Nvidia Earnings in Focus; SpaceX Eyes Historic IPO | Bloomberg Tech 5/20/2026

Key Concepts

• AI Infrastructure & Compute: The shift from training to inferencing and the resulting demand for high-performance GPUs.
• Supply Chain Bottlenecks: Constraints in memory (HBM), networking, and silicon photonics.
• Custom Silicon (ASICs): The rise of proprietary chips (like Google’s TPUs) competing with Nvidia’s general-purpose GPUs.
• Geopolitical Overhang: The impact of US-China trade restrictions on Nvidia’s H200 chip sales.
• Re-industrialization: The domestic push to manufacture critical aerospace and defense components in the US.
• Independent AI Evaluation: The need for third-party benchmarking of LLMs regarding factual accuracy and bias.

---

1. Nvidia Earnings and Market Outlook

Nvidia remains the "kingmaker" of the AI trade, with the market closely watching its earnings to gauge the sustainability of the AI boom.

• Financial Expectations: Analysts anticipate 80% topline growth and 85% growth in earnings per share.
• The "Bar": Bloomberg Intelligence notes that a standard "beat and raise" is insufficient; the market is looking for a guide that clears a $90 billion hurdle for the next quarter.
• Valuation: Despite the stock's rally, Nvidia’s forward P/E ratio has compressed due to astronomical earnings growth, making it appear relatively "cheap" compared to historical averages.
• The "Buy the Rumor, Sell the News" Phenomenon: Historically, Nvidia shares have often fallen the day after earnings, even following blowout reports, as investors take profits.

2. SpaceX IPO and Strategic Acquisitions

SpaceX is moving toward a potential IPO, with reports suggesting a valuation of over $2 trillion and a $75 billion capital raise.

• Banking Structure: Goldman Sachs is reportedly in the "lead left" position, though Morgan Stanley is also heavily involved. Historical precedents (like the Alibaba IPO) suggest the top five banks may receive similar fees despite the "lead" designation.
• Cursor Acquisition: SpaceX plans to acquire the coding startup Cursor within 30 days of its IPO. The deal is structured as an all-cash transaction with a $10 billion breakup fee.
• Strategic Synergy: The acquisition is designed to integrate Cursor’s coding capabilities with xAI’s massive compute infrastructure (specifically the "Colossus 2" data center), creating a vertically integrated AI-coding powerhouse.

3. Re-industrialization: The Case of AMCA

AMCA, a company that recently raised $300 million at a $1 billion valuation, exemplifies the trend of domestic manufacturing for critical defense and aerospace needs.

• Methodology: Unlike standard contract manufacturers, AMCA designs, qualifies, and manufactures critical components (e.g., sensors, capacitors, wire harnesses) that are currently single-sourced from offshore locations.
• Strategic Goal: To bridge the gap between US defense requirements and the dwindling domestic supply base for high-precision electronic components.

4. AI Accuracy and Geopolitics

Forum AI’s "Newsbench" study highlights significant failures in current LLMs regarding news and political information.

• Key Findings: 90% failure rate on election-related questions across four major chatbots.
• Bias: Claude and Gemini showed a left-leaning bias in 100% of election-related tests, while ChatGPT showed 95% and Grok showed 85% right-leaning bias.
• The "Grading Homework" Problem: Campbell Brown (CEO of Forum AI) argues that AI companies currently grade their own performance. She advocates for an ecosystem of independent, third-party evaluators to ensure factual accuracy, especially as enterprise and government sectors begin to rely on these models.

5. Global Chip Ecosystem Perspectives

Paulina McPatteren (Baillie Gifford) provided an international view on the semiconductor supply chain:

• TSMC’s Dominance: TSMC remains the most critical player due to its ability to coordinate a complex, globalized supply chain. Their guidance of 56% AI-related growth over the next five years serves as a key indicator for the industry.
• Memory (SK Hynix): The memory market is shifting from a cyclical, commodity-based model to one defined by long-term purchase agreements and "lock-in" effects due to the specialized nature of High Bandwidth Memory (HBM).

---

Synthesis and Conclusion

The current tech landscape is defined by a massive disconnect between the skyrocketing demand for AI compute and the physical constraints of the global supply chain. While Nvidia remains the central pillar of this growth, the market is shifting its focus toward inferencing efficiency and custom silicon. Simultaneously, the private sector is aggressively pursuing vertical integration (e.g., SpaceX/xAI/Cursor) to secure compute and talent. As AI models become integrated into critical infrastructure, the industry faces a growing mandate for independent verification of accuracy and bias, moving beyond the "self-grading" model that has characterized the early phase of the AI boom.