OpenAI’s plan to develop semiconductor chips for advanced AI models could receive financial support from the United Arab Emirates. According to a Financial Times report cited by AI News, Abu Dhabi state-backed investment group MGX was in discussions to support the effort. The report relied on two people familiar with the talks, making the potential investment a developing proposal rather than a completed agreement.
The discussions form part of OpenAI’s attempt to secure enormous amounts of capital for in-house chip development. The company was reportedly seeking investment measured in trillions of dollars from sources around the world. Building its own semiconductor capability could reduce reliance on Nvidia, whose GPUs had become the dominant hardware for training and operating advanced AI models.
Compute Supply Has Become a Strategic Constraint
Model developers depend on access to large clusters of accelerators. As models and user demand grow, limited supply can slow research, raise operating costs and constrain product capacity. OpenAI’s interest in custom chips reflects a desire to influence one of the most important inputs to its business rather than accepting hardware availability and design choices determined by another company.
Designing chips internally could also optimise hardware around the company’s workloads. A purpose-built accelerator may improve performance or efficiency for particular training and inference patterns. However, semiconductor development requires specialised engineering, manufacturing partnerships, packaging, memory and a long supply chain. Funding is necessary, but money alone does not eliminate the technical and industrial difficulty.
MGX Gives the UAE a Role in AI Infrastructure
MGX was launched as an AI-focused fund and is led by UAE national security adviser Sheikh Tahnoon Bin Zayed al-Nahyan. It was established with G42 and Mubadala. G42 had already partnered with OpenAI in October 2023 as part of the model developer’s expansion in the Middle East, creating an existing relationship around which further infrastructure discussions could develop.
Potential support for chip development would move the relationship closer to the foundation of the AI stack. Instead of investing only in applications, the UAE would gain exposure to the compute infrastructure on which advanced models depend. For OpenAI, a state-backed fund can offer the scale and long-term perspective required by a capital-intensive semiconductor programme.
Reducing Dependence Creates New Dependencies
Moving away from Nvidia would not make OpenAI self-sufficient. Even a custom design must be fabricated, assembled and integrated into data-centre systems. The company would need access to manufacturing capacity, advanced memory and networking, along with software capable of making the hardware useful. The project would therefore replace one supplier relationship with a broader set of industrial partnerships.
Execution time is another consideration. Nvidia can continue improving its own architecture while a new entrant develops a first chip. OpenAI must decide whether custom hardware provides enough strategic and economic benefit to justify the delay and risk. It may also continue buying commercial GPUs during development, creating a hybrid approach rather than an immediate replacement.
Semiconductor Funding Is Becoming Geopolitical
The source placed the UAE discussions alongside UK participation in the European Union’s Chips Joint Undertaking. That programme gave British semiconductor research access to a €1.3 billion Horizon Europe funding pool through 2027, backed by initial UK government support of £5 million and a further £30 million planned for participation between 2025 and 2027.
The two initiatives differ greatly in scale and purpose, but both show governments and state-linked investors treating semiconductor capability as strategic. Chips affect AI competitiveness, scientific research and supply-chain resilience. Capital decisions therefore carry geopolitical implications alongside commercial ones, especially when investors, model developers and manufacturing capacity span several regions.
OpenAI’s reported valuation had already risen above $80 billion following a Thrive Capital deal, illustrating investor confidence in the company’s software and model business. Semiconductor ambition operates on a different financial level. Fabrication and infrastructure require sustained capital long before the resulting hardware can produce revenue, making partners with deep resources particularly important.
The potential MGX investment is best understood as an early signal of how AI competition is expanding. Model companies want greater control over compute; nations want positions in the infrastructure economy; and investors are willing to consider projects once associated mainly with established chipmakers. Whether the talks produce a working OpenAI processor remains uncertain. What is clear is that access to specialised silicon has become a strategic question powerful enough to reshape international investment relationships.
Governance would be a central part of any large cross-border arrangement. Investors and OpenAI would need agreement on ownership, access and the location of resulting capabilities. Because MGX is state backed and led by a national security adviser, commercial discussions would naturally sit alongside questions about strategic technology and international relationships.
The scale reported for OpenAI’s broader fundraising ambition also suggests that a single investor would be only one part of the solution. Chip design, fabrication and data-centre expansion could involve different partners and financing structures. Coordinating them becomes a major management challenge before the first custom processor proves its value.
For Nvidia, the project is evidence that its largest customers are considering alternatives, but it is not an immediate loss of demand. OpenAI would continue to require substantial compute during development, and Nvidia’s mature ecosystem remains difficult to reproduce. Custom silicon would need to deliver meaningful gains after software and migration costs.
The talks therefore reveal both ambition and dependency. OpenAI wants more control over a critical resource, while the UAE wants a deeper place in the AI economy. A partnership could advance both goals, yet its success would rest on patient capital, semiconductor expertise and an industrial network far wider than either party alone.
