Apple Weighs Using Intel, Samsung Processors | Bloomberg Tech 5/5/2026

Key Concepts

• Semiconductor Supply Chain Diversification: Apple’s strategic shift to reduce reliance on TSMC (Taiwan) by exploring US-based manufacturing with Intel and Samsung.
• AI Model Governance: The expansion of US government pre-release review agreements to include Alphabet, Microsoft, and xAI.
• Agentic AI: AI systems capable of performing autonomous tasks, currently a major focus for companies like Anthropic and Palantir.
• Ontology (Data Integration): Palantir’s specialized framework for mapping siloed enterprise data into a single, real-time "source of truth."
• Fit-for-Purpose Models: The trend of using smaller, specialized AI models rather than expensive, general-purpose models to achieve higher efficiency and lower costs.
• Zero-Day Vulnerabilities: Previously unknown security flaws that AI can now identify and exploit at high speeds.

---

1. Apple’s Supply Chain Rethink

Apple is actively seeking to diversify its processor manufacturing beyond Taiwan (TSMC) due to geopolitical risks and supply chain vulnerabilities.

• Strategy: Apple is in exploratory talks with Intel and Samsung to utilize their new US-based fabrication plants (fabs) for A-series and M-series chips.
• Motivation: Mark Gurman notes that relying on a single geography (Taiwan) for 90% of global chip production is a significant risk. While Apple has invested in TSMC’s Phoenix, Arizona facility, they require additional capacity and geographic redundancy to avoid potential tariffs and geopolitical disruptions.
• Political Context: The move aligns with US government efforts to bolster domestic semiconductor manufacturing, with Donald Trump recently highlighting Intel’s growth and investment.

2. AI Governance and Regulation

The US Department of Commerce is formalizing oversight of AI development.

• The Agreement: Alphabet, Microsoft, and xAI have joined OpenAI and Anthropic in agreeing to provide the US government with early access to their models for pre-release review.
• Scope: Maggie Eastland clarifies that this center (within the Commerce Department) focuses on evaluation rather than direct regulation. However, it signals a growing appetite for oversight, potentially paving the way for future cybersecurity executive orders.

3. Market Trends: AI Execution vs. Panic

Investment banking experts at Piper Sandler observe a shift in investor sentiment.

• From Panic to Discernment: Investors are moving away from general "AI anxiety" toward identifying companies with clear execution strategies.
• Enterprise Moats: Companies with deep enterprise integration (where "rip and replace" is difficult) are viewed as safer bets.
• Defense Tech: There is a significant rotation into defense technology, particularly firms that combine hardware with software, driven by increased national security spending.

4. Corporate Spotlights

• Pinterest: CEO Bill Ready reported 11 consecutive quarters of record users (630M+). Pinterest is positioning itself as an "AI-powered shopping assistant." By using "fit-for-purpose" models trained on their unique visual data, they claim 30% better relevancy than general-purpose models at 10% of the cost.
• Grab: The Southeast Asian "super app" reported 24% GMV growth. CFO Peter Oey highlighted that despite fuel price volatility, the company is leveraging EV adoption and lending programs to maintain a healthy marketplace.
• Palantir: Despite strong earnings, the stock faced pressure due to high valuation multiples and concerns over US commercial growth. Executive VP Josh Harris emphasized that Palantir’s "ontology" layer is essential for the next generation of "agentic AI," providing the reliable data foundation that general-purpose models lack.

5. Cybersecurity in the Age of AI

Seth Boro of Thoma Bravo discussed the escalating threat landscape.

• Speed of Threats: AI allows attackers to find "zero-day" vulnerabilities in minutes—a process that previously took humans years.
• Layered Defense: Portfolio companies like ProofPoint and Darktrace are using network effects (analyzing millions of malicious emails) to detect and respond to threats faster than ever.
• Inference Costs: Boro noted that while current AI inference costs are high, the industry is moving toward specialized, efficient models rather than relying solely on expensive general-purpose LLMs.

---

Synthesis/Conclusion

The tech sector is currently defined by a transition from the "hype" phase of AI to a phase of operational execution and risk management. Key themes include the urgent need for geographic diversification in hardware (Apple), the formalization of AI safety protocols with the US government, and a focus on "fit-for-purpose" AI models that prioritize efficiency and specific commercial utility over general-purpose capabilities. While market valuations remain high, investors are increasingly discerning, favoring companies that can demonstrate a clear "moat" through proprietary data, enterprise integration, or specialized AI applications.