Nvidia Earnings Brush Off AI Bubble Fears — For Now
Timestamps are as accurate as they can be but may be slightly off. We encourage you to listen to the full context.
Podcast Summary
This episode of Prof G Markets features Ed Elson analyzing NVIDIA's record-breaking Q3 earnings of $35.6 billion in revenue, with data center sales up 66% year-over-year. (00:45) Gil Luria from D.A. Davidson discusses whether these earnings put AI bubble concerns to bed, warning about excessive borrowing by companies like CoreWeave to purchase GPUs. The episode also covers Meta's antitrust victory, with former DOJ official Jonathan Kanter expressing concern about the ruling that dismissed Meta's monopoly case based on current competition from TikTok and YouTube. (17:04) Finally, Ed critiques Trump's meeting with Saudi Crown Prince Mohammed bin Salman, questioning the substance behind announced trillion-dollar investment commitments.
- Main theme: Examining AI market sustainability, big tech antitrust enforcement, and geopolitical business relationships through the lens of recent NVIDIA earnings, Meta's legal victory, and US-Saudi investment announcements.
Speakers
Ed Elson
Ed Elson is the host of Prof G Markets, a financial and business analysis podcast. He provides market commentary and interviews industry experts on technology, antitrust, and economic policy matters.
Gil Luria
Gil Luria serves as Head of Technology Research at D.A. Davidson, where he analyzes semiconductor and technology companies. He provides investment research and market insights on major tech stocks including NVIDIA and AI infrastructure companies.
Jonathan Kanter
Jonathan Kanter is the former Assistant Attorney General for the Antitrust Division of the US Department of Justice. He led major antitrust cases against big tech companies during his tenure and continues to provide expert commentary on competition policy and enforcement.
Key Takeaways
Distinguish Between Real AI Winners and Speculative Players
Gil Luria emphasizes the critical importance of separating legitimate AI companies from those engaging in unsustainable financial practices. (04:13) While companies like Amazon, Microsoft, Google, and Meta are purchasing GPUs with cash flow and retained earnings, others like CoreWeave are borrowing heavily to fund their operations. Luria warns that companies borrowing 90-100% of their capital to build speculative AI infrastructure while generating insufficient profits to cover interest expenses represent dangerous bubble territory. This creates a precarious situation where leveraged players could default, potentially dragging down the entire ecosystem.
NVIDIA's Guidance Provides Unprecedented Market Visibility
For the first time since the AI era began, NVIDIA provided guidance extending five quarters out, projecting $37 billion in Blackwell and Rubin chip sales between this year and next. (06:05) This extended visibility, compared to their typical one-quarter guidance, gives investors confidence that the data center buildout will continue through 2026. However, Luria notes this doesn't eliminate cyclical concerns - semiconductors remain a cyclical industry where peaks are eventually followed by declines, especially when demand is artificially inflated through excessive borrowing.
Antitrust Enforcement Requires Faster Action
Jonathan Kanter argues that the Meta antitrust case's failure highlights the critical importance of blocking problematic acquisitions in real-time rather than attempting to unwind them years later. (21:32) The FTC should have blocked Instagram and WhatsApp acquisitions in 2012 and 2014 when the evidence was available, instead of filing suit five years later. Kanter emphasizes that waiting 10-15 years to address anticompetitive behavior allows companies to become entrenched, making remedies more difficult and less effective.
Courts Are Applying Problematic Legal Standards to Big Tech Cases
The Meta ruling demonstrates a concerning trend where courts evaluate monopoly power based on current market conditions rather than when the alleged anticompetitive conduct occurred. (23:17) Judge Boesburg ruled that Meta's acquisitions weren't illegal because TikTok and YouTube now compete with Facebook, ignoring the monopolistic behavior from 2012-2014. This creates a dangerous precedent where companies can engage in anticompetitive acquisition strategies without consequences, as long as future competition eventually emerges from other sources.
Investment Announcements Without Legal Framework Are Marketing Stunts
Ed Elson's analysis of the Saudi Arabia investment announcement reveals a pattern of substance-free deals in the current administration. (33:54) The trillion-dollar Saudi investment commitment lacks any treaty, contract, written terms, or signed agreements - making it essentially a press release rather than a genuine business deal. This mirrors previous announcements from Japan, Europe, and China that never materialized as promised, highlighting the importance of scrutinizing the legal framework behind high-profile international business commitments.
Statistics & Facts
- NVIDIA reported record Q3 revenue of $35.6 billion, representing a 94% increase year-over-year, with data center sales up 66% from the previous year. (02:52) Gil Luria noted this demonstrates continued strong demand for AI infrastructure, though he cautioned about sustainability given the borrowing patterns of some customers.
- Over $100 billion in loans have been made to build data centers, mostly at high interest rates, according to Gil Luria's analysis. (08:21) This debt accumulation represents a potential systemic risk if compute prices decline and these leveraged data centers cannot service their interest payments.
- Saudi Arabia's entire sovereign wealth fund is worth less than the trillion dollars they committed to invest in the US, as Ed Elson pointed out. (35:35) This mathematical impossibility underscores the lack of substance behind the investment announcement.
