OpenAI’s Ambitious Infrastructure Plans Unveiled

OpenAI CEO Sam Altman is embarking on an ambitious project to develop artificial intelligence hardware and systems worth tens of billions of dollars, according to Bloomberg, citing sources.

In February, the agency reported Altman’s efforts to secure U.S. government approval for a large-scale chip manufacturing venture essential for neural networks. That same month, media outlets learned of Altman’s negotiations with investors to raise between $5 trillion and $7 trillion for these purposes.

According to Bloomberg, Altman and his team are now discussing the construction of data centers, increasing power capacity and transmission using turbines and generators, and expanding semiconductor production.

Potential sponsors include investors from Canada, Korea, Japan, and the UAE, with expected participation from Microsoft and other companies.

OpenAI executives, including Altman, have been meeting with investors in recent weeks to promote the deal. OpenAI’s Chief Strategy Officer Jason Kwon recently visited Japan and Korea, while OpenAI’s Vice President for Global Policy Chris Lehane engaged with investors in Canada, and Altman discussed cooperation with investors from the UAE, the agency reports.

In recent months, OpenAI has held meetings with the U.S. National Security Council regarding the planned investments, Bloomberg noted.

During negotiations, Altman and other OpenAI leaders emphasize that the global infrastructure plan will bolster America’s geopolitical advantage by creating a multinational coalition to compete with China’s infrastructure consortium. Foreign capital, which might otherwise flow to China, will be invested in the U.S.

The discussions occur as OpenAI nears the closure of a funding round valuing the company at over $100 billion.

One advantage of participating in the initiative is that other companies and U.S. ally countries will benefit from OpenAI’s infrastructure projects. Jobs will also be created.

Strawberry, Orion, and GPT Next

Meanwhile, media outlets are uncovering details about OpenAI’s upcoming AI models.

A representative of OpenAI’s Japanese subsidiary, Tadao Nagasaki, announced plans for the imminent launch of a new neural network called GPT Next, according to IT Media.

He stated that it will be “almost 100 times more advanced than previous ones.”

In July, it emerged that the startup was working on a new project codenamed Strawberry.

With Strawberry, Altman’s firm aims to significantly enhance AI models’ reasoning processes and enable them to plan actions in advance, allowing them to “autonomously and reliably” explore the internet and conduct “in-depth research.”

Previously, the project was called Q* and was considered a breakthrough for creating artificial general intelligence (AGI) that “could threaten humanity.”

Another language model OpenAI is working on is Orion. Media note that it is positioned as a flagship model and could potentially replace GPT-4 in language understanding and generation with additional multimodal capabilities.

In August, OpenAI introduced a new feature for customizing GPT-4o to suit the unique characteristics of individual companies.

Earlier, the startup introduced the ability to generate up to two images per day in the free version of ChatGPT using the DALL-E 3 AI model.