E-commerce and trade

Mistral AI expands European footprint with acquisition of Koyeb

Mistral AI has strengthened its position in Europe’s AI sector through the acquisition of Koyeb. The deal forms part of its strategy to build end-to-end capacity for deploying advanced AI systems across European infrastructure.

The company has been expanding beyond model development into large-scale computing. It is currently building new data centre facilities, including a primary site in France and a €1.2 billion facility in Sweden, both aimed at supporting high-performance AI workloads.

The acquisition follows a period of rapid growth for Mistral AI, which reached a valuation of €11.7 billion after investment from ASML. French public support has also played a role in accelerating its commercial and research progress.

Mistral AI now positions itself as a potential European technology champion, seeking to combine model development, compute infrastructure and deployment tools into a fully integrated AI ecosystem.

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

Rising DRAM prices push memory to the centre of AI strategy

The cost of running AI systems is shifting towards memory rather than compute, as the price of DRAM has risen sharply over the past year. Efficient memory orchestration is now becoming a critical factor in keeping inference costs under control, particularly for large-scale deployments.

Analysts such as Doug O’Laughlin and Val Bercovici of Weka note that prompt caching is turning into a complex field.

Anthropic has expanded its caching guidance for Claude, with detailed tiers that determine how long data remains hot and how much can be saved through careful planning. The structure enables significant efficiency gains, though each additional token can displace previously cached content.

The growing complexity reflects a broader shift in AI architecture. Memory is being treated as a valuable and scarce resource, with optimisation required at multiple layers of the stack.

Startups such as Tensormesh are already working on cache optimisation tools, while hyperscalers are examining how best to balance DRAM and high-bandwidth memory across their data centres.

Better orchestration should reduce the number of tokens required for queries, and models are becoming more efficient at processing those tokens. As costs fall, applications that are currently uneconomical may become commercially viable.

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

China boosts AI leadership with major model launches ahead of Lunar New Year

Leading Chinese AI developers have unveiled a series of advanced models ahead of the Lunar New Year, strengthening the country’s position in the global AI sector.

Major firms such as Alibaba, ByteDance, and Zhipu AI introduced new systems designed to support more sophisticated agents, faster workflows and broader multimedia understanding.

Industry observers also expect an imminent release from DeepSeek, whose previous model disrupted global markets last year.

Alibaba’s Qwen 3.5 model provides improved multilingual support across text, images and video while enabling rapid AI agent deployment instead of slower generation pipelines.

ByteDance followed up with updates to its Doubao chatbot and the second version of its image-to-video tool, SeeDance, which has drawn copyright concerns from the Motion Picture Association due to the ease with which users can recreate protected material.

Zhipu AI expanded the landscape further with GLM-5, an open-source model built for long-context reasoning, coding tasks, and multi-step planning. The company highlighted the model’s reliance on Huawei hardware as part of China’s efforts to strengthen domestic semiconductor resilience.

Meanwhile, excitement continues to build for DeepSeek’s fourth-generation system, expected to follow the widespread adoption and market turbulence associated with its V3 model.

Authorities across parts of Europe have restricted the use of DeepSeek models in public institutions because of data security and cybersecurity concerns.

Even so, the rapid pace of development in China suggests intensifying competition in the design of agent-focused systems capable of managing complex digital tasks without constant human oversight.

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

AI startup raises $100m to predict human behaviour

Artificial intelligence startup Simile has raised $100m to develop a model designed to predict human behaviour in commercial and corporate contexts. The funding round was led by Index Ventures with participation from Bain Capital Ventures and other investors.

The company is building a foundation model trained on interviews, transaction records and behavioural science research. Its AI simulations aim to forecast customer purchases and anticipate questions analysts may raise during earnings calls.

Simile says the technology could offer an alternative to traditional focus groups and market testing. Retail trials have included using the system to guide decisions on product placement and inventory.

Founded by Stanford-affiliated researchers, the startup recently emerged from stealth after months of development. Prominent AI figures, including Fei-Fei Li and Andrej Karpathy, joined the funding round as it seeks to scale predictive decision-making tools.

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

Tokyo semiconductor profits surge amid AI boom

Major semiconductor companies in Tokyo have reported strong profit growth for the April to December period, buoyed by rising demand for AI related chips. Several firms also raised their full year forecasts as investment in AI infrastructure accelerates.

Kioxia expects net profit to climb sharply for the year ending in March, citing demand from data centres in Tokyo and devices equipped with on device AI. Advantest and Tokyo Electron also upgraded their outlooks, pointing to sustained orders linked to AI applications.

Industry data suggest the global chip market will continue expanding, with World Semiconductor Trade Statistics projecting record revenues in 2026. Growth is being driven largely by spending on AI servers and advanced semiconductor manufacturing.

In Tokyo, Rapidus has reportedly secured significant private investment as it prepares to develop next generation chips. However, not all companies in Japan share the optimism, with Screen Holdings forecasting lower profits due to upfront capacity investments.

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

AI governance becomes urgent for mortgage lenders

Mortgage lenders face growing pressure to govern AI as regulatory uncertainty persists across the United States. States and federal authorities continue to contest oversight, but accountability for how AI is used in underwriting, servicing, marketing, and fraud detection already rests with lenders.

Effective AI risk management requires more than policy statements. Mortgage lenders need operational governance that inventories AI tools, documents training data, and assigns accountability for outcomes, including bias monitoring and escalation when AI affects borrower eligibility, pricing, or disclosures.

Vendor risk has become a central exposure. Many technology contracts predate AI scrutiny and lack provisions on audit rights, explainability, and data controls, leaving lenders responsible when third-party models fail regulatory tests or transparency expectations.

Leading US mortgage lenders are using staged deployments, starting with lower-risk use cases such as document processing and fraud detection, while maintaining human oversight for high-impact decisions. Incremental rollouts generate performance and fairness evidence that regulators increasingly expect.

Regulatory pressure is rising as states advance AI rules and federal authorities signal the development of national standards. Even as boundaries are debated, lenders remain accountable, making early governance and disciplined scaling essential.

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

Hybrid AI could reshape robotics and defence

Investors and researchers are increasingly arguing that the future of AI lies beyond large language models. In London and across Europe, startups are developing so-called world models designed to simulate physical reality rather than simply predict text.

Unlike LLMs, which rely on static datasets, world models aim to build internal representations of cause and effect. Advocates say these systems are better suited to autonomous vehicles, robotics, defence and industrial simulation.

London based Stanhope AI is among companies pursuing this approach, claiming its systems learn by inference and continuously update their internal maps. The company is reportedly working with European governments and aerospace firms on AI drone applications.

Supporters argue that safety and explainability must be embedded from the outset, particularly under frameworks such as the EU AI Act. Investors suggest that hybrid systems combining LLMs with physics aware models could unlock large commercial markets across Europe.

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

Next-gen AI infrastructure boosted by Samsung HBM4

Samsung Electronics has commenced mass production and commercial shipments of its next-generation HBM4 memory, marking the first industry deployment of the advanced high-bandwidth solution.

The launch strengthens the company’s position in AI infrastructure hardware as demand for accelerated computing intensifies.

Built on sixth-generation 10nm-class DRAM and a 4nm logic base die, HBM4 delivers transfer speeds of 11.7Gbps, with performance scalable to 13Gbps. Bandwidth per stack has surged, reducing data bottlenecks as AI models and processing demands grow.

Engineering upgrades extend beyond raw speed. Enhanced stacking architecture, low-power design integration, and thermal optimisation have improved energy efficiency and heat dissipation, supporting large-scale data centre deployments and sustained GPU workloads.

Production scale-up is already in motion, backed by expanded manufacturing capacity and industry partnerships. Samsung expects HBM revenue growth to accelerate into 2026, with next-generation variants and custom configurations scheduled for future release cycles.

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

Illicit trafficking payments rise across blockchain channels

Cryptocurrency flows linked to suspected human trafficking services surged sharply in 2025, with transaction volumes rising 85% year-on-year, according to new blockchain analysis.

Investigators say the financial activity reflects the rapid expansion of digitally enabled exploitation networks operating across borders.

Growth is linked to Southeast Asia-based illicit networks, including scam compounds, gambling platforms, and laundering groups operating via encrypted messaging channels.

Analysts identified multiple trafficking service categories, each with distinct transaction structures and payment preferences.

Stablecoins became the dominant payment method, especially for escort networks, thanks to their price stability and ease of conversion. Larger transfers and structured pricing models indicate increasingly professionalised operations supported by organised financial infrastructure.

Despite the scale of the activity, blockchain transparency continues to provide enforcement advantages. Transaction tracing has aided investigations, shutdowns, and arrests, strengthening digital forensics in combating trafficking-linked financial crime.

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

EU considers blanket crypto ban targeting Russia

European Union officials are weighing a sweeping prohibition on cryptocurrency transactions involving Russia, signalling a more rigid sanctions posture against alternative financial networks.

Policymakers argue that the rapid emergence of replacement crypto service providers has undermined existing restrictions.

Internal European Commission discussions indicate concern that digital assets are facilitating trade flows supporting Russia’s war economy. Authorities say platform-specific sanctions are ineffective, as new entities quickly replicate restricted services.

Proposals under review extend beyond private crypto platforms. Measures could include sanctions on additional Russian banks, restrictions linked to the digital ruble, and scrutiny of payments infrastructure tied to sanctioned trade channels.

The consensus remains uncertain, with some states warning that a blanket ban could shift activity to non-European markets. Parallel trade controls targeting dual-use exports to Kyrgyzstan are also being considered as part of broader anti-circumvention efforts.

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