📁 Market

The PHLX Semiconductor Index plunged roughly 12% in two sessions, after a first half that gained over 80%. A signal that reshapes priorities for those building local stacks around GPUs and on-prem LLMs.

The German enterprise software giant is cutting operational costs to redirect cash into artificial intelligence projects. The move highlights competitive pressure in the enterprise AI space and reignites debate on deployment models, especially in Europe where data sovereignty and on-premise architectures remain critical.

June saw more but smaller funding rounds across Europe. Openchip raised 115 million for semiconductors, IQM became the first European quantum company to list in the US. AI-RADAR analysis on what this means for those adopting on-premise stacks and self-hosted LLMs.

Currys, the UK electronics giant, warns that PCs, phones and TVs will get more expensive in the coming months due to AI-related memory pressure. The same dynamic is making it harder and costlier to source hardware for on-premise LLM deployments.

With the acquisition of Common Room, Zoom embeds AI into the sales process before the video call. The startup analyzes buying signals, pushing Zoom into enterprise software. The deal, announced without financial details, marks a step toward an ecosystem where the platform controls data and conversations, raising sovereignty questions for those evaluating on-premise.

Starling Bank cuts 130 roles across banking and tech to streamline operations and speed product delivery, after profit and revenue dipped. The UK challenger bank is deploying an agentic AI assistant, a first in the country. The cuts and the AI push raise questions about data sovereignty and infrastructure choices: cloud vs on-prem. A restructuring that shows how AI adoption is reshaping work in financial services.

The company, specialized in modular data centers powered by wasted natural gas, could triple its valuation to $30 billion. The funding round highlights the growing demand for AI compute and pushes toward distributed infrastructure, with potential implications for on-premise deployment and data sovereignty.

Danish fund Climentum Capital held a €60 million first close for its second climate tech vehicle, backed by EIF, EIFO and IDA. The focus is on hardware for energy, industry and supply chain sovereignty. For those running LLM inference on-premise, these technologies directly affect energy cost and availability, a critical TCO factor.

Chinese NOR Flash supplier Giantec has announced a 25% price hike, signaling ongoing memory market pressure. The increase could impact hardware costs for self-hosted LLM inference, particularly in edge and embedded scenarios where NOR Flash is common.

The Bavarian startup Quantum Systems has closed a $1.2 billion Series D round, more than doubling its valuation to around $8 billion. Co-led by Blackstone, Noteus, Airbus, and Advent, it ranks among the largest ever for a European defense startup. The autonomous drone market is accelerating, driven by growing demand for unmanned surveillance and combat capabilities.

Swiss biotech startup ALP Bio has secured €161,000 from Venture Kick to advance its platform combining human immune models with artificial intelligence. The goal is to detect immunogenicity risks of biologic drugs early, reducing clinical failures and improving safety. Initial pilot projects with pharma partners will test the technology in real-world settings and strengthen the company’s commercial position.

According to DIGITIMES, global server demand will remain robust through 2027, with widening supply chain pressures. For organizations planning on-premise deployments of Large Language Models, this translates into longer lead times, budget revisions, and a need for inference optimization. TCO analysis becomes critical to avoid delays and maintain data sovereignty.

A new $2.5 billion unit with engineers embedded at client sites aims to accelerate enterprise AI adoption. It signals just how hard it is to bring AI out of the lab and into real workflows.

DIGITIMES reveals the strategy behind TSMC's effort to strengthen semiconductor supply chain resilience by building a 'second fleet' of alternative suppliers and capacities. For those running LLM inference on-premises, GPU availability is a cost and risk factor: a more stable supply chain could lower TCO and ease AI infrastructure planning.

Beijing is taking action against excessively large EV batteries, according to an AFP analysis. The goal is to curb public spending and manage raw material supply chain pressures.

Meta's reported cloud acceleration fuels debate over real-world AI chip demand. As NVIDIA cements its lead, observers question impacts on procurement and on-premise deployment choices. AI-RADAR explores the trade-offs.

Nvidia is reportedly expanding its financing push for AI cloud providers using a revenue-sharing model. The move could speed up high-end GPU adoption but raises questions about technological lock-in and the impact on on-premise deployment strategies.

The Innoscience-Infineon legal battle reveals how Chinese courts are becoming a strategic weapon in the gallium nitride race, a semiconductor critical for powering on-premise AI clusters and data centers.

Both cybersecurity firms are expanding AI integrations for enterprises. A move that reopens the debate on where to run models: cloud or on-premise, between data sovereignty and latency.

Talk of BYD taking over Volkswagen exposes Europe’s industrial strain. An auto sector reshuffle could reshape advanced chip demand, directly affecting availability and cost of on-premise LLM infrastructure.