Semiconductors

Jetson AGX Thor brings Blackwell-powered compute to robots and autonomous vehicles

Nvidia has introduced Jetson AGX Thor, its Blackwell-powered robotics platform that succeeds the 2022 Jetson Orin. Designed for autonomous driving, factory robots, and humanoid machines, it comes in multiple models, with a DRIVE OS kit for vehicles scheduled for release in September.

Thor delivers 7.5 times more AI compute, 3.1 times greater CPU performance, and double the memory of Orin. The flagship Thor T5000 offers up to 2,070 teraflops of AI compute, paired with 128 GB of memory, enabling the execution of generative AI models and robotics workloads at the edge.

The platform supports Nvidia’s Isaac, Metropolis, and Holoscan systems, and features multi-instance GPU capabilities that enable the simultaneous execution of multiple AI models. It is compatible with Hugging Face, PyTorch, and leading AI models from OpenAI, Google, and other sources.

Adoption has begun, with Boston Dynamics utilising Thor for Atlas and firms such as Volvo, Aurora, and Gatik deploying DRIVE AGX Thor in their vehicles. Nvidia stresses it supports robot-makers rather than building robots, with robotics still a small but growing part of its business.

Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!

NVIDIA eyes recovery in China after export deal ahead of Q2 report

NVIDIA is due to report its Q2 2026 financial results after the US market closes on 27 August, and analysts are expecting strong performance.

Consensus forecasts place revenue at around US $45.9 billion, up about 50 percent year-on-year, driven by ongoing demand for Blackwell GPUs, data centre expansion and redistribution of AI infrastructure investments globally.

Export changes are also pivotal. After entering a deal to resume H20 chip sales to China, despite revenue-sharing conditions, NVIDIA could reclaim as much as US$8 billion during Q2, mitigating past losses caused by restrictions.

Beyond geopolitical shifts, the Blackwell Ultra GPU is central to growth. Offering up to 50 times faster AI inference than earlier models, it is increasingly stocked by cloud providers and hyperscalers. Markets view this as a strategic advantage, fueling long-term AI momentum.

Risks remain. Gross margins may recover from prior pressure due to licensing charges, but margin expansion depends on supply and TAM realisation. China’s policy environment is also uncertain, making future guidance cautious for some analysts.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

Arm to build its own chips with AI focus

Arm Holdings has hired Amazon’s AI chip lead, Rami Sinno, to design its own complete chips. Known for Amazon’s Trainium and Inferentia processors, Sinno brings key expertise to Arm’s new direction in chip manufacturing.

Arm has traditionally licensed chip designs to companies like Apple and Nvidia, but now aims to build chips and complete systems. The firm is expanding teams with experience from HPE, Intel and Qualcomm, signalling a significant shift in its business model.

Backed by SoftBank, Arm plans to invest profits in chip development to rival Nvidia and reduce reliance on traditional licensing.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

Microsoft and AMD develop new gaming chips

Microsoft plans to equip its next-generation Xbox console with AI-focused hardware, including a dedicated neural processing unit.

Vice President Jason Ronald confirmed that the company is working with AMD to develop chips for gaming consoles, PCs and cloud platforms.

New AI capabilities are expected to transform gameplay and provide developers with tools to create immersive, previously unattainable experiences.

Microsoft’s experimental Xbox Ally X device, developed with ASUS, is already used to test AI integration in real-world scenarios.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

Sam Altman urges rethink of US–China AI strategy

OpenAI CEO Sam Altman has warned that the United States may be underestimating China’s rapid advances in AI. He argued that export controls on semiconductors are unlikely to be a reliable long-term solution to the global AI race.

At a press briefing in San Francisco, Altman said the competition cannot be reduced to a simple scoreboard. China can expand inference capacity more quickly, even as Washington tightens restrictions on advanced semiconductor exports.

He expressed doubts about the effectiveness of purely policy-driven approaches. ‘You can export-control one thing, but maybe not the right thing… workarounds exist,’ Altman said. He stressed that chip controls may not keep pace with technological realities.

His comments come as US policy becomes increasingly complex. President Trump halted advanced chip supplies in April, while the Biden administration recently allowed ‘China-safe’ chips, requiring Nvidia and AMD to share revenue. Critics call the rules contradictory and difficult to enforce.

Meanwhile, Chinese firms are accelerating efforts to replace US suppliers, with Huawei and others building domestic alternatives. Altman suggested this push for self-sufficiency could undermine Washington’s goals, raising questions about America’s strategy in the AI race.

Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!

Nvidia prepares new AI chip for China amid Washington’s hesitation

As if Trump’s recent shifts in chip export policy regarding the scaled-down chip models were not enough to reopen supply to the Chinese market, after all the earlier tariffs and bans, Nvidia is now quietly developing a new AI chip for China, even as Washington continues to debate how much cutting-edge US technology Beijing should be allowed to access.

According to Nvidia’s latest statements, the chip, codenamed B30A, will be based on Nvidia’s latest Blackwell architecture and is expected to outperform the company’s current China-approved model, the H20.

Namely, the novelty comes just days after President Donald Trump weighed permitting scaled-down versions of Nvidia’s most advanced chips to be sold in China. His comments marked a potential shift in US policy, but the approval remains uncertain, with lawmakers in both parties warning that even weaker versions of top-end chips could still give Beijing an edge in the global AI race.

Technically, the B30A will be less potent than Nvidia’s flagship B300, but it retains advanced features such as high-bandwidth memory and NVLink connectivity, which are crucial for fast data processing.

Nvidia hopes to send early samples to Chinese customers next month, though final specifications have yet to be confirmed.

‘Everything we offer is with full government approval and designed for commercial use,’ the company said in a statement.

The stakes are high, as China accounted for 13% of Nvidia’s revenue last year, and losing that market could push customers toward domestic rivals like Huawei.

Analysts note that Huawei’s chips are improving, particularly in raw computing power, though they still lag in software support and memory performance, areas where Nvidia remains dominant.

At the same time, Beijing has been pushing back. Chinese experts recently raised concerns that Nvidia’s chips could pose security risks, and regulators have reportedly warned Chinese tech firms about buying the H20.

Nvidia denies any such vulnerabilities, but the warnings illustrate how political friction is weighing on commercial strategy.

Alongside the B30A, Nvidia is also preparing another chip, the RTX6000D, built for AI inference rather than training. That model has weaker specifications designed to comply with strict US export thresholds.

Nvidia plans to start shipping small batches of the RTX6000D to Chinese clients as early as September, which seems to indicate that the company is trying to balance Washington’s restrictions with the need to preserve its foothold in one of the world’s most lucrative AI markets.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

SoftBank invests $2 billion in Intel to boost US semiconductor industry

Japanese technology giant SoftBank has announced plans to buy a $2 billion stake in Intel, signalling a stronger push into the American semiconductor industry.

The investment comes as Washington debates greater government involvement in the sector, with reports suggesting President Donald Trump is weighing a US government stake in the chipmaker.

SoftBank will purchase Intel’s common stock at $23 per share. Its chairman, Masayoshi Son, said semiconductors remain the backbone of every industry and expressed confidence that advanced chip manufacturing will expand in the US, with Intel playing a central role.

The move follows SoftBank’s increasing investments in the US, including its role in the $500 billion ‘Stargate’ AI project announced earlier this year.

Once a dominant force in Silicon Valley, Intel has struggled against rivals such as Nvidia and AMD. Under new CEO Lip-Bu Tan, the company is cutting 15% of its workforce and reducing costs to stabilise operations.

After a private meeting, Trump recently criticised Tan’s leadership but later softened his stance.

Shares in both companies slipped following the announcement, with SoftBank down 2.2% in Tokyo and Intel falling 3.7% in New York.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

US may take stake in Intel to boost chip production

The US government is reportedly considering acquiring a stake in Intel to support its domestic chip manufacturing plans. Talks began after Intel CEO Lip-Bu Tan met with Trump administration officials on 11 August, following calls for his resignation over alleged China ties.

President Trump has pushed for greater control over the semiconductor sector and recently criticised Tan, prompting political pressure on Intel’s board.

While Intel declined to comment on a possible deal, it stated its commitment to supporting US technology and manufacturing leadership.

The proposed stake would aid Intel’s delayed Ohio chip factory project and expand its US production capacity.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

DeepSeek delays next AI model amid Huawei chip challenges

Chinese AI company DeepSeek has postponed the launch of its R2 model after repeated technical problems using Huawei’s Ascend processors for training. The delay highlights Beijing’s ongoing struggle to replace US-made chips with domestic alternatives.

Authorities had encouraged DeepSeek to shift from Nvidia hardware to Huawei’s chips after the release of its R1 model in January. However, training failures, slower inter-chip connections, stability issues, and weaker software performance led the start-up to revert to Nvidia chips for training, while continuing to explore Ascend for inference tasks.

Despite Huawei deploying engineers to assist on-site, DeepSeek was unable to complete a successful training run using Ascend processors. The company is also contending with extended data-labelling timelines for its updated model, adding to the delays.

The situation underscores how far Chinese chip technology lags behind Nvidia for advanced AI development, even as Beijing pressures domestic firms to use local products. Industry observers say Huawei is facing “growing pains” but could close the gap over time. Meanwhile, competitors like Alibaba’s Qwen3 have integrated elements of DeepSeek’s design more efficiently, intensifying market pressure.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

South Korea unveils megagrowth plan with AI expressway and energy reform

President Lee Jae-myung has announced a sweeping national megagrowth plan that positions South Korea at the forefront of AI and energy transformation.

The initiative includes the creation of an ‘AI expressway’, starting with the Ulsan AI data centre, underpinned by bold tax incentives and regulatory reforms to attract private sector investment. Complementing this is a proposed investment of 100 trillion won to accelerate AI innovation, next-generation semiconductors, and the development of AI infrastructure and innovation zones.

On the energy front, the government has launched a dedicated task force to develop an AI-powered next-generation power grid. This ‘electric highway’ aims to integrate AI technology into renewable energy distribution and grid modernisation without needing vast new infrastructure.

Complementing the power grid overhaul, Korea Electric Power Corp. (KEPCO) plans to invest around 73 trillion won by 2038 to expand transmission lines and upgrade the power infrastructure serving major semiconductor complexes.

Together, these measures form a robust blueprint that aligns digital transformation with energy security, aimed at keeping South Korea globally competitive while responding to rising electricity demands from AI and other tech industries.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!