What happened
Mubadala CEO Khaldoon Al Mubarak told Semafor that the rapid expansion of AI models and services is hitting a “ shortage” that could force a slowdown in the sector. He described the situation as a potential “winter” for AI, but one that might be mild rather than severe. Al Mubarak highlighted Mubadala’s large investments in AI infrastructure, data‑center developers, and stakes in OpenAI and Anthropic via the MGX vehicle, positioning the fund as a major player in the U.S. AI ecosystem. He also noted that while capital and energy are readily available, the scarcity of compute capacity is the biggest bottleneck for continued AI growth.
In an unsponsored interview with Semafor’s Silicon Valley & The World series, Khaldoon Al Mubarak said the global AI build‑out is entering a “reset” phase, likening it to a mild winter rather than a catastrophic collapse.
He emphasized that while capital and energy supplies are sufficient, the primary obstacle is securing enough power to sustain the rapid development of large AI models.
Mubadala, the Abu Dhabi sovereign‑wealth fund, has positioned itself as a major investor in AI infrastructure, holding stakes in data‑center developers and in AI firms such as OpenAI and Anthropic through its MGX vehicle.
Al Mubarak noted that about 44% of Mubadala’s assets are in North America and that the fund is aligning with U.S. efforts to dominate the AI race, rejecting any “middle‑ground” approach between the U.S. and China.
Why it matters
The warning from a leading sovereign‑wealth fund signals that the AI industry may soon confront a supply‑side constraint that could temper the current hype and investment frenzy. —primarily GPUs and specialized chips—is essential for training and running large language models; a shortage can delay product rollouts, increase costs, and shift competitive dynamics. Mubadala’s focus on U.S. AI infrastructure and its alignment with U.S. policy underscores the geopolitical dimension of the compute race, especially as firms choose between U.S. and Chinese technology stacks. If the shortage persists, it could spur new investments in chip manufacturing, data‑center capacity, or alternative architectures, reshaping the AI supply chain and influencing which firms can scale their models.
scarcity can raise the cost of training and running large models, potentially slowing innovation and limiting the number of firms that can compete at scale.
The geopolitical framing—choosing the U.S. technology stack over China—highlights how national policy and supply‑chain decisions are becoming intertwined with AI development strategies.
Mubadala’s investment focus may accelerate the build‑out of U.S. data‑center capacity, influencing where future AI workloads are hosted and which regions attract AI talent and capital.
If the shortage persists, it could trigger a wave of new funding for alternative hardware solutions, such as custom AI chips or more efficient model architectures, reshaping the AI hardware landscape.
Interactive Mechanism: How It Actually Works
Explore the underlying technology behind this development interactively.
What should a useful AI forecast state?
What to watch next
Watch for announcements from major chip makers (e.g., Nvidia, AMD, Intel) and data‑center operators about expanding capacity, as well as any policy initiatives from the U.S. government aimed at boosting domestic resources. Track Mubadala’s subsequent investment moves, especially any new funding rounds for AI infrastructure startups, which could indicate how the fund plans to mitigate the shortage. Finally, monitor pricing trends for cloud compute services, as tighter supply may drive up costs for AI developers.
Announcements from GPU manufacturers about new production capacity or next‑generation chips that could alleviate the bottleneck.
Policy measures from the U.S. government aimed at increasing domestic semiconductor manufacturing or incentivizing data‑center construction.
Mubadala’s future investment activity, particularly any new commitments to AI‑focused data‑center projects or hardware startups.
Changes in cloud‑service pricing for ‑intensive workloads, which would reflect market pressure from the shortage.