Futurum CEO Daniel Newman on report Nvidia is halting Intel's 18A tests

Key Concepts

• Intel 18A Process: Intel’s advanced chip manufacturing process node, tested by NVIDIA but ultimately deemed unsuitable for their needs.
• TSMC: Taiwan Semiconductor Manufacturing Company, currently the dominant leader in chip manufacturing, known for reliability and predictability.
• AI Capacity Strain: The current global shortage of chip manufacturing capacity driven by the rapid growth of Artificial Intelligence.
• US Government Investment: Significant financial support provided by the US government to bolster domestic chip manufacturing capabilities.
• Hyperscalers: Large-scale cloud providers like Amazon and Meta, heavily reliant on advanced chips for their operations.
• Project Rainier: Amazon’s initiative to ramp up its AI infrastructure, including NVIDIA GPUs and internally developed silicon.

The Strategic Value of Intel’s Manufacturing Capacity & AI Chip Demand

The discussion centers around the implications of NVIDIA’s testing and subsequent abandonment of Intel’s 18A chip manufacturing process, and the broader context of the current AI-driven demand for semiconductors. Daniel Neumann argues against the notion that NVIDIA’s decision was due to Intel’s process being “not strategic enough to let it fail.” Instead, he posits that the substantial investments – $10 billion from the US government, $5 billion from NVIDIA, and multi-billion dollar contributions from ARM and SoftBank – demonstrate a critical need for expanded chip manufacturing capacity to meet the demands of the AI boom. The core argument is that simply having more capacity available, even if not surpassing TSMC in quality, is paramount in the current environment.

Intel 18A Testing & the Shift in Manufacturing Dynamics

NVIDIA’s testing of the 18A process, and its subsequent discontinuation, was already a known issue, as reported by sources like Liputin. Intel’s focus is shifting towards the 14A node for near-term production, with 27 for risk and 28 for full production. This shift is crucial because it represents the point at which companies like NVIDIA and Apple will seriously consider expanding their manufacturing partnerships beyond TSMC. The fact that Intel was even considered for 18A is seen as a positive development, potentially influenced by the company’s relationship with the US government.

Neumann clarifies that all major chip designers (Broadcom, NVIDIA, AMD) routinely test new nodes from various manufacturers (Intel, Samsung, TSMC) for potential advantages in performance and efficiency. However, TSMC currently holds a significant advantage due to its established reliability and predictability, even at a higher cost. The current capacity constraints are unprecedented, forcing companies to explore all available options. “We’ve never been in a situation where we simply can’t build enough,” Neumann states, highlighting the urgency of the situation. This is driving interest in Intel, Google’s potential move to Samsung, and overall increased demand for all available chip manufacturing capacity.

Government Influence & the US Semiconductor Strategy

The discussion highlights the significant role of the US government in incentivizing domestic chip manufacturing. The substantial financial investment is explicitly aimed at establishing a US-led, US-based manufacturer of leading-edge chips. Neumann suggests a degree of “favoritism” may be at play, with Intel benefiting from its relationship with the government, but emphasizes that this is part of a broader strategic effort to secure US leadership in semiconductor production. The policy is designed to reduce reliance on foreign manufacturers and ensure a stable supply of critical components.

2026 Predictions: Amazon, Meta, and Potential IPOs

Looking ahead to 2026, Neumann offers several predictions. He believes Amazon and Meta trailed behind other tech giants in 2024 but are poised for improvement. Specifically, he anticipates Meta’s investment in AI talent will yield positive results next year, despite initial underperformance. He also believes Amazon’s investments in NVIDIA GPUs and its internally developed silicon (Project Rainier) will allow it to leverage its position as the largest hyperscale cloud provider.

Finally, Neumann predicts a significant IPO event in 2026, specifically mentioning Allen & Sam, suggesting a potential large-scale public offering from a currently private entity. “I think there’ll be $2 trillion IPOs next year. It’s going to be Allen and Sam,” he stated.

The Changing Landscape of Manufacturing Partnerships

The conversation underscores a fundamental shift in the dynamics of chip manufacturing. Historically, companies prioritized reliability and predictability, leading to strong relationships with TSMC. However, the current capacity shortage is forcing them to re-evaluate these priorities and consider alternative manufacturers like Intel and Samsung, even if they come with increased risk. The need to secure sufficient capacity is overriding traditional considerations, creating opportunities for companies like Intel to gain market share.

Conclusion

The core takeaway is that the current AI boom has created an unprecedented demand for semiconductors, leading to a global capacity shortage. This shortage is forcing companies to explore all available manufacturing options, including Intel, despite its historical challenges. The US government’s significant investment in domestic chip manufacturing is aimed at securing US leadership in this critical industry. The situation is dynamic, with companies like Amazon and Meta poised for growth in the AI space, and the potential for significant IPO activity in the coming years. The traditional dominance of TSMC is being challenged, and the future of chip manufacturing is likely to be more diversified and competitive.