Brad Gerstner, Founder and CEO of Altimeter Capital, provides a deep dive into the 2026 market landscape, focusing heavily on the "AI Super Cycle," infrastructure demands, and the diverging paths of software companies.
[00:02:22]() - AI Infrastructure: The $500B Capex Super Cycle
Disclaimer: Orignal content owned by or sourced from third parties. It does not represent the views of 'Nuggets' platform or it's team. AI is used extensively across this platform including for summaries. Accuracy is not guaranteed, there can be mistakes. Any info or content on this platform is not a financial, legal, or investment advice. Do your own research. Refer for complete disclosures:- Terms of Use · Full Disclaimer
Gerstner describes 2026 as a "stock picker's market," distinguishing it from the broad recoveries seen in 2023 and 2024 [00:22].
Key Drivers: Lower taxes, a rate-cutting cycle, controlled inflation, and earnings growth [01:01].
Investment Themes: He highlights a massive investment super cycle driven by re-industrialization and the onshoring of national supply chains [01:16].
Corporate Efficiency: Companies are leveraging AI to become "more fit," leading to a "golden age of margin expansion," despite the challenges this creates for the employment picture [01:30].
The AI Infrastructure Super Cycle
Gerstner emphasizes that the market is still underestimating the scale of the AI buildout.
Capex Explosion: In 2023, major tech companies spent $150 billion on capex; in 2026, that figure is expected to exceed $500 billion [02:44].
Power Constraints: The bottleneck for AI deployment is no longer just chips, but "powered shell"—the physical infrastructure and energy required to run data centers [03:03]. For example, Amazon is expected to deploy 6 gigawatts of power this year [03:08].
Top Picks: Altimeter’s biggest bets are in AI infrastructure, including NVIDIA, TSMC, SK Hynix, Samsung, Amazon, Microsoft, Coreweave, and Google [02:22].
NVIDIA and the Hardware Race
Valuation: Gerstner argues NVIDIA is not "demanding" at 25-26x earnings, noting it peaked at 41x in July 2024 and troughed at 21x in April 2025 [01:55].
Performance Leaps: Jensen Huang's recent keynote revealed that the new Reuben chips improve training performance by 3.5x and inference performance by 5x [15:21]. This allows for Blackwell-level performance at 1/5th the cost [15:34].
Competitive Moat: He believes NVIDIA's rapid innovation makes it extremely difficult for custom ASICs (from Meta or OpenAI) or competitors like AMD to catch up [15:55].
Cloud Giants and AI Private Players
Alphabet (Google): After being critical of Google's slow start in 2024, Gerstner praises their turnaround. He notes Gemini's dramatic improvement and points out that ChatGPT still maintains over 900 million weekly average users [05:33].
Amazon: Gerstner calls Amazon "super cheap" and expects a breakout year [09:32]. AWS is accelerating as it integrates more NVIDIA chips and benefits from its investment in Anthropic.
Anthropic & OpenAI: He predicts both companies will eventually go public to allow retail investors a chance to participate [10:32]. He dismisses rumors of an Amazon acquisition, stating the founders likely wish to remain independent [09:10].
Coreweave: Altimeter has added to its position in Coreweave, viewing it as a superior "Neocloud" due to its strategic importance to NVIDIA and execution capabilities [04:10].
The Software "Crisis" and Opportunities
Gerstner delivers a stark warning about the software sector, which is trading at a two-decade low multiple (less than 5x forward revenue) [16:13].
The 90% Rule: He claims 90% of software companies deserve to be down because their business models (seat-based subscriptions) are being disrupted by AI-generated applications [18:22, 20:16].
The Winners (The 10%): The "smart money" is looking for the 10% that are AI-durable. He identifies Snowflake and Databricks as key beneficiaries because they provide the "data substrate" required for agentic AI solutions [18:53].
Shift in Compute: He asserts that all future software must be "accelerated" (running on GPUs/TPUs) to produce intelligent "thinking tokens" rather than just being "dumb" pre-compiled software running on a CPU [12:21].
Key Figures & Facts Summary
NVIDIA Multiple: 25x-26x (Current) vs. 41x (July '24 Peak) [02:03].
2026 (Projected): Over $500 billion in non-speculative purchase orders [02:44].
Power Deployment: Amazon alone is expected to deploy 6 gigawatts of power for data centers this year [03:08].
Anthropic Spend: AWS is the beneficiary of approximately $5 billion of Anthropic’s spend, a number expected to grow [08:08].
NVIDIA/Microsoft Deal: A massive deal in the works (referenced regarding Anthropic) could be worth upwards of $50 billion this year [08:57].
Software Sector Metrics
Valuation Collapse:
Current Forward Revenue Multiple:4.2x (a generational low) [17:54].
Peak Forward Revenue Multiple (2021): Almost 17x [17:54].
Survival Rate: Gerstner estimates that 90% of the software companies currently down deserve to be down; only the remaining 10% are "AI durable" [20:16].
Consumer & Enterprise AI Data
ChatGPT Reach: Currently has upwards of 900 million weekly average users [05:33].
Google Turnaround: Despite a slow start, Google's stock was up 65% in the previous year [05:07].
Inference Vision: Jensen Huang predicted 1 billion-x inference growth a year ago, which Gerstner believes was actually an understatement [12:47].
Anecdotes & Quotes
The "DeepSeek Scare": Mentioned as a moment at the beginning of last year where the market mistakenly thought one model could do everything, temporarily cooling AI investment [12:41].
Stan Druckenmiller Anecdote: Gerstner cites his hero Druckenmiller, noting his greatest strength is that he "forgets the last shot," emphasizing the need for mental flexibility in investing [05:52].
The " hour-long CRM" Anecdote: Gerstner describes the current fear in software: a belief that a user can go to Claude Code or OpenAI's Codex and build a custom CRM in about an hour, rendering traditional software seats obsolete [18:11].
"Brookfield's the largest infrastructure owner in the world... We drew a pipeline and we showed all the different components of the payments ecosystem on a pipeline and said it's like a pipe that moves any commodity except what it's moving…