the opportunity to showcase the kind of AI inferencing performance uplifts customers can expect from its next-generation EPYC "Turin" server processors powering these MI300X machines. "Turin" features "Zen 5" CPU cores, sporting a 512-bit FPU datapath, and improved performance in AI-relevant 512-bit SIMD instruction-sets, such as AVX-512, and VNNI. The MI300X, on the other hand, banks on the strengths of its memory sub-system, FP8 data format support, and efficient KV cache management.

The MLPerf Inference v4.1 benchmark focused on the 70 billion-parameter LLaMA2-70B model. AMD's submissions included machines featuring the Instinct MI300X, powered by the current EPYC "Genoa" (Zen 4), and next-gen EPYC "Turin" (Zen 5). The GPUs are backed by AMD's ROCm open-source software stack. The benchmark evaluated inference performance using 24,576 Q&A samples from the OpenORCA dataset, with each sample containing up to 1024 input and output tokens. Two scenarios were assessed: the offline scenario, focusing on batch processing to maximize throughput in tokens per second, and the server scenario, which simulates real-time queries with strict latency limits (TTFT ≤ 2 seconds, TPOT ≤ 200 ms). This lets you see the chip's mettle in both high-throughput and low-latency queries.

A new startup emerged out of stealth mode today to power the next generation of generative AI. Etched is a company that makes an application-specific integrated circuit (ASIC) to process "Transformers." The transformer is an architecture for designing deep learning models developed by Google and is now the powerhouse behind models like OpenAI's GPT-4o in ChatGPT, Anthropic Claude, Google Gemini, and Meta's Llama family. Etched wanted to create an ASIC for processing only the transformer models, making a chip called Sohu. The claim is Sohu outperforms NVIDIA's latest and greatest by an entire order of magnitude. Where a server configuration with eight NVIDIA H100 GPU clusters pushes Llama-3 70B models at 25,000 tokens per second, and the latest eight B200 "Blackwell" GPU cluster pushes 43,000 tokens/s, the eight Sohu clusters manage to output 500,000 tokens per second.

Why is this important? Not only does the ASIC outperform Hopper by 20x and Blackwell by 10x, but it also serves so many tokens per second that it enables an entirely new fleet of AI applications requiring real-time output. The Sohu architecture is so efficient that 90% of the FLOPS can be used, while traditional GPUs boast a 30-40% FLOP utilization rate. This translates into inefficiency and waste of power, which Etched hopes to solve by building an accelerator dedicated to power transformers (the "T" in GPT) at massive scales. Given that the frontier model development costs more than one billion US dollars, and hardware costs are measured in tens of billions of US Dollars, having an accelerator dedicated to powering a specific application can help advance AI faster. AI researchers often say that "scale is all you need" (resembling the legendary "attention is all you need" paper), and Etched wants to build on that.

The full-stack NVIDIA accelerated computing platform has once again demonstrated exceptional performance in the latest MLPerf Training v4.0 benchmarks. NVIDIA more than tripled the performance on the large language model (LLM) benchmark, based on GPT-3 175B, compared to the record-setting NVIDIA submission made last year. Using an AI supercomputer featuring 11,616 NVIDIA H100 Tensor Core GPUs connected with NVIDIA Quantum-2 InfiniBand networking, NVIDIA achieved this remarkable feat through larger scale - more than triple that of the 3,584 H100 GPU submission a year ago - and extensive full-stack engineering.

Thanks to the scalability of the NVIDIA AI platform, Eos can now train massive AI models like GPT-3 175B even faster, and this great AI performance translates into significant business opportunities. For example, in NVIDIA's recent earnings call, we described how LLM service providers can turn a single dollar invested into seven dollars in just four years running the Llama 3 70B model on NVIDIA HGX H200 servers. This return assumes an LLM service provider serving Llama 3 70B at $0.60/M tokens, with an HGX H200 server throughput of 24,000 tokens/second.

The 63rd edition of the TOP500 reveals that Frontier has once again claimed the top spot, despite no longer being the only exascale machine on the list. Additionally, a new system has found its way into the Top 10.

The Frontier system at Oak Ridge National Laboratory in Tennessee, USA remains the most powerful system on the list with an HPL score of 1.206 EFlop/s. The system has a total of 8,699,904 combined CPU and GPU cores, an HPE Cray EX architecture that combines 3rd Gen AMD EPYC CPUs optimized for HPC and AI with AMD Instinct MI250X accelerators, and it relies on Cray's Slingshot 11 network for data transfer. On top of that, this machine has an impressive power efficiency rating of 52.93 GFlops/Watt - putting Frontier at the No. 13 spot on the GREEN500.

Quantum computing. Drug discovery. Fusion energy. Scientific computing and physics-based simulations are poised to make giant steps across domains that benefit humanity as advances in accelerated computing and AI drive the world's next big breakthroughs. NVIDIA unveiled at GTC in March the NVIDIA Blackwell platform, which promises generative AI on trillion-parameter large language models (LLMs) at up to 25x less cost and energy consumption than the NVIDIA Hopper architecture.

Blackwell has powerful implications for AI workloads, and its technology capabilities can also help to deliver discoveries across all types of scientific computing applications, including traditional numerical simulation. By reducing energy costs, accelerated computing and AI drive sustainable computing. Many scientific computing applications already benefit. Weather can be simulated at 200x lower cost and with 300x less energy, while digital twin simulations have 65x lower cost and 58x less energy consumption versus traditional CPU-based systems and others.

NVIDIA's next-gen Blackwell platform, which includes B-series GPUs and integrates NVIDIA's own Grace Arm CPU in models such as the GB200, represents a significant development. TrendForce points out that the GB200 and its predecessor, the GH200, both feature a combined CPU+GPU solution, primarily equipped with the NVIDIA Grace CPU and H200 GPU. However, the GH200 accounted for only approximately 5% of NVIDIA's high-end GPU shipments. The supply chain has high expectations for the GB200, with projections suggesting that its shipments could exceed millions of units by 2025, potentially making up nearly 40 to 50% of NVIDIA's high-end GPU market.

Although NVIDIA plans to launch products such as the GB200 and B100 in the second half of this year, upstream wafer packaging will need to adopt more complex and high-precision CoWoS-L technology, making the validation and testing process time-consuming. Additionally, more time will be required to optimize the B-series for AI server systems in aspects such as network communication and cooling performance. It is anticipated that the GB200 and B100 products will not see significant production volumes until 4Q24 or 1Q25.

During the Vision 2024 event, Intel announced its latest Gaudi 3 AI accelerator, promising significant improvements over its predecessor. Intel claims the Gaudi 3 offers up to 70% improvement in training performance, 50% better inference, and 40% better efficiency than Nvidia's H100 processors. The new AI accelerator is presented as a PCIe Gen 5 dual-slot add-in card with a 600 W TDP or an OAM module with 900 W. The PCIe card has the same peak 1,835 TeraFLOPS of FP8 performance as the OAM module despite a 300 W lower TDP. The PCIe version works as a group of four per system, while the OAM HL-325L modules can be run in an eight-accelerator configuration per server. This likely will result in a lower sustained performance, given the lower TDP, but it confirms that the same silicon is used, just finetuned with a lower frequency. Built on TSMC's N5 5 nm node, the AI accelerator features 64 Tensor Cores, delivering double the FP8 and quadruple FP16 performance over the previous generation Gaudi 2.

The Gaudi 3 AI chip comes with 128 GB of HBM2E with 3.7 TB/s of bandwidth and 24 200 Gbps Ethernet NICs, with dual 400 Gbps NICs used for scale-out. All of that is laid out on 10 tiles that make up the Gaudi 3 accelerator, which you can see pictured below. There is 96 MB of SRAM split between two compute tiles, which acts as a low-level cache that bridges data communication between Tensor Cores and HBM memory. Intel also announced support for the new performance-boosting standardized MXFP4 data format and is developing an AI NIC ASIC for Ultra Ethernet Consortium-compliant networking. The Gaudi 3 supports clusters of up to 8192 cards, coming from 1024 nodes comprised of systems with eight accelerators. It is on track for volume production in Q3, offering a cost-effective alternative to NVIDIA accelerators with the additional promise of a more open ecosystem. More information and a deeper dive can be found in the Gaudi 3 Whitepaper.

NVIDIA's freshly unveiled "Blackwell" B200 and GB200 AI GPUs will be getting plenty of coverage this year, but many organizations will be sticking with current or prior generation hardware. Team Green is in the process of shipping out compromised "Hopper" designs to customers in China, but the region's appetite for powerful AI-crunching hardware is growing. Last year's China-specific H800 design, and the older "Ampere" A800 chip were deemed too potent—new regulations prevented further sales. Recently, AMD's Instinct MI309 AI accelerator was considered "too powerful to gain unconditional approval from the US Department of Commerce." Natively-developed solutions are catching up with Western designs, but some institutions are not prepared to queue up for emerging technologies.

NVIDIA's new H20 AI GPU as well as Ada Lovelace-based L20 PCIe and L2 PCIe models are weakened enough to get a thumbs up from trade regulators, but likely not compelling enough for discerning clients. The Telegraph believes that NVIDIA's uncompromised H100 AI GPU is currently in use at several Chinese establishments—the report cites information presented within four academic papers published on ArXiv, an open access science website. The Telegraph's news piece highlights one of the studies—it was: "co-authored by a researcher at 4paradigm, an AI company that was last year placed on an export control list by the US Commerce Department for attempting to acquire US technology to support China's military." Additionally, the Chinese Academy of Sciences appears to have conducted several AI-accelerated experiments, involving the solving of complex mathematical and logical problems. The article suggests that this research organization has acquired a very small batch of NVIDIA H100 GPUs (up to eight units). A "thriving black market" for high-end NVIDIA processors has emerged in the region—last Autumn, the Center for a New American Security (CNAS) published an in-depth article about ongoing smuggling activities.

At GTC on Monday, Microsoft Corp. and NVIDIA expanded their longstanding collaboration with powerful new integrations that leverage the latest NVIDIA generative AI and Omniverse technologies across Microsoft Azure, Azure AI services, Microsoft Fabric and Microsoft 365.

"Together with NVIDIA, we are making the promise of AI real, helping to drive new benefits and productivity gains for people and organizations everywhere," said Satya Nadella, Chairman and CEO, Microsoft. "From bringing the GB200 Grace Blackwell processor to Azure, to new integrations between DGX Cloud and Microsoft Fabric, the announcements we are making today will ensure customers have the most comprehensive platforms and tools across every layer of the Copilot stack, from silicon to software, to build their own breakthrough AI capability."

"AI is transforming our daily lives - opening up a world of new opportunities," said Jensen Huang, founder and CEO of NVIDIA. "Through our collaboration with Microsoft, we're building a future that unlocks the promise of AI for customers, helping them deliver innovative solutions to the world."

NVIDIA today announced its next-generation AI supercomputer—the NVIDIA DGX SuperPOD powered by NVIDIA GB200 Grace Blackwell Superchips—for processing trillion-parameter models with constant uptime for superscale generative AI training and inference workloads.

Featuring a new, highly efficient, liquid-cooled rack-scale architecture, the new DGX SuperPOD is built with NVIDIA DGX GB200 systems and provides 11.5 exaflops of AI supercomputing at FP4 precision and 240 terabytes of fast memory—scaling to more with additional racks.

Powering a new era of computing, NVIDIA today announced that the NVIDIA Blackwell platform has arrived—enabling organizations everywhere to build and run real-time generative AI on trillion-parameter large language models at up to 25x less cost and energy consumption than its predecessor.

The Blackwell GPU architecture features six transformative technologies for accelerated computing, which will help unlock breakthroughs in data processing, engineering simulation, electronic design automation, computer-aided drug design, quantum computing and generative AI—all emerging industry opportunities for NVIDIA.

NVIDIA today announced that TSMC and Synopsys are going into production with NVIDIA's computational lithography platform to accelerate manufacturing and push the limits of physics for the next generation of advanced semiconductor chips. TSMC, the world's leading foundry, and Synopsys, the leader in silicon to systems design solutions, have integrated NVIDIA cuLitho with their software, manufacturing processes and systems to speed chip fabrication, and in the future support the latest-generation NVIDIA Blackwell architecture GPUs.

"Computational lithography is a cornerstone of chip manufacturing," said Jensen Huang, founder and CEO of NVIDIA. "Our work on cuLitho, in partnership with TSMC and Synopsys, applies accelerated computing and generative AI to open new frontiers for semiconductor scaling." NVIDIA also introduced new generative AI algorithms that enhance cuLitho, a library for GPU-accelerated computational lithography, dramatically improving the semiconductor manufacturing process over current CPU-based methods.

GIGABYTE Technology and Giga Computing, a subsidiary of GIGABYTE and an industry leader in enterprise solutions, will showcase their solutions at the GIGABYTE booth #1224 at NVIDIA GTC, a global AI developer conference running through March 21. This event will offer GIGABYTE the chance to connect with its valued partners and customers, and together explore what the future in computing holds.

The GIGABYTE booth will focus on GIGABYTE's enterprise products that demonstrate AI training and inference delivered by versatile computing platforms based on NVIDIA solutions, as well as direct liquid cooling (DLC) for improved compute density and energy efficiency. Also not to be missed at the NVIDIA booth is the MGX Pavilion, which features a rack of GIGABYTE servers for the NVIDIA GH200 Grace Hopper Superchip architecture.

TSMC is experiencing unprecedented demand from AI chip customers—unnamed parties have (fancifully) requested the construction of entirely new fabrication facilities. Taiwan's leading semiconductor contract manufacturer seems to concentrating on "sensible" expansions, mainly in the area of CoWoS packaging output—according to an Economic Daily report, company leadership and local government were negotiating over the construction of four new advanced packaging plants. Insiders propose that plans have been revised—an investment in excess of 500 billion yuan ($16 billion) will enable the founding of six new CoWoS-focused facilities. TSMC is expected to make an official announcement next month—industry moles reckon that construction work will start in April. Two (of the six total) advanced packaging plants could become fully operational before the conclusion of 2024.

Lately, TSMC has initiated an ambitious recruitment drive—targeting around 6000 new workers. A touring entity is tasked with the attraction of "talents with high enthusiasm for semiconductors." The majority of new recruits are likely heading to new or expanded Taiwan-based facilities. The Economic Daily report proposes that Chiayi City's technological hub will play host to TSMC's new CoWoS packaging plants. A DigiTimes Asia news piece (from January) posited that TSMC leadership anticipates CoWoS output reaching 44,000 units by the end of 2024. This predicted tally could grow, thanks to the (rumored) activation of additional factories. CoWoS packaging is considered to be a vital aspect of AI accelerators—insiders believe that TSMC's latest investment will boost production of NVIDIA H100 GPUs. The combined output of six new CoWoS plants will assist greatly in the creation of next-gen B100 chips.

Stability AI, the developers behind the popular Stable Diffusion generative AI model, have run some first-party performance benchmarks for Stable Diffusion 3 using popular data-center AI GPUs, including the NVIDIA H100 "Hopper" 80 GB, A100 "Ampere" 80 GB, and Intel's Gaudi2 96 GB accelerator. Unlike the H100, which is a super-scalar CUDA+Tensor core GPU; the Gaudi2 is purpose-built to accelerate generative AI and LLMs. Stability AI published its performance findings in a blog post, which reveals that the Intel Gaudi2 96 GB is posting a roughly 56% higher performance than the H100 80 GB.

With 2 nodes, 16 accelerators, and a constant batch size of 16 per accelerator (256 in all), the Intel Gaudi2 array is able to generate 927 images per second, compared to 595 images for the H100 array, and 381 images per second for the A100 array, keeping accelerator and node counts constant. Scaling things up a notch to 32 nodes, and 256 accelerators or a batch size of 16 per accelerator (total batch size of 4,096), the Gaudi2 array is posting 12,654 images per second; or 49.4 images per-second per-device; compared to 3,992 images per second or 15.6 images per-second per-device for the older-gen A100 "Ampere" array.