In context: TSMC has steadily raised the prices of its most advanced semiconductor process nodes over the past several years – so much so that one analysis suggests the cost per transistor hasn't decreased in over a decade. Further price hikes, driven by tariffs and rising development costs, are reinforcing the notion that Moore's Law is truly dead. The Commercial Times reports that TSMC's upcoming N2 2nm semiconductors will cost $30,000 per wafer, a roughly 66% increase over the company's 3nm chips. Future nodes are expected to be even more expensive and likely reserved for the largest manufacturers. TSMC has justified these price increases by citing the massive cost of building 2nm fabrication plants, which can reach up to $725 million. According to United Daily News, major players such as Apple, AMD, Qualcomm, Broadcom, and Nvidia are expected to place orders before the end of the year despite the higher prices, potentially bringing TSMC's 2nm Arizona fab to full capacity. Also see: How profitable are TSMC's nodes: crunching the numbers Unsurprisingly, Apple is getting first dibs. The A20 processor in next year's iPhone 18 Pro is expected to be the first chip based on TSMC's N2 process. Intel's Nova Lake processors, targeting desktops and possibly high-end laptops, are also slated to use N2 and are expected to launch next year. Earlier reports indicated that yield rates for TSMC's 2nm process reached 60% last year and have since improved. New data suggests that 256Mb SRAM yield rates now exceed 90%. Trial production is likely already underway, with mass production scheduled to begin later this year. // Related Stories With tape-outs for 2nm-based designs surpassing previous nodes at the same development stage, TSMC aims to produce tens of thousands of wafers by the end of 2025. TSMC also plans to follow N2 with N2P and N2X in the second half of next year. N2P is expected to offer an 18% performance boost over N3E at the same power level and 36% greater energy efficiency at the same speed, along with significantly higher logic density. N2X, slated for mass production in 2027, will increase maximum clock frequencies by 10%. As semiconductor geometries continue to shrink, power leakage becomes a major concern. TSMC's 2nm nodes will address this issue with gate-all-around (GAA) transistor architectures, enabling more precise control of electrical currents. Beyond 2nm lies the Angstrom era, where TSMC will implement backside power delivery to further enhance performance. Future process nodes like A16 (1.6nm) and A14 (1.4nm) could cost up to $45,000 per wafer. Meanwhile, Intel is aiming to outpace TSMC's roadmap. The company recently began risk production of its A18 node, which also features gate-all-around and backside power delivery. These chips are expected to debut later this year in Intel's upcoming laptop CPUs, codenamed Panther Lake.

Taiwanese chip titan TSMC said Tuesday that it expected to see record earnings this year as it increased semiconductor production capacity to meet soaring demand for AI technology.

Wafer Inspection Equipment Market Revenue Growth 2025,  openPR.com

From large language models (LLMs) to reasoning agents, today’s AI tools bring unprecedented computational demands. Trillion-parameter models, workloads running on-device, and swarms of agents collaborating to complete tasks all require a new paradigm of computing to become truly seamless and ubiquitous. Silicon’s mid-life crisis AI has evolved from classical ML to deep learning to generative AI. The most recent chapter, which took AI mainstream, hinges on two phases—training and inference—that are data and energy-intensive in terms of computation, data movement, and cooling. At the same time, Moore’s Law, which determines that the number of transistors on a chip doubles every two years, is reaching a physical and economic plateau. For the last 40 years, silicon chips and digital technology have nudged each other forward—every step ahead in processing capability frees the imagination of innovators to envision new products, which require yet more power to run. That is happening at light speed in the AI age. As models become more readily available, deployment at scale puts the spotlight on inference and the application of trained models for everyday use cases. This transition requires the appropriate hardware to handle inference tasks efficiently. Central processing units (CPUs) have managed general computing tasks for decades, but the broad adoption of ML introduced computational demands that stretched the capabilities of traditional CPUs. This has led to the adoption of graphics processing units (GPUs) and other accelerator chips for training complex neural networks, due to their parallel execution capabilities and high memory bandwidth that allow large-scale mathematical operations to be processed efficiently. But CPUs are already the most widely deployed and can be companions to processors like GPUs and tensor processing units (TPUs). AI developers are also hesitant to adapt software to fit specialized or bespoke hardware, and they favor the consistency and ubiquity of CPUs. Chip designers are unlocking performance gains through optimized software tooling, adding novel processing features and data types specifically to serve ML workloads, integrating specialized units and accelerators, and advancing silicon chip innovations, including custom silicon. AI itself is a helpful aid for chip design, creating a positive feedback loop in which AI helps optimize the chips that it needs to run. These enhancements and strong software support mean modern CPUs are a good choice to handle a range of inference tasks. Beyond silicon-based processors, disruptive technologies are emerging to address growing AI compute and data demands. The unicorn start-up Lightmatter, for instance, introduced photonic computing solutions that use light for data transmission to generate significant improvements in speed and energy efficiency. Quantum computing represents another promising area in AI hardware. While still years or even decades away, the integration of quantum computing with AI could further transform fields like drug discovery and genomics. The developments in ML theories and network architectures have significantly enhanced the efficiency and capabilities of AI models. Today, the industry is moving from monolithic models to agent-based systems characterized by smaller, specialized models that work together to complete tasks more efficiently at the edge—on devices like smartphones or modern vehicles. This allows them to extract increased performance gains, like faster model response times, from the same or even less compute. Researchers have developed techniques, including few-shot learning, to train AI models using smaller datasets and fewer training iterations. AI systems can learn new tasks from a limited number of examples to reduce dependency on large datasets and lower energy demands. Optimization techniques like quantization, which lower the memory requirements by selectively reducing precision, are helping reduce model sizes without sacrificing performance.  New system architectures, like retrieval-augmented generation (RAG), have streamlined data access during both training and inference to reduce computational costs and overhead. The DeepSeek R1, an open source LLM, is a compelling example of how more output can be extracted using the same hardware. By applying reinforcement learning techniques in novel ways, R1 has achieved advanced reasoning capabilities while using far fewer computational resources in some contexts. The integration of heterogeneous computing architectures, which combine various processing units like CPUs, GPUs, and specialized accelerators, has further optimized AI model performance. This approach allows for the efficient distribution of workloads across different hardware components to optimize computational throughput and energy efficiency based on the use case. As AI becomes an ambient capability humming in the background of many tasks and workflows, agents are taking charge and making decisions in real-world scenarios. These range from customer support to edge use cases, where multiple agents coordinate and handle localized tasks across devices. With AI increasingly used in daily life, the role of user experiences becomes critical for mass adoption. Features like predictive text in touch keyboards, and adaptive gearboxes in vehicles, offer glimpses of AI as a vital enabler to improve technology interactions for users. Edge processing is also accelerating the diffusion of AI into everyday applications, bringing computational capabilities closer to the source of data generation. Smart cameras, autonomous vehicles, and wearable technology now process information locally to reduce latency and improve efficiency. Advances in CPU design and energy-efficient chips have made it feasible to perform complex AI tasks on devices with limited power resources. This shift toward heterogeneous compute enhances the development of ambient intelligence, where interconnected devices create responsive environments that adapt to user needs. Seamless AI naturally requires common standards, frameworks, and platforms to bring the industry together. Contemporary AI brings new risks. For instance, by adding more complex software and personalized experiences to consumer devices, it expands the attack surface for hackers, requiring stronger security at both the software and silicon levels, including cryptographic safeguards and transforming the trust model of compute environments. More than 70% of respondents to a 2024 DarkTrace survey reported that AI-powered cyber threats significantly impact their organizations, while 60% say their organizations are not adequately prepared to defend against AI-powered attacks. Collaboration is essential to forging common frameworks. Universities contribute foundational research, companies apply findings to develop practical solutions, and governments establish policies for ethical and responsible deployment. Organizations like Anthropic are setting industry standards by introducing frameworks, such as the Model Context Protocol, to unify the way developers connect AI systems with data. Arm is another leader in driving standards-based and open source initiatives, including ecosystem development to accelerate and harmonize the chiplet market, where chips are stacked together through common frameworks and standards. Arm also helps optimize open source AI frameworks and models for inference on the Arm compute platform, without needing customized tuning.  How far AI goes to becoming a general-purpose technology, like electricity or semiconductors, is being shaped by technical decisions taken today. Hardware-agnostic platforms, standards-based approaches, and continued incremental improvements to critical workhorses like CPUs, all help deliver the promise of AI as a seamless and silent capability for individuals and businesses alike. Open source contributions are also helpful in allowing a broader range of stakeholders to participate in AI advances. By sharing tools and knowledge, the community can cultivate innovation and help ensure that the benefits of AI are accessible to everyone, everywhere. Learn more about Arm’s approach to enabling AI everywhere. This content was produced by Insights, the custom content arm of MIT Technology Review. It was not written by MIT Technology Review’s editorial staff. This content was researched, designed, and written entirely by human writers, editors, analysts, and illustrators. This includes the writing of surveys and collection of data for surveys. AI tools that may have been used were limited to secondary production processes that passed thorough human review.

From large language models (LLMs) to reasoning agents, today’s AI tools bring unprecedented computational demands. Trillion-parameter models, workloads running on-device, and swarms of agents collaborating to complete tasks all require a new paradigm of computing to become truly seamless and ubiquitous. Silicon’s mid-life crisis AI has evolved from classical ML to deep learning to generative AI. The most recent chapter, which took AI mainstream, hinges on two phases—training and inference—that are data and energy-intensive in terms of computation, data movement, and cooling. At the same time, Moore’s Law, which determines that the number of transistors on a chip doubles every two years, is reaching a physical and economic plateau. For the last 40 years, silicon chips and digital technology have nudged each other forward—every step ahead in processing capability frees the imagination of innovators to envision new products, which require yet more power to run. That is happening at light speed in the AI age. As models become more readily available, deployment at scale puts the spotlight on inference and the application of trained models for everyday use cases. This transition requires the appropriate hardware to handle inference tasks efficiently. Central processing units (CPUs) have managed general computing tasks for decades, but the broad adoption of ML introduced computational demands that stretched the capabilities of traditional CPUs. This has led to the adoption of graphics processing units (GPUs) and other accelerator chips for training complex neural networks, due to their parallel execution capabilities and high memory bandwidth that allow large-scale mathematical operations to be processed efficiently. But CPUs are already the most widely deployed and can be companions to processors like GPUs and tensor processing units (TPUs). AI developers are also hesitant to adapt software to fit specialized or bespoke hardware, and they favor the consistency and ubiquity of CPUs. Chip designers are unlocking performance gains through optimized software tooling, adding novel processing features and data types specifically to serve ML workloads, integrating specialized units and accelerators, and advancing silicon chip innovations, including custom silicon. AI itself is a helpful aid for chip design, creating a positive feedback loop in which AI helps optimize the chips that it needs to run. These enhancements and strong software support mean modern CPUs are a good choice to handle a range of inference tasks. Beyond silicon-based processors, disruptive technologies are emerging to address growing AI compute and data demands. The unicorn start-up Lightmatter, for instance, introduced photonic computing solutions that use light for data transmission to generate significant improvements in speed and energy efficiency. Quantum computing represents another promising area in AI hardware. While still years or even decades away, the integration of quantum computing with AI could further transform fields like drug discovery and genomics. The developments in ML theories and network architectures have significantly enhanced the efficiency and capabilities of AI models. Today, the industry is moving from monolithic models to agent-based systems characterized by smaller, specialized models that work together to complete tasks more efficiently at the edge—on devices like smartphones or modern vehicles. This allows them to extract increased performance gains, like faster model response times, from the same or even less compute. Researchers have developed techniques, including few-shot learning, to train AI models using smaller datasets and fewer training iterations. AI systems can learn new tasks from a limited number of examples to reduce dependency on large datasets and lower energy demands. Optimization techniques like quantization, which lower the memory requirements by selectively reducing precision, are helping reduce model sizes without sacrificing performance.  New system architectures, like retrieval-augmented generation (RAG), have streamlined data access during both training and inference to reduce computational costs and overhead. The DeepSeek R1, an open source LLM, is a compelling example of how more output can be extracted using the same hardware. By applying reinforcement learning techniques in novel ways, R1 has achieved advanced reasoning capabilities while using far fewer computational resources in some contexts. The integration of heterogeneous computing architectures, which combine various processing units like CPUs, GPUs, and specialized accelerators, has further optimized AI model performance. This approach allows for the efficient distribution of workloads across different hardware components to optimize computational throughput and energy efficiency based on the use case. As AI becomes an ambient capability humming in the background of many tasks and workflows, agents are taking charge and making decisions in real-world scenarios. These range from customer support to edge use cases, where multiple agents coordinate and handle localized tasks across devices. With AI increasingly used in daily life, the role of user experiences becomes critical for mass adoption. Features like predictive text in touch keyboards, and adaptive gearboxes in vehicles, offer glimpses of AI as a vital enabler to improve technology interactions for users. Edge processing is also accelerating the diffusion of AI into everyday applications, bringing computational capabilities closer to the source of data generation. Smart cameras, autonomous vehicles, and wearable technology now process information locally to reduce latency and improve efficiency. Advances in CPU design and energy-efficient chips have made it feasible to perform complex AI tasks on devices with limited power resources. This shift toward heterogeneous compute enhances the development of ambient intelligence, where interconnected devices create responsive environments that adapt to user needs. Seamless AI naturally requires common standards, frameworks, and platforms to bring the industry together. Contemporary AI brings new risks. For instance, by adding more complex software and personalized experiences to consumer devices, it expands the attack surface for hackers, requiring stronger security at both the software and silicon levels, including cryptographic safeguards and transforming the trust model of compute environments. More than 70% of respondents to a 2024 DarkTrace survey reported that AI-powered cyber threats significantly impact their organizations, while 60% say their organizations are not adequately prepared to defend against AI-powered attacks. Collaboration is essential to forging common frameworks. Universities contribute foundational research, companies apply findings to develop practical solutions, and governments establish policies for ethical and responsible deployment. Organizations like Anthropic are setting industry standards by introducing frameworks, such as the Model Context Protocol, to unify the way developers connect AI systems with data. Arm is another leader in driving standards-based and open source initiatives, including ecosystem development to accelerate and harmonize the chiplet market, where chips are stacked together through common frameworks and standards. Arm also helps optimize open source AI frameworks and models for inference on the Arm compute platform, without needing customized tuning.  How far AI goes to becoming a general-purpose technology, like electricity or semiconductors, is being shaped by technical decisions taken today. Hardware-agnostic platforms, standards-based approaches, and continued incremental improvements to critical workhorses like CPUs, all help deliver the promise of AI as a seamless and silent capability for individuals and businesses alike. Open source contributions are also helpful in allowing a broader range of stakeholders to participate in AI advances. By sharing tools and knowledge, the community can cultivate innovation and help ensure that the benefits of AI are accessible to everyone, everywhere. Learn more about Arm’s approach to enabling AI everywhere. This content was produced by Insights, the custom content arm of MIT Technology Review. It was not written by MIT Technology Review’s editorial staff. This content was researched, designed, and written entirely by human writers, editors, analysts, and illustrators. This includes the writing of surveys and collection of data for surveys. AI tools that may have been used were limited to secondary production processes that passed thorough human review.

From large language models (LLMs) to reasoning agents, today’s AI tools bring unprecedented computational demands. Trillion-parameter models, workloads running on-device, and swarms of agents collaborating to complete tasks all require a new paradigm of computing to become truly seamless and ubiquitous. First, technical progress in hardware and silicon design is critical to pushing the boundaries of compute. Second, advances in machine learning (ML) allow AI systems to achieve increased efficiency with smaller computational demands. Finally, the integration, orchestration, and adoption of AI into applications, devices, and systems is crucial to delivering tangible impact and value. Silicon’s mid-life crisis AI has evolved from classical ML to deep learning to generative AI. The most recent chapter, which took AI mainstream, hinges on two phases—training and inference—that are data and energy-intensive in terms of computation, data movement, and cooling. At the same time, Moore’s Law, which determines that the number of transistors on a chip doubles every two years, is reaching a physical and economic plateau. For the last 40 years, silicon chips and digital technology have nudged each other forward—every step ahead in processing capability frees the imagination of innovators to envision new products, which require yet more power to run. That is happening at light speed in the AI age. As models become more readily available, deployment at scale puts the spotlight on inference and the application of trained models for everyday use cases. This transition requires the appropriate hardware to handle inference tasks efficiently. Central processing units (CPUs) have managed general computing tasks for decades, but the broad adoption of ML introduced computational demands that stretched the capabilities of traditional CPUs. This has led to the adoption of graphics processing units (GPUs) and other accelerator chips for training complex neural networks, due to their parallel execution capabilities and high memory bandwidth that allow large-scale mathematical operations to be processed efficiently. But CPUs are already the most widely deployed and can be companions to processors like GPUs and tensor processing units (TPUs). AI developers are also hesitant to adapt software to fit specialized or bespoke hardware, and they favor the consistency and ubiquity of CPUs. Chip designers are unlocking performance gains through optimized software tooling, adding novel processing features and data types specifically to serve ML workloads, integrating specialized units and accelerators, and advancing silicon chip innovations, including custom silicon. AI itself is a helpful aid for chip design, creating a positive feedback loop in which AI helps optimize the chips that it needs to run. These enhancements and strong software support mean modern CPUs are a good choice to handle a range of inference tasks. Beyond silicon-based processors, disruptive technologies are emerging to address growing AI compute and data demands. The unicorn start-up Lightmatter, for instance, introduced photonic computing solutions that use light for data transmission to generate significant improvements in speed and energy efficiency. Quantum computing represents another promising area in AI hardware. While still years or even decades away, the integration of quantum computing with AI could further transform fields like drug discovery and genomics. Understanding models and paradigms The developments in ML theories and network architectures have significantly enhanced the efficiency and capabilities of AI models. Today, the industry is moving from monolithic models to agent-based systems characterized by smaller, specialized models that work together to complete tasks more efficiently at the edge—on devices like smartphones or modern vehicles. This allows them to extract increased performance gains, like faster model response times, from the same or even less compute. Researchers have developed techniques, including few-shot learning, to train AI models using smaller datasets and fewer training iterations. AI systems can learn new tasks from a limited number of examples to reduce dependency on large datasets and lower energy demands. Optimization techniques like quantization, which lower the memory requirements by selectively reducing precision, are helping reduce model sizes without sacrificing performance.  New system architectures, like retrieval-augmented generation (RAG), have streamlined data access during both training and inference to reduce computational costs and overhead. The DeepSeek R1, an open source LLM, is a compelling example of how more output can be extracted using the same hardware. By applying reinforcement learning techniques in novel ways, R1 has achieved advanced reasoning capabilities while using far fewer computational resources in some contexts. The integration of heterogeneous computing architectures, which combine various processing units like CPUs, GPUs, and specialized accelerators, has further optimized AI model performance. This approach allows for the efficient distribution of workloads across different hardware components to optimize computational throughput and energy efficiency based on the use case. Orchestrating AI As AI becomes an ambient capability humming in the background of many tasks and workflows, agents are taking charge and making decisions in real-world scenarios. These range from customer support to edge use cases, where multiple agents coordinate and handle localized tasks across devices. With AI increasingly used in daily life, the role of user experiences becomes critical for mass adoption. Features like predictive text in touch keyboards, and adaptive gearboxes in vehicles, offer glimpses of AI as a vital enabler to improve technology interactions for users. Edge processing is also accelerating the diffusion of AI into everyday applications, bringing computational capabilities closer to the source of data generation. Smart cameras, autonomous vehicles, and wearable technology now process information locally to reduce latency and improve efficiency. Advances in CPU design and energy-efficient chips have made it feasible to perform complex AI tasks on devices with limited power resources. This shift toward heterogeneous compute enhances the development of ambient intelligence, where interconnected devices create responsive environments that adapt to user needs. Seamless AI naturally requires common standards, frameworks, and platforms to bring the industry together. Contemporary AI brings new risks. For instance, by adding more complex software and personalized experiences to consumer devices, it expands the attack surface for hackers, requiring stronger security at both the software and silicon levels, including cryptographic safeguards and transforming the trust model of compute environments. More than 70% of respondents to a 2024 DarkTrace survey reported that AI-powered cyber threats significantly impact their organizations, while 60% say their organizations are not adequately prepared to defend against AI-powered attacks. Collaboration is essential to forging common frameworks. Universities contribute foundational research, companies apply findings to develop practical solutions, and governments establish policies for ethical and responsible deployment. Organizations like Anthropic are setting industry standards by introducing frameworks, such as the Model Context Protocol, to unify the way developers connect AI systems with data. Arm is another leader in driving standards-based and open source initiatives, including ecosystem development to accelerate and harmonize the chiplet market, where chips are stacked together through common frameworks and standards. Arm also helps optimize open source AI frameworks and models for inference on the Arm compute platform, without needing customized tuning.  How far AI goes to becoming a general-purpose technology, like electricity or semiconductors, is being shaped by technical decisions taken today. Hardware-agnostic platforms, standards-based approaches, and continued incremental improvements to critical workhorses like CPUs, all help deliver the promise of AI as a seamless and silent capability for individuals and businesses alike. Open source contributions are also helpful in allowing a broader range of stakeholders to participate in AI advances. By sharing tools and knowledge, the community can cultivate innovation and help ensure that the benefits of AI are accessible to everyone, everywhere. Learn more about Arm’s approach to enabling AI everywhere. This content was produced by Insights, the custom content arm of MIT Technology Review. It was not written by MIT Technology Review’s editorial staff. This content was researched, designed, and written entirely by human writers, editors, analysts, and illustrators. This includes the writing of surveys and collection of data for surveys. AI tools that may have been used were limited to secondary production processes that passed thorough human review.

The US has ordered companies that make software used to design semiconductors to stop selling to China without first obtaining export licenses. The restrictions go beyond software alone, covering chemicals for semiconductors, butane and ethane, machine tools, and aviation equipment, Reuters reported, citing two people familiar with the development. “On May 23, the US Government informed the Electronic Design Automation (EDA) industry about new export controls on EDA software to China and Chinese military end users globally,” said a Siemens EDA spokesperson. “Siemens has supported customers in China for more than 150 years and will continue to work with our customers globally to mitigate the impact of these new restrictions while operating in compliance with applicable national export control regimes. The company continues to support our employees and customers around the world who are using our technology to transform the everyday.” This represents the latest chapter in a tech war that began with restrictions on selling actual semiconductors to China. Now, the US is targeting the tools needed to design those chips — a potentially more damaging approach. Strategic shift to upstream controls Electronic design automation software makers — including industry leaders Cadence, Synopsys, and Siemens EDA — were sent notifications by the Commerce Department last Friday to cease supplying their technology to Chinese customers, the report said. The department will review license requests on a case-by-case basis, it added. The financial implications are substantial. Synopsys and Cadence earn annual revenue of about 16% and 12% from their China business. “With Cadence and Synopsys being US-based companies and Siemens contributing to more than 90% share of the EDA tools globally, this move further tightens EDA software sales in China,” said Neil Shah, VP for research and partner at Counterpoint Research. “EDA tools cannot be substituted and are the foundation to chip design and manufacturing.” What makes this strategically different is its upstream focus. Manish Rawat, semiconductor analyst at TechInsights, explained that, unlike previous hardware restrictions, “the new focus on EDA software targets the critical tools essential for designing advanced chips (5nm and below). This upstream control aims to block innovation before chips are manufactured, making it a more preemptive and disruptive tactic.” Why now? The timing reflects broader strategic recalibration. Rawat noted that “the US has shifted its strategy, now seeing China’s push for tech self-sufficiency — especially in AI and semiconductors — as a growing national security threat.” Since the 2020 CHIPS Act, coordinated export controls with allies like Japan and the Netherlands have strengthened US resolve. Sanchit Vir Gogia, chief analyst and CEO at Greyhound Research, observed that targeting design-phase technologies “seeks to constrain the conceptual stage of advanced chip development — not merely production.” The timing may also serve as “a strategic bargaining tool amid paused tariffs and ongoing diplomacy,” Rawat suggested, signaling US willingness to escalate tech restrictions to strengthen its negotiating position. The EDA software packages from companies like Synopsys and Cadence are central to modeling, simulation, and verification of complex semiconductor architectures. “The software lifecycle of these tools is super important with updates, patches, and support to be at the forefront of leading edge, which will stop with the restrictions on licensing,” Shah pointed out. This ongoing dependency means even alternative tools would struggle to keep pace with rapidly evolving chip design requirements without continuous vendor support. China’s long road to independence For China, developing viable alternatives presents enormous challenges. While Chinese companies like Empyrean, Primarius, and Entasys have emerged as domestic providers, they remain far behind. “Developing advanced EDA software on par with Synopsys or Cadence is highly complex, requiring decades of R&D,” Rawat explained. “Fully closing the gap — especially for cutting-edge sub-7nm chip design — could take 5 to 10 years or more.” Gogia added that “while notable progress has been made in selected areas of analog and layout tooling, full-stack integration across simulation, IP compatibility, and foundry certification continues to lag.” The gap is widening. Shah noted that Cadence recently announced M2000 Supercomputers, integrating advanced AI into EDA workflows. “This widens the gap between what China can build with an indigenous toolchain, as these US companies are miles ahead.” However, China may have breathing room. “China has been relegated to access to advanced process nodes, so in the near to mid-term, they might not need an advanced toolchain as they won’t be able to design or manufacture advanced chips,” Shah observed. Beijing’s likely response China’s response will likely be multifaceted. “Beijing is likely to accelerate funding through increased subsidies and incentives for domestic EDA startups,” Rawat said. “It will also aggressively recruit global experts and repatriate Chinese talent with semiconductor software expertise.” Beyond domestic development, “China may build alternative chip design ecosystems less reliant on US intellectual property, though these will initially lag in sophistication,” Rawat added. Diplomatic measures may include reciprocal restrictions on US firms or supply chains involving Chinese technology. Toward a bifurcated design world The restrictions are accelerating what analysts see as an inevitable split. Gogia described emerging “parallel EDA stacks” where “global design ecosystems may begin to diverge, with export controls catalyzing separate compliance frameworks and IP governance models.” “This is accelerating a split into two spheres: a US-led system using Western tools and IP protections, and a China-led system focused on domestic tools and foundries,” Rawat added. This separation isn’t just technical — it’s institutional. “Engineering workflows, legal oversight, cloud infrastructure, and partner ecosystems are all being restructured to manage compliance in a fractured regulatory environment,” Gogia said. Global industry implications For multinational companies, this fragmentation creates significant challenges. “Multinational firms may need to adopt dual design workflows and navigate stricter compliance, affecting partnerships and operational efficiency,” Rawat said. Organizations face maintaining duplicate systems and complex compliance across jurisdictions. Smaller firms may find duplication costs force market exits or a narrowed geographic focus. To mitigate risks, companies “are likely to diversify supply chains and expand in neutral regions like India, Vietnam, and Singapore, emerging as new semiconductor design hubs,” Rawat pointed out. The EDA software restrictions represent the latest evolution in US-China tech competition, moving from end-product controls to fundamental design capabilities. “US continues to find stranglehold on China with critical software and hardware to cut off access to critical and advanced tools,” Shah said. For enterprise technology leaders, this signals an era where geopolitical considerations increasingly shape technology architecture decisions, requiring strategic planning for an increasingly fragmented world. Cadence and Synopsys did not respond to requests for comment by publication time.

Tencent and Baidu said stockpiling chips, optimizing AI models and using home-grown semiconductors have helped them progress with the tech.

NVIDIA's next big rival is not another Big Tech company from the West but Huawei, the firm on track to dominate the AI segment in China. Huawei Could Very Well Make NVIDIA Irrelevant In The Chinese AI Markets, Marking a Massive Blow To The Company's Business When it comes to acknowledging competition, Team Green has done little despite being in the market for more than three decades, where you have rivals like Intel or AMD. Under NVIDIA's CEO Jensen Huang, the firm has managed to dominate its business segment for several years now, but now there's an "undeniable" panic within Jensen's camp, which is mainly due to how big of a rival has Huawei emerged itself to be, to a point where NVIDIA now feels threatened by the company. When looking at Chinese markets alone, Huawei dominates every space, including the AI segment. One of the reasons why Huawei has become such a significant factor for NVIDIA is that the US restrictions have forced Team Green to distance itself from the Chinese AI markets, which has given Huawei a massive opportunity, and the firm has capitalized on this. Based on what we know until now, the Chinese tech giant offers its Ascend 910B and 910C chips for the region, which are said to be equivalent in performance relative to the likes of NVIDIA's H100 AI accelerators, which are some of the most high-end chips out there. When you look at NVIDIA's position, the company is forced to sell its stripped-down solutions to the Chinese markets, and Team Green's upcoming AI chip is said to further bridge the performance gap by integrating relatively weaker technologies such as GDDR7. In terms of raw power, NVIDIA's next chip might not match what Huawei offers, which could be a massive blow to the company. However, Jensen will likely rely on his company's software ecosystem in order to be competent in domestic markets, but we know Huawei is fond of developing capable alternatives. Huawei isn't just building up AI chips; it has managed to vertically integrate the supply chain, from semiconductors to memory production. The Chinese giant is said to operate multiple fabs under different subsidiaries, and is claimed to have achieved persistent 7nm production. So NVIDIA isn't just rivaling a tech company here, it is in competition with one of China's largest business entities in terms of production, hence becoming obsolete isn't too far away for NVIDIA in China. NVIDIA's urgency to maintain its position in China is evident from Jensen's recent statements about US policies and how Team Green has been affected by the AI Diffusion policy. The company's future in China is definitely uncertain, and even after desperate attempts, NVIDIA will have a tough time maintaining its position. Deal of the Day

President Donald Trump has disrupted global trade and roiled markets in an effort to bring manufacturing jobs back to the U.S. Some of his top tech allies, however, have backed ventures that replace human workers with robots.Elon Musk, a top donor and adviser to Trump, has touted humanoid robots as a future growth area for electric-carmaker Tesla. "You can produce any product,” Musk said of the robots’ potential capacity during a February interview with Dubai's World Governments Summit.Amazon founder Jeff Bezos, who Trump last month called “terrific,” has invested in several advanced robotics firms.Bezos last year poured funds into Figure, a humanoid robot company that says its initial rollout will focus on manufacturers and warehouses, among other business applications. “We believe humanoids will revolutionize a variety of industries,” the company says on its website.Nvidia CEO Jensen Huang and OpenAI CEO Sam Altman – both of whom joined Trump on his recent trip to the Middle East – helmed their respective companies as each invested in Figure. OpenAI ended its partnership with Figure last year.“Trump is talking about bringing back the jobs, and he’s not understanding the tension between that goal and automation, which the tech bros have enthusiasm for,” Harry Holzer, a professor of public policy at Georgetown University and a former chief economist at the U.S. Department of Labor, told ABC News. “There’s a fundamental conflict between those goals.”MORE: Trump wants Apple to make iPhones in the US. Will it ever happen?Musk did not immediately respond to ABC News’ request for comment made through Musk-owned firm SpaceX. Neither Bezos, Huang nor Altman responded to ABC News’ request.Speaking at a conference in April, Huang said the onset of artificial intelligence would fuel "new types of factories," which in turn would create jobs in construction and steelmaking, as well as in trades such as plumbing and electricity.Even more, Huang said, AI is set to trigger a surge in productivity at companies that adopt the new technology, allowing them to add employees as the firms increase output and revenue."New jobs will be created, some jobs will be lost, every job will be changed," Huang said. "Remember, it's not AI that's going to take your job. It's not AI that's going to destroy your company. It's the company and the person who uses AI that's going to take your job. And so that's something to internalize."Even after a rollback of some levies, consumers face the highest overall average effective tariff rate since 1934, the Yale Budget Lab found earlier this month.A key reason for the tariffs, White House officials say: Reshoring factories and rejuvenating employment in the manufacturing industry.Commerce Secretary Howard Lutnick said this month in an interview with Fox News that Trump's vision for ushering in a "golden age" for America involved enticing manufacturers to open factories and build in the United States."We're going to have huge jobs in manufacturing. You've heard the president talk about trillions and trillions of factories being built in America," he said in the interview on May 11.In response to ABC News' request for comment, White House Spokesperson Kush Desai said "the importance of President Trump’s push to reinvigorate American industry goes beyond creating good-paying jobs for everyday Americans.""Supply chain shocks of critical pharmaceuticals, medical equipment, and semiconductors during the COVID era prove that America cannot rely on foreign imports. The Trump administration remains committed to reshoring manufacturing that’s critical to our national and economic security with a multifaceted approach of tariffs, tax cuts, rapid deregulation, and domestic energy production," Desai added.The share of U.S. workers in manufacturing has plummeted for decades. Roughly 8% of U.S. workers currently hold positions in manufacturing, which marks a steep decline from about a quarter of all employees as recently as 1970.Researchers attribute such decline to overlapping trends, including the offshoring of manufacturing to low-wage markets overseas and the adoption of labor-saving technology throughout the sector.Long before current advances, automation significantly increased productivity in U.S. factories, meaning the same number of workers could produce many more goods, researchers at Ball State University found in 2015. As a result, they said, manufacturing employment stagnated for decades even as output climbed.Popular Reads“Automation is something we’ve seen for a long time,” Philipp Kircher, a professor of industrial and labor relations at Cornell University, told ABC News.CEO of Meta and Facebook Mark Zuckerberg, Lauren Sanchez, Amazon founder Jeff Bezos, Google CEO Sundar Pichai and Tesla and SpaceX CEO Elon Musk attend the inauguration ceremony of Donald Trump in Washington, D.C., on Jan. 20, 2025.Julia Demaree Nikhinson, Pool via AFP via Getty Images, FILESome of Trump’s tech allies have backed firms that seek to further automate manufacturing, touting a new wave of artificial-intelligence equipped robots as a replacement for some workers and salve for labor shortages.Robotics outfit Vicarious boasts $250 million in investments from a set of backers that includes Bezos, Musk and Meta CEO Mark Zuckerberg – all of whom flanked Trump during his inauguration.On a webpage displaying photos of robots for use in warehouse settings, Vicarious tells potential clients that the products can “reduce both your costs and person-hour needs.”In 2022, Vicarious was acquired by Alphabet-backed robotics software firm Intrinsic. Alphabet CEO Sundar Pichai also sat alongside tech leaders at Trump’s inauguration.Alphabet did not respond to ABC News' request for comment. Meta declined to comment.Yong Suk Lee, a professor of economics and technology at the University of Notre Dame, described the views on automation among Trump’s tech allies and some of his trade advisers as “opposed.”The tech position, Lee said, would likely win out, even if some firms do open plants in the U.S.“If you want to reshore, are you going to pay the same wages as Vietnam? Probably not,” Lee said. “Companies are faced with higher labor costs. In that case, they’ll probably automate.”Discordant views among some tech leaders and White House officials surfaced in April, when Musk sharply criticized tariff-advocate Peter Navarro, Trump’s senior counselor for trade and manufacturing. Navarro, Musk said, is “truly a moron.”In an interview with CNBC, Navarro responded, saying Musk "isn't a car manufacturer — he's a car assembler.”MORE: Target CEO says tariffs risk 'massive' costs, but price hikes a 'last resort'To be sure, analysts said, automation in manufacturing would likely continue regardless of support from Trump’s tech allies, since producers are locked in a competition to lower costs and increase output. The precise outlook for manufacturing employment is unclear, they added, since additional technology may add jobs for those maintaining and optimizing the machinery.“Whether it’s the companies that currently support the U.S. president or not, somebody would be doing this innovation, maybe slightly slower,” Kircher said.Even at current employment levels, a labor shortage bedevils U.S. manufacturers. Roughly one of every five U.S. factories that failed to produce at full capacity cited a shortage of workers, Jason Miller, a professor of supply chain management at Michigan State University, found in a January study analyzing government data.Agility Robots, an Amazon-backed firm building humanoid robots, identifies the current push for rejuvenated U.S. manufacturing as an opportunity for greater adoption of technology.“Manufacturing companies are seeing a massive reshoring movement spanning various industries,” Agility Robots says on its website. “Adding a humanoid robot to your manufacturing facility is a great way to stay on the leading edge of automation.”In response to ABC News' request for comment, an Amazon spokesperson pointed to previous remarks about robotics made by a company executive."Our goal is to ensure these systems improve safety and productivity. Technology should be used to help us retain and grow our talent through skill development and reimagining how we make our workplace better, both in productivity and safety. If we do this well, we’re certain to always innovate for our customers," Tye Brady, chief technologist at Amazon Robotics, said in a September blog post.Amazon has "created more U.S. jobs in the last decade than any other company," Amazon said this month.