Tether has launched an AI framework that runs large language models on smartphones and non-NVIDIA GPUs. The system is part of its QVAC platform and uses Microsoft’s BitNet architecture, along with LoRA techniques to reduce memory and computational requirements.
The framework enables cross-platform training on AMD, Intel, Apple Silicon, and mobile GPUs, allowing models with up to 1 billion parameters to be fine-tuned on phones in under 2 hours.
Larger models with up to 13 billion parameters are also supported on mobile devices. BitNet’s 1-bit architecture reduces VRAM requirements by nearly 78%, enabling larger models to run on limited hardware.
Performance improvements benefit inference, with mobile GPUs outperforming CPUs, enabling on-device training and federated learning. By reducing reliance on cloud infrastructure, the system offers more flexible AI development for distributed environments.
Tether’s expansion into AI mirrors a broader trend in the crypto sector, where companies are investing in AI infrastructure, autonomous agents, and high-performance computing.
Industry activity includes record revenue growth for AI and HPC operations, blockchain-integrated AI agents, and new tools for secure on-chain transactions.
Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!
Chey Tae-won warned that the global memory chip shortage could last for years, with structural supply constraints likely to continue into the next decade. Speaking on the sidelines of Nvidia GTC 2026 in San Jose, he said limited wafer capacity remains a key bottleneck for the semiconductor industry.
‘The shortage stems from a lack of wafer capacity, and securing additional wafers takes at least four to five years,’ Chey said. ‘We expect the industry-wide supply shortfall to persist at over 20 percent through 2030.’
He added that SK Hynix is implementing initiatives such as adjusting production schedules and diversifying supplier partnerships to stabilise prices. CEO Kwak Noh-jung is expected to provide further details on these new steps to manage volatility linked to the memory chip shortage.
Despite growing pressure to expand manufacturing overseas, Chey stressed that the group will prioritise domestic production to better respond to the ongoing memory chip shortage. ‘Building capacity outside Korea takes the same amount of time, regardless of location,’ he said. ‘Korea already has the infrastructure in place, allowing for a much faster response.’
He also highlighted the challenges of building fabrication plants abroad, including the need for reliable electricity and water supplies, as well as access to skilled engineering talent.
On competition in the high-bandwidth memory market, Chey noted that rising demand driven by artificial intelligence is reshaping supply dynamics. ‘AI requires graphics processing units (GPUs), and GPUs require HBM. We will do our best,’ he said, while cautioning that excessive focus on HBM could worsen the memory chip shortage for conventional DRAM used in smartphones and personal computers.
Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!
The European Commission has delayed a flagship tech sovereignty package for the second time, according to its latest College agenda. The measures are now scheduled for adoption on 27 May, after previously being postponed from March to April.
The tech sovereignty package includes several major initiatives aimed at strengthening EU tech sovereignty, such as the Cloud and AI Development Act, the Chips Act 2, an open-source strategy, and a roadmap for digitalisation and AI in energy. European Commission officials have not provided a reason for the latest delay.
The Cloud and AI Development Act is expected to define what constitutes a ‘sovereign’ cloud and simplify rules for building data centres. The proposal is designed to accelerate infrastructure development as Europe seeks to compete in the global AI race.
Chips Act 2 will follow up on the EU’s earlier semiconductor strategy, which struggled to boost domestic chip production significantly. The new proposal is expected to refine industrial policy efforts to reduce reliance on foreign suppliers.
Meanwhile, the planned open source strategy aims to support European software ecosystems and reduce dependence on large US technology firms. By encouraging commercially viable open source projects, the EU hopes to strengthen its long-term digital autonomy.
Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!
AI agents are rapidly gaining traction, raising questions about whether existing EU rules can keep pace. Unlike chatbots, these systems can act autonomously and interact with digital tools on behalf of users.
Experts warn that AI agents require deeper access to personal data and online services to function effectively. Regulators in Europe are monitoring potential risks as the technology becomes more integrated into daily life.
Lawmakers are examining whether current legislation, such as the AI Act and GDPR, adequately covers agent-based systems. Legal experts highlight challenges around contracts, liability and accountability when AI acts independently.
Despite concerns, many governments remain reluctant to introduce new rules, citing regulatory fatigue. Policymakers may rely on existing frameworks unless major incidents force a reassessment of AI oversight.
Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!
Legal pressure is increasing on OpenAI as Encyclopaedia Britannica and Merriam-Webster file a lawsuit accusing the company of large-scale copyright violations.
According to the complaint, nearly 100,000 copyrighted articles were allegedly used without authorisation to train large language models. Publishers also argue that AI-generated outputs can reproduce parts of their content, raising concerns about unauthorised distribution.
Additional claims focus on how AI systems retrieve and present information. The lawsuit argues that retrieval-augmented generation tools may rely on proprietary databases, potentially undermining publishers’ business models by reducing traffic to original sources.
Concerns are also raised about inaccurate outputs attributed to publishers, which could affect trust in established information providers. The case highlights ongoing tensions between AI development and intellectual property protections.
Growing legal disputes involving media organisations, including The New York Times, suggest that courts will play a key role in defining how copyrighted material can be used in AI training.
Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!
Australia is advancing plans to regulate digital asset platforms under its financial services framework. The Senate committee recommended passing the Digital Assets Framework Bill 2025, bringing Australia closer to licensing crypto exchanges and tokenisation platforms.
Industry groups have raised concerns about definitions such as ‘digital token’ and ‘factual control.’ Broad wording could inadvertently cover infrastructure providers, including multi-party wallet systems, potentially classifying them as financial service operators.
Ripple Labs emphasised the need for precise language to avoid unintended regulation.
The committee supported the Treasury’s approach while planning to refine technical details through future regulations. Coinbase welcomed the progress but noted ongoing banking challenges for crypto firms.
The bill now proceeds to the Senate for debate and a final vote, which could reshape digital asset operations in Australia.
Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!
A service disruption has affected users of Microsoft Exchange Online, and Microsoft has confirmed ongoing investigations into mailbox access issues affecting enterprise customers worldwide.
Reports indicate that Microsoft users encountered difficulties connecting via multiple access points, including the Microsoft Outlook desktop and mobile applications and browser-based email services. The issue affects specific connection methods rather than the entire platform.
Organisations relying on cloud-based communication tools experienced interruptions in email workflows, calendar scheduling, and shared mailbox functionality. Such disruptions can significantly disrupt operational continuity, particularly for businesses that depend on real-time communication systems.
Updates through Microsoft’s service health channels suggest that engineering teams are working to identify the root cause, though no definitive explanation has yet been provided.
Such incidents highlight broader concerns around resilience in cloud infrastructure, as enterprises increasingly depend on centralised platforms for critical communication services.
Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!
X has submitted a compliance proposal to the European Commission outlining how it intends to modify its blue check verification system following regulatory concerns under the Digital Services Act.
The EU regulators concluded that the platform’s system allowed users to obtain verification simply by paying for a subscription without meaningful identity checks, potentially misleading users about the authenticity of accounts.
The Commission imposed a €120 million fine in December and gave the company 60 working days to propose corrective measures. Officials confirmed that X met the deadline for submitting a plan, which regulators will now assess.
The platform, owned by Elon Musk, must also pay the penalty while the Commission evaluates the proposed changes. The company has challenged the enforcement decision before the EU’s General Court.
Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!
In the EU, France is calling for planned European AI ‘gigafactories’ to focus on testing and scaling European technologies rather than primarily increasing demand for hardware from companies such as Nvidia.
The large computing facilities are intended to provide the infrastructure needed to train advanced AI systems. However, officials in France argue that the projects should strengthen Europe’s technological capabilities rather than reinforce reliance on foreign suppliers.
Several EU countries, including Poland, Austria and Lithuania, support using the infrastructure to improve Europe’s digital resilience.
The initiative forms part of the European Commission’s wider plans to expand computing capacity and support the development of a stronger European AI ecosystem.
Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!
According to reports, OpenAI is preparing to integrate its AI video generator Sora directly into ChatGPT, a move that could expand the platform’s capabilities beyond text and image generation.
Sora currently operates as a standalone application and web service. Integrating the tool into ChatGPT could dramatically increase its visibility and usage, particularly given the chatbot’s massive global user base.
The company released an updated version of the model in 2025 that allows users to create, remix and even appear inside AI-generated videos. Bringing those features into ChatGPT would represent a major step toward making video generation a mainstream function within conversational AI systems.
Competition in the generative video market is intensifying. Companies, including Google, are developing similar technologies, with the company’s Gemini platform offering video creation powered by the Veo system. Other developers are also launching text-to-video models as the field rapidly expands.
Despite the potential growth, integrating video generation into ChatGPT may significantly increase operating costs. Running large AI systems requires vast computing resources and energy, and the chatbot already costs billions of dollars annually to operate.
Although OpenAI earns revenue from subscriptions, the majority of ChatGPT users currently use the free version. The company is therefore exploring additional monetisation strategies, including advertising and new premium services.
Integrating Sora into ChatGPT could therefore serve both strategic and financial goals, strengthening the platform’s position in the competitive generative AI market while expanding the types of content users can create.
Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!
