Capital & Industrial Strategy

Moonshot AI is raising up to $1 billion in an expanded funding round at approximately $18 billion valuation, more than quadrupling its value from roughly $4 billion just three months earlier, according to Bloomberg. The rapid appreciation reflects growing investor conviction that Chinese AI developers can compete with Silicon Valley frontier labs despite US export controls on advanced chips. Moonshot, which develops large language models, is benefiting from both domestic capital seeking alternatives to restricted US technology and Beijing's industrial policy prioritising AI sovereignty.

The valuation trajectory suggests Chinese AI startups are no longer trading at a discount to Western peers due to geopolitical risk — instead, investors are pricing in the scale of China's domestic market and state support. This creates a parallel ecosystem where capital, talent, and deployment pathways diverge structurally from the US-led stack, reducing the leverage of US chip export controls over the long term.

The US Army awarded Anduril a contract worth up to $20 billion, consolidating more than 120 separate procurement actions into a single enterprise arrangement, TechCrunch reports. This structure marks a shift from traditional defence contracting, where multiple vendors compete for discrete programmes, toward integrated platform providers that can deliver autonomous systems, AI-enabled surveillance, and battlefield decision support as a unified stack. Anduril has positioned itself as a software-first defence company, and this contract validates that positioning at material scale.

The consolidation reduces fragmentation in military AI deployment and signals that the Pentagon is willing to concentrate spending on vendors who can integrate AI across domains rather than procuring point solutions. It also demonstrates that defence is becoming a primary monetisation pathway for advanced AI — potentially on par with consumer and enterprise software — as governments prioritise capabilities over cost efficiency in strategic competition with China.

Meta is considering layoffs that could affect up to 20% of its workforce, TechCrunch reports, as the company attempts to offset aggressive capital expenditures on AI infrastructure, acquisitions, and specialised hiring. Meta has committed tens of billions to GPU clusters, custom silicon development through its MTIA programme, and acquisitions of AI talent and companies, but faces investor pressure to demonstrate return on investment. The layoffs would primarily target non-AI roles, reflecting a strategic reallocation of human capital toward AI research and deployment while shedding legacy functions.

This mirrors broader patterns across Big Tech, where AI capex is crowding out other investments and forcing workforce restructuring. The scale — potentially 20,000+ employees — suggests Meta believes AI productivity gains can absorb the workload reduction, effectively using AI deployment to justify headcount cuts. It also indicates that even well-capitalised incumbents face trade-offs between maintaining current operations and funding AI transformation, rather than simply layering AI spending atop existing budgets.

Amazon Web Services announced a partnership with Cerebras to offer inference computing using the startup's wafer-scale engine chips, which deliver exceptionally low latency due to their architecture, The Wall Street Journal reports. The deal extends AWS's strategy of providing customers with multiple compute options — including Nvidia GPUs, its own Trainium and Inferentia chips, and now Cerebras — to reduce reliance on any single supplier and capture more of the AI workload stack. For Cerebras, the partnership provides a distribution channel and validation after its recent IPO, while for AWS it creates competitive differentiation against Google Cloud and Microsoft Azure in latency-sensitive inference workloads.

The partnership reflects a broader shift in cloud strategy: rather than simply reselling Nvidia capacity, hyperscalers are curating a portfolio of specialised silicon to lock in customers and capture margin. This creates opportunities for chip startups like Cerebras, Groq, and SambaNova to access enterprise customers through cloud partnerships, but also concentrates power with the hyperscalers as the gatekeepers of AI infrastructure.