OpenAI plans to spend $750 billion on infrastructure by 2030

What happened?

OpenAI has revealed that it plans to spend up to $750 billion on artificial intelligence infrastructure by 2030, as reported by TechCrunch. This figure, comparable to Sweden's GDP (which was approximately $590 billion in 2022, according to the World Bank), includes investments in data centers, custom semiconductors, and energy systems. The scale of this spending is historic: it exceeds the annual defense budget of the United States (about $850 billion for 2024) and is more than double the total AI infrastructure investment of the entire industry in 2023. To put it in perspective, the construction of the Large Hadron Collider cost about $4.75 billion, and NASA's Apollo program totaled about $25 billion adjusted for inflation. This OpenAI plan is, therefore, an order of magnitude greater than any previous technological project.

Why is it important?

This level of unprecedented spending reflects the race to dominate AI on a global scale. OpenAI seeks to secure the computational capacity necessary to train increasingly larger models, such as GPT-5 and future versions. According to industry estimates, training GPT-4 required about $100 million in compute; GPT-5 could cost more than $1 billion. The investment could also redefine the technology supply chain, with implications for chip manufacturers like NVIDIA (whose shares fell 2% after the announcement) and TSMC. In addition, OpenAI is developing its own AI chips, which could reduce its dependence on NVIDIA, whose profit margin on its H100 GPUs exceeds 80%. Microsoft, OpenAI's main investor with $13 billion committed, would also benefit from integrating these models into Azure. However, this massive spending could trigger an investment war with competitors like Google (which invested $30 billion in capital expenditures in 2023) and Anthropic.

What will be the consequences?

The plan could accelerate the adoption of AI in key sectors such as healthcare, finance, and logistics, but it also raises concerns about energy consumption and environmental impact. A state-of-the-art AI data center can consume up to 100 megawatts, equivalent to 80,000 homes. If OpenAI builds dozens of these facilities, its carbon footprint could rival that of small countries. Furthermore, it raises questions about OpenAI's financial sustainability: the company spent $540 million in 2022 and brought in only $28 million, according to The Information. With this plan, it would need to generate massive revenue through subscriptions (ChatGPT Plus, enterprise API) and deals like the $10 billion agreement with Microsoft. There is also the risk that the investment will not be recouped if AI adoption does not grow at the expected rate. On the other hand, market consolidation around a few players with access to massive resources could reduce competition and increase the prices of AI services.

What should readers know?

Readers should understand that this investment is not just an OpenAI bet, but an indicator that AI infrastructure is becoming a strategic asset comparable to telecommunications networks. Companies and governments will need to evaluate their own position regarding this trend. For example, the European Union plans to invest 4 billion euros in AI chips and data centers, and China has allocated $10 billion to its AI ecosystem. At the user level, this could translate into better AI services (faster and more accurate models) but also greater corporate control over the technology. In addition, dependence on AI infrastructure could create new geopolitical vulnerabilities, such as the concentration of chip production in Taiwan (TSMC) and South Korea (Samsung). In short, this announcement marks a turning point: AI is no longer just software, but a capital-intensive industry that will redefine the global balance of technological power.