OpenAI’s next major model may arrive alongside a much larger hardware push. According to the source article, Sam Altman is pursuing talks tied to AI chip production while OpenAI works on a GPT-4 successor described as a "major upgrade" and expected to be released later this year.
The connection is straightforward: more capable AI models require more computing power to train and run. If OpenAI expects demand for future systems to keep rising, chip supply becomes not just a procurement issue, but a strategic constraint.
A reported GPT-4 successor puts compute back in focus
According to the Financial Times, OpenAI is working on a new AI model that will be a "major upgrade" to GPT-4. The report says the model is expected to be released later this year.
At the World Economic Forum in Davos, Altman described GPT-4 as a "preview" of future developments. That framing matters because it suggests GPT-4 is not being treated as the endpoint of OpenAI’s model roadmap. It is being positioned as part of a continuing escalation in capability.
That escalation has a hardware consequence. Larger and more widely used AI models rely on chips for both training and deployment. The source article states that OpenAI’s reliance on AI chips will continue to grow.
For users, the visible result may be better AI systems. For companies building those systems, the less visible challenge is whether enough advanced computing power will be available at the right scale.
Altman is looking beyond today’s chip supply chain
The source article says Altman is in talks with Middle Eastern investors, including Sheikh Tahnoon bin Zayed al-Nahyan of the United Arab Emirates. The goal is to develop chips needed for training and running AI models, while reducing OpenAI’s dependence on Nvidia.
Altman is also in talks with Taiwan Semiconductor Manufacturing Co (TSMC) about a chip manufacturing partnership. It remains unclear whether the proposed chip venture would operate as a subsidiary of OpenAI or as a separate entity. However, OpenAI is expected to be the main customer of the new company.
Bloomberg is also cited in the source article as reporting that Altman wants to raise billions of dollars for a chip company that would build a global network of semiconductor fabs. The publication cited several people with knowledge of the matter.
The project would reportedly involve talks with various large potential investors, including Abu Dhabi-based G42 and SoftBank Group Corp., and would partner with leading chipmakers.
The core concern is long-term AI chip capacity
Altman is reportedly concerned that, as AI becomes more widespread, there will not be enough chips available for large-scale deployment. The source article says some current forecasts for AI chip production fall short of projected demand.
That concern is not only about today’s shortage or today’s model training runs. It is about whether the industry can prepare enough supply by the end of the decade. Altman believes the industry needs to act now, according to the source article.
The plan is ambitious because semiconductor fabs are not quick or cheap to build. A single state-of-the-art fab can cost tens of billions of dollars to build, and a global network of chip factories would require significant investment and take years.
That makes the reported effort different from the approach taken by several other major technology companies. The source article notes that Amazon, Google, and Microsoft are focusing on designing custom semiconductor products while outsourcing manufacturing to outside companies.
Nvidia’s dominance shapes the market
The current AI boom has made Nvidia central to the market for AI computing power. Google, Amazon, Meta, OpenAI, and Microsoft are all using Nvidia GPUs to train AI and deploy models to customers.
Meta alone plans to install 340,000 Nvidia H100 GPUs in its server farms by the end of the year. That number shows the scale at which leading AI companies are now thinking about compute infrastructure.
The source article says Nvidia currently dominates the market for AI computing power and sets the prices, resulting in extreme revenue growth. It also notes that chip startups like Graphcore find it difficult to compete with Nvidia’s dominance because the advantage comes from the interaction between software and hardware.
In other words, the challenge is not simply to design a faster chip. Any alternative must also fit into the broader software and hardware environment that AI developers already depend on.
Big AI companies are building alternatives
The source article describes a broader push among major AI players to reduce reliance on existing chip supply. Microsoft unveiled its first AI chips at the end of November, while Meta did so in the spring of 2023. Google, through TPU, and Amazon, through Trainium, have been developing these chips for years.
Microsoft is also working closely with Advanced Micro Devices (AMD) on AMD’s MI300X AI chip, which has been available since early December 2023.
Altman is also an investor in RainAI, which is developing a neuromorphic processing unit (NPU) that mimics human-like functions and promises significantly higher processing power and energy efficiency than today’s GPUs. OpenAI will purchase $51 million worth of AI chips from Rain AI, described in the source article as a tiny fraction of the total amount being spent on AI chips.
Taken together, the reports point to a clear strategic direction. OpenAI’s model roadmap and its chip strategy are becoming harder to separate. If the GPT-4 successor is indeed a "major upgrade," the race for AI chips will likely remain central to how quickly such systems can be trained, deployed, and scaled.