Nscale vs NVIDIA

Nscale's IPO tests Wall Street's appetite for concentrated AI bets

British neocloud Nscale is preparing for an IPO on the NYSE, aiming for a $35 billion valuation and raising $3 billion. The company, spun out of Arkon Energy two years ago, has amassed over $103 billion in contracts, but 85% of that is tied to two customers: Microsoft ($43.8 billion) and Anthropic ($44.6 billion). Anthropic's deal is contingent on Nscale securing financing and meeting stringent milestones. Nscale reported revenue of $140.6 million for H1 2026, with net losses of $1.02 billion. The IPO will test investor appetite for companies with high customer concentration, a trend highlighted by a Sona Asset Management paper. Competitors include CoreWeave, Nebius, Lambda, and Crusoe.

• Nscale plans to list on NYSE with a target valuation of $35 billion.
• The company seeks to raise $3 billion in its IPO.
• Nscale has $103 billion in contracts, with 85% from Microsoft and Anthropic.

Nscale Files for US IPO Targeting $30B Valuation

British AI infrastructure company Nscale has filed its IPO prospectus with the US SEC, aiming to list on the New York Stock Exchange under the ticker 'NSCL'. The company seeks a valuation of around $30 billion, as reported by CNBC via Reuters, a significant jump from its last private round valuation of $14.6 billion. Nscale, which transitioned from Bitcoin mining to AI compute, reported first-half 2026 revenue of $140.6 million, a 1,252% increase, but also a net loss of $1.02 billion. Its order book stands at $103.4 billion, with Microsoft and Anthropic contracts totaling $88.4 billion. The company recently secured $3.1 billion in convertible notes, with Nvidia as a key investor. Nscale faces high customer concentration, a going-concern note, and dependence on a few suppliers. Goldman Sachs, J.P. Morgan, and Morgan Stanley are lead underwriters.

• Nscale filed an IPO prospectus with the SEC to list on the NYSE under the ticker 'NSCL'.
• Targeted valuation is approximately $30 billion.
• First-half 2026 revenue was $140.6 million, up 1,252% year-over-year; net loss was $1.02 billion.

Nscale Files Registration Statement for Proposed Initial Public Offering

Nscale has filed a registration statement for a proposed initial public offering, as announced on September 18, 2026.

Apple Targets AI Compute Market with Powerful Macs

Apple is repositioning its Mac mini and Mac Studio models as professional AI infrastructure, emphasizing local inference, large language models, and predictable compute costs. The new Mac Studio with M5 Ultra, now shipping in the US, offers up to 512GB unified memory and 1.2TB/s bandwidth, with up to 36 CPU and 80 GPU cores. Apple also demonstrated clustering four Mac Studios via Thunderbolt 5 to run a trillion-parameter model. This move challenges Nvidia's dominance in AI compute and Windows' enterprise market share, positioning Apple as an alternative for on-premises AI workloads. Pricing starts at $899 for Mac mini and $5,499 for Mac Studio, with high-end configurations near $20,000. Apple emphasizes energy efficiency and total cost of ownership, targeting businesses with sustained AI workloads.

• Apple's new Mac Studio with M5 Ultra ships in the US starting September 22, 2026.
• Mac Studio M5 Ultra offers up to 512GB unified memory and 1.2TB/s memory bandwidth.
• Apple demonstrated running a trillion-parameter model on a cluster of four Mac Studios via Thunderbolt 5.

Microsoft Reveals AI-Centric Windows Plans for Agents

The Pragmatic Engineer interviews Microsoft Windows team executives about their strategy to integrate AI and agentic capabilities into the operating system. Key features include agent identity via Entra ID, a registry for MCP tools (ODR), an isolation mechanism for agent tools (MXC), a new hardware abstraction layer for local AI models (Windows ML), and embedded small language models (SLMs). Microsoft aims to win back developers by addressing past criticisms and supporting Linux via WSL. The article includes market share data showing Windows declining among developers, and highlights partnerships with NVIDIA for on-device AI performance.

• Microsoft plans native agent identification using Entra ID and an On-Device Agent Registry (ODR) for MCP tools.
• Microsoft Execution Containers (MXC) provide cross-platform isolation for agent tools with configurable containment policies.
• Windows ML is a new hardware-agnostic layer for running local AI models across GPU, NPU, and CPU, based on ONNX runtime.

Taiwan Chip Expert: Geopolitics Trumps Tech in US

In an interview at the Taiwan Innotech Expo 2026, Professor Tzi-Dar Chiueh of National Taiwan University discusses Taiwan's semiconductor dominance, the cultural work ethic behind TSMC's success, and the challenges of replicating it abroad. He notes that TSMC holds 70% of global contract manufacturing, but Taiwan's overall chip design share is below 20%. Chiueh argues that geopolitical motives, not chip industry success, drive China's interest in Taiwan, and he warns that US support for Taiwan's defense may come with costs. He also explains Taiwan's AI strategy: focusing on AI infrastructure and leveraging strengths rather than competing in large model development.

• TSMC holds about 70% of the global semiconductor foundry market.
• TSMC's top customer is NVIDIA, with Apple now second.
• Chip production at TSMC's US plant costs 30-40% more than in Taiwan, according to Chiueh.