Consider the upcoming Figma IPO, a high-quality software company with a sticky product, valued at a reasonable 15x revenue multiple. A more durable investment strategy is to focus on the "picks and shovels" of the AI boom, specifically the infrastructure all AI companies require. The demand for electricity to power data centers is expected to surge, making utilities and energy producers attractive long-term investments. This massive buildout also directly benefits companies that supply the core components for data centers. The most direct public company play on this trend remains the primary GPU supplier, NVIDIA (NVDA).

Anthropic (Pre-IPO)

• Valuation: The company is reportedly seeking a new funding round at a $150 billion to $170 billion valuation, led by a Middle Eastern sovereign wealth fund.
  • This is a significant increase from its last funding round in March, which was at a $61.5 billion valuation (a 143% increase in about four months).
• Financials: Anthropic is reported to have $4 billion in Annual Recurring Revenue (ARR).
  • The new valuation represents a 37x revenue multiple.
  • Revenue has grown 4x since January, an increase of $3 billion in ARR in just six months.
• Sentiment & Strategy: The sentiment among the hosts is mixed but leans positive, contingent on continued growth.
  • Bullish Case: The product, Claude, is considered "world-class" and is a preferred model for important tasks among some hosts. The company is effectively targeting specific verticals, such as its new Claude for financial services, which partners with Databricks, FactSet, Morningstar, PitchBook, and S&P Global. This is seen as a direct challenge to incumbents like Bloomberg. The stickiness of being integrated with Amazon Web Services (AWS) is also seen as a strong competitive advantage (a "moat").
  • Bearish Case / Risks: The 37x sales valuation is described as "absurd" by one host, who questions if the "insane growth rate" can be maintained. Unlike competitors OpenAI and xAI, Anthropic does not own its own GPU infrastructure, instead partnering with cloud providers. This strategy has led to significant shareholder dilution in past funding rounds.

Takeaways

• Anthropic is a high-growth, high-valuation play in the AI large language model (LLM) space. Its valuation is heavily dependent on its ability to maintain its extraordinary revenue growth.
• The company's strategy of targeting specific, high-value industries like finance could be a key driver of future growth. The partnerships with major financial data providers (Databricks, FactSet, S&P Global, etc.) are a strong validation of this approach.
• Investors should weigh the "best-in-class" product and strong enterprise strategy against the extremely high valuation and the competitive risks from players like OpenAI and Google.
• The company's reliance on partners like Amazon (AMZN) for infrastructure is a key strategic difference from its peers. This creates a symbiotic relationship but also a dependency.

xAI (Pre-IPO)

• Infrastructure Strategy: xAI has entered a deal where its venture capital investor, Valor, will purchase $12 billion worth of GPUs (estimated at 350,000 units) and lease them back to xAI.
  • This is viewed as a "creative financing" structure that helps xAI acquire necessary hardware without using its own capital, potentially circumventing covenants from previous debt raises.
  • It's also a strategic move to take a large number of GPUs off the market, making them unavailable to competitors.
• Scale Ambitions: Elon Musk aims to have 1 million GPUs for training its models.
  • The podcast notes that Grok 4 was trained on less than 200,000 GPUs and is already considered a very good model, implying a model trained on 1 million GPUs could be exceptionally powerful.
• Training vs. Inference: An important distinction was made: the 1 million GPUs are for training the models. The actual day-to-day running of the models (inference) is done on rented chips from cloud providers, which could include chips from the semiconductor company Grok (GROQ).

Takeaways

• xAI is making a massive, capital-intensive bet that scaling up GPU count will lead to a superior AI model. This is a high-risk, high-reward strategy.
• The creative leasing deal with Valor highlights the intense competition for GPUs and the innovative financial engineering required to secure them.
• The separation of training and inference infrastructure is a key insight. While xAI is building a massive training cluster, it still relies on the broader cloud ecosystem for inference, creating opportunities for other chip makers like Grok (GROQ) and cloud providers.

Figma (Pre-IPO, soon to IPO)

• IPO Details: Figma is expected to begin trading on Thursday with an IPO valuation between $15 billion and $16 billion.
• Valuation Context: This is below the $20 billion acquisition offer from Adobe in 2022, which was blocked by regulators. However, the hosts note that an acquisition price is different from a funding or IPO valuation.
  • With approximately $1 billion in ARR, the IPO valuation is around a 15x revenue multiple.
• Business Health: The company is viewed as a "solid" and "strong business" with a very "sticky" product that is essential for technology companies and designers.
  • Its revenue growth has recovered to nearly 50% after a slowdown in 2022-2023, which was attributed to cost-cutting by its tech and startup customer base.
• Secondary Market Data: On private secondary markets, Figma's valuation peaked near $18 billion a few months ago and has since settled to around $14 billion. The IPO pricing is seen as reasonable in this context.

Takeaways

• Figma is a mature, high-quality software company with a strong user base and a sticky product, making it a potentially attractive IPO for investors looking for exposure to the creator economy.
• The 15x revenue multiple is considered "rich" but potentially justified by its market leadership and recovering growth.
• While AI is a potential long-term risk, the hosts believe it's more likely that Figma will successfully integrate AI into its platform, further strengthening its product and market position.
• The IPO's performance will be a key indicator of the market's appetite for new tech listings.

Investment Theme: AI Infrastructure

• Core Idea: One of the most compelling investment strategies discussed is to invest in the "picks and shovels" of the AI boom rather than trying to pick the winning AI model. This means focusing on the fundamental infrastructure that all AI companies need.
• Electricity: This was highlighted as a potential "goldmine."
  • The demand for electricity to power data centers and GPUs is expected to "go through the roof."
  • An estimate from former Google CEO Eric Schmidt suggests the AI industry will need 90 gigawatts of power (equivalent to 90 nuclear power plants).
  • Electricity prices are already up 35% year-over-year, indicating a strong inflationary trend in this sector.
• GPUs & Data Centers: The intense competition for GPUs, exemplified by the xAI/Valor deal, underscores their critical importance. Companies that build and supply the components for data centers are direct beneficiaries. NVIDIA (NVDA) is the primary example.
• Political Tailwinds: A recent executive order from the Trump administration was mentioned, which is seen as a "green light" to fast-track the buildout of AI infrastructure, including deregulation for energy production.

Takeaways

• Investing in the AI infrastructure supply chain, particularly electricity generation and transmission, could be a more durable and less speculative way to play the AI revolution.
• The massive capital expenditure required for this buildout ($2 trillion estimated by Apollo) suggests a long-term investment cycle.
• Regulatory and political shifts appear to be aligning to accelerate this buildout, potentially reducing hurdles for energy and construction projects. Look for opportunities in utilities, energy producers (especially natural gas), and industrial companies involved in building data centers.