• Treat Anthropic as a potential IPO opportunity only if final pricing is attractive: podcast estimates ranged from $1.5 trillion–$1.6 trillion at offering to about $1.7 trillion after trading, but growth, cash needs, and execution remain key risks.
• Approach SpaceX cautiously around its IPO: shares reportedly swung from $135 to nearly $220 and back toward $150, so wait for evidence that its AI and data-center plans translate into reported revenue.
• Favor the broader AI compute, hardware, and data-center theme selectively; prioritize companies that can secure GPUs and power and turn capacity into durable revenue rather than relying on valuation momentum.

Anthropic

• The speakers viewed Anthropic’s S-1 as broadly positive: they cited revenue growth, a more diversified customer base, and cash burn they considered manageable given the company’s cash.
• They said the reported $80 billion net loss included $72 billion in accounting charges related to multi-year data-center contracts and commitments. They characterized the remaining $8 billion as an operating loss for the first half of the year, against $130 billion in cash on the balance sheet.
• Valuation and IPO expectations varied:
  • One speaker said a $2 trillion IPO valuation seemed realistic and predicted the stock could rise after listing.
  • Another said the secondary market was around $1.55 trillion and suggested an IPO price of $1.5 trillion to $1.6 trillion could be more reasonable, potentially settling near $1.7 trillion after early trading volatility.
  • A private-market investor discussed on the podcast viewed buying at a discount as a possible trade, hoping to sell after lockups ended; the hosts noted that this strategy had many moving parts.
• The speakers said Anthropic would need to keep growing revenue and demonstrate that its growth story is achievable. They pointed to competition in AI assistants and the challenge of securing enough compute as important considerations.
• One speaker argued that AI companies could serve as a hedge for investors with exposure to traditional software, while another stressed that the public market would still require evidence of revenue growth.

Takeaways

• Treat the valuation figures as expectations and opinions discussed on the podcast, not established IPO terms. Compare any eventual offering price with the company’s revenue growth and cash needs.
• The accounting loss was described as largely non-cash, but the discussion still highlighted operating losses, large data-center commitments, competition, and execution as factors to monitor.
• An IPO “pop” or a lockup-related trading strategy is uncertain; the speakers specifically anticipated volatility and differed on how well the stock might perform.

SpaceX

• The speakers described SpaceX as increasingly an AI and data-center business, alongside its space and Starlink operations. They cited plans to build data centers and argued that the company’s ability to secure GPUs, power, and facilities would matter to its growth.
• One speaker said Elon Musk had discussed reaching $100 billion in annualized revenue in December and said they believed that was achievable. The transcript also refers to a projection of growth from roughly $18 billion to $500 billion, though the timing and basis of that figure are not fully clear in the conversation.
• The speakers cited SpaceX’s stock moving from an IPO price of $135 per share to a brief high near $220, then pulling back to around $150. They described early trading and lockup-related supply as potential sources of volatility.
• One speaker expected SpaceX to have more revenue than Anthropic by the end of 2027. They also said the company had been profitable in the prior quarter, while emphasizing that future growth depended on execution and demand for compute.
• The discussion mentioned a “GrokBot” product and Cursor in connection with SpaceX’s AI plans, but did not provide specific revenue figures for them.

Takeaways

• The investment case presented rests heavily on whether SpaceX can deliver its data-center and AI plans, not just on its existing space-related businesses.
• Monitor whether announced revenue expectations translate into reported results, and whether the company can secure compute resources and power as described.
• The podcast’s account of the share-price swings and lockup dynamics underscores the possibility of sharp price movements around an IPO.

OpenAI

• One speaker said OpenAI was targeting a $1.25 trillion valuation and described that as a potentially smart pricing strategy compared with a higher-priced IPO.
• The speakers suggested that a lower offering valuation could attract demand and leave room for gains, but the conversation offered no detailed financial analysis of OpenAI.

Takeaways

• The valuation was discussed as a target, not a confirmed IPO price. Any assessment would depend on the eventual offering terms and the company’s financial results, neither of which were detailed here.

Fireworks AI

• Fireworks AI was described as an AI inference platform. The speakers said it raised a round at $17.5 billion in July and might pursue another round at $30 billion to $40 billion, with one speaker guessing it could land around $35 billion to $40 billion.
• One speaker cited approximately $1 billion in annualized revenue in July and said the company appeared to be demonstrating revenue growth. The podcast did not provide further financial details.
• The speakers saw potential in companies that help customers run open-source AI models and provide inference services.

Takeaways

• The valuation increase could reflect strong investor interest, but the discussion does not establish whether that price is supported by durable growth or profitability. Compare future valuation terms with reported revenue and business performance.

Databricks

• The speakers cited a valuation of about $212 billion, calling Databricks one of the largest pre-IPO companies.
• One speaker estimated that it had been valued around $75 billion to $80 billion roughly a year to a year and a half earlier, while noting uncertainty about the exact figures.
• Databricks was used as a comparison for Fireworks AI and other companies in the AI data and inference ecosystem.

Takeaways

• The podcast presented Databricks as evidence of investor appetite for AI-related infrastructure and data businesses. The valuation history was approximate, so it should not be treated as a precise performance benchmark.

Together AI

• Together AI was described as an AI company in the inference and open-source-model ecosystem. One speaker cited a secondary-market valuation of about $10 billion and said it had more revenue than Fireworks AI.
• The speaker expressed particularly positive interest in the company, but did not provide revenue figures or a specific valuation analysis.

Takeaways

• The discussion points to investor interest in AI inference and open-source model services, but provides too little financial detail to assess Together AI’s valuation independently.

Etched

• Etched was mentioned as a comparison in a discussion of sharp valuation increases. The speakers referred to a move from $12 billion to $21 billion, with consideration of $41 billion, and noted that the company had no revenue at the time discussed.

Takeaways

• The example illustrates the risk of valuing private AI companies primarily on expectations when revenue is absent. The speakers did not offer a specific investment recommendation on Etched.

AI Compute, Hardware, and Data Centers

• The speakers argued that investment attention was shifting from software toward hardware, compute, and data centers.
• They said AI companies would need access to GPUs, power, and data-center capacity to deliver their products. One speaker argued that Anthropic might need to become more closely tied to hardware and compute to make its models more capable.
• They also suggested that demand for compute could be substantial, while noting that companies still need to build capacity and convert demand into revenue.

Takeaways

• The discussion identifies AI infrastructure as a major investment theme, but the opportunity depends on companies’ ability to build capacity, secure power and hardware, and generate revenue from it.
• The speakers’ view that AI could pressure traditional software is a broad market thesis, not a company-specific forecast.

Traditional Software and SaaS

• One speaker argued that investors focused exclusively on traditional software could be exposed to disruption from AI and suggested that AI companies might act as a hedge.
• The conversation also acknowledged competition and uncertainty about how quickly AI products will generate revenue.

Takeaways

• Consider the potential impact of AI on software businesses when evaluating the sector, but the podcast did not identify specific software stocks or quantify the expected effects.