[trading places]

Nvidia $4.4T vs Cerebras $23B Val | Why Private Prices Are Wrong with Nick Fusco of PM Insights | Episode 27

March 18, 2026

Key Takeaways Copied to clipboard!

  • Anthropic's reported ARR growth ($5B added in February) suggests it is rapidly catching up to or potentially surpassing OpenAI in momentum, despite OpenAI's larger valuation. 
  • Figure Robotics' $39B valuation is considered extremely ambitious and potentially unjustified given its focus on humanoid form over broader robotics function and low current revenue. 
  • The influx of massive capital into top-tier VC firms (like General Catalyst, Founders Fund, Spark Capital) is concentrating resources toward capital-intensive sectors like AI and defense, potentially squeezing emerging market managers. 
  • Increased transparency and data from secondary markets are making VC analysis easier, allowing investors to gain comfort with illiquid asset exposure and better track true market valuations. 
  • Cerebras, despite a recent $23B valuation, faces significant risk due to extreme customer concentration, with 83-87% of its revenue coming from a single customer, G42, which could jeopardize a successful IPO under current terms. 
  • The market is trending toward public-private crossover, evidenced by new funds like the Robinhood Private Markets Venture Fund, as institutional investors seek pure-play AI investment opportunities before major private players like OpenAI or Anthropic go public. 

Segments

Private Mag 7 Discussion
Copied to clipboard!
(00:00:50)
  • Key Takeaway: SpaceX, Anthropic, and OpenAI form the core of the private ‘Mag 7,’ with Databricks and Stripe also being strong contenders.
  • Summary: The hosts debate the composition of the private ‘Mag 7,’ noting that SpaceX, Anthropic, and OpenAI are difficult to argue against based on size and momentum. Databricks and Stripe are considered strong inclusions for the top six. XAI’s inclusion was questioned before its absorption into SpaceX.
OpenAI vs. Anthropic Growth
Copied to clipboard!
(00:04:26)
  • Key Takeaway: Anthropic demonstrated superior recent growth, adding $5 billion in ARR in February alone, making it a potentially better trade than OpenAI.
  • Summary: OpenAI’s current cash flow dynamics and fundraising efforts are viewed as shaky compared to Anthropic’s performance. Anthropic added $5 billion in ARR in February, significantly outpacing Databricks’ annual ARR generation. This rapid growth suggests Anthropic is gaining traction against OpenAI.
Figure Robotics Valuation Critique
Copied to clipboard!
(00:05:25)
  • Key Takeaway: Figure Robotics’ $39B valuation is deemed too high, suggesting an overemphasis on humanoid form (‘figure’) rather than functional, non-humanoid robotics solutions.
  • Summary: The $39B valuation for Figure Robotics, up from $3B, is criticized as ambitious based on current revenue. The principle ‘form follows function’ suggests robot design should prioritize utility over mimicking the human form. Broader robotics investment may be more warranted than focusing solely on humanoid personal assistants.
Geopolitical Impact on Markets
Copied to clipboard!
(00:08:02)
  • Key Takeaway: Conflict in Iran is increasing market volatility and slightly raising inflation forecasts (Goldman Sachs revised PCE up to 2.9%), potentially delaying IPOs.
  • Summary: The conflict in Iran has increased day-to-day volatility, pushing Brent Crude prices between $85 and $115, which impacts inflation predictions. Goldman Sachs raised its 2024 PCE inflation forecast from 2.2% to 2.9% due to disrupted trade in the Strait of Hormuz. Prediction markets suggest the conflict may not resolve or normalize Strait of Hormuz traffic until May or later.
Tech Layoffs and Efficiency Metrics
Copied to clipboard!
(00:14:55)
  • Key Takeaway: Meta is achieving best-in-class revenue per employee ($2.5M) through efficiency efforts, contrasting sharply with Oracle’s low efficiency ($0.4M) amid heavy AI CapEx.
  • Summary: Oracle is facing pressure due to its high debt-to-market cap ratio and negative free cash flow resulting from massive AI infrastructure commitments (Stargate initiative). Meta has significantly improved operating efficiency, moving from $1.4M revenue per employee in 2015 to $2.5M currently, setting a new standard. Investors will increasingly scrutinize revenue per employee metrics across the tech sector.
VC Mega-Fund Raising Trends
Copied to clipboard!
(00:19:43)
  • Key Takeaway: Mega-platform VCs (>$300B AUM) are raising capital at a rate disproportionate to their current market share, concentrating funds toward capital-intensive AI/defense bets.
  • Summary: Firms like General Catalyst ($10B), Founders Fund ($6B), and Spark Capital ($3B) recently announced very large fundraises, indicating a trend toward larger funds. These large funds are necessary to pursue capital-intensive sectors like AI and defense, moving away from capital-light strategies. Emerging market managers are facing a tough cycle as capital concentrates at the top.
Robotics Investment Surge
Copied to clipboard!
(00:28:11)
  • Key Takeaway: Venture investment in robotics surged over 70% year-over-year to $40 billion last year, highlighted by Travis Kalanick’s rebranding of Cloud Kitchens to Atoms, focusing on ‘gainfully employed robots.’
  • Summary: Travis Kalanick’s Atoms is focusing on specialized, productive robots for mining and transport, reinforcing the idea that robots do not need to be humanoid to be useful. Mind Robotics, a Rivian spin-out, raised $500M at a $2B valuation to build AI-powered robotics. Replit also saw a massive valuation jump to $9B after projecting $1B ARR by year-end.
Private Market Pricing Discrepancies
Copied to clipboard!
(00:33:28)
  • Key Takeaway: Significant valuation dispersion exists among private market participants, exemplified by Databricks shares being marked at $102 by one party and $58 by another, often due to differing assumptions or investor type (institutional vs. retail).
  • Summary: PM Insights focuses on providing market truth based on willing buyers and sellers, avoiding model assumptions common in traditional valuation methods. They track data showing that institutional transactions (>$5M) are typically more aligned with fair value than smaller, retail-sized transactions. GPs should generally not hold stale marks (pre-2022 valuations) for more than two years due to public market repricing.
VC Liquidity and Data Transparency
Copied to clipboard!
(01:01:54)
  • Key Takeaway: Secondary market players enhance VC liquidity, providing data that validates valuations and allows for more comfortable, alpha-generating portfolio exposure.
  • Summary: Analysis in the private market is improving due to increased transparency and data availability. Secondary platforms address the historical illiquidity of VC access by demonstrating true liquidity. This data allows investors to justify larger portfolio allocations with confidence in the underlying asset valuation.
Public-Private Crossover Trend
Copied to clipboard!
(01:02:43)
  • Key Takeaway: The launch of the Robinhood Private Markets Venture Fund signals a growing trend of public-private crossover as fund complexes respond to high investor demand for private stock exposure.
  • Summary: The launch of the Robinhood Private Markets Venture Fund indicates a clear demand for access to private assets from public-facing platforms. This trend suggests that fund complexes are actively creating products reflective of this demand. This movement points toward increased virtualization or tokenization of stocks in the future.
Cerebras Valuation Corner Introduction
Copied to clipboard!
(01:04:17)
  • Key Takeaway: Cerebras achieved a $23 billion valuation in a recent private round led by Tiger, significantly up from its $8 billion valuation during a pulled IPO attempt in late 2024.
  • Summary: Aman Verjee introduces the ‘Valuation Corner’ segment focusing on Cerebras, which recently secured a $23 billion valuation with Tiger leading the round. This follows a failed IPO attempt in October 2024, where they had secured an $8 billion valuation via a private round after pulling the public offering. Cerebras designs AI infrastructure chips, claiming to be significantly faster and more efficient than NVIDIA’s GPUs using WaferScale technology.
Cerebras IPO Projections and Risks
Copied to clipboard!
(01:06:20)
  • Key Takeaway: Cerebras projects strong revenue growth (potentially 5x to $1B in 2025) justifying a valuation multiple similar to NVIDIA, but faces an existential risk due to customer concentration with G42.
  • Summary: If Cerebras hits a $1 billion revenue run rate in 2025, its valuation implies a multiple of 23 times forward revenues, aligning with NVIDIA’s current multiple. The company shows massive top-line acceleration, projecting nearly five times revenue growth year-over-year. However, the S1 filing reveals that Group 42 Holdings accounted for 83% to 87% of total revenue in 2023 and the first half of 2024, creating severe concentration risk.
G42 Concentration and Geopolitical Risk
Copied to clipboard!
(01:08:02)
  • Key Takeaway: The heavy reliance on G42, a Middle Eastern entity, exposes Cerebras to renegotiation risks and potential geopolitical export control issues, which must be mitigated before a successful IPO.
  • Summary: The dependence on G42 presents risks such as contract renegotiation or customer switching to competitors like NVIDIA. Unlike NVIDIA’s customer base of hyperscalers like Microsoft and Meta, Cerebras’s primary customer is G42. Geopolitical instability in the Middle East could lead to export controls, posing an existential threat to Cerebras’s business model.
Dilution from Preferred Stock Agreement
Copied to clipboard!
(01:09:29)
  • Key Takeaway: A sweet deal granted to G42, allowing them to purchase shares at a $14.66 strike price via preferred stock, will cause significant shareholder dilution upon IPO if the share price exceeds this low strike price.
  • Summary: G42 holds the right to buy 23 million shares of Series F2 convertible preferred stock at only $14.66 per share. This strike price is significantly lower than the $36.23 per-share price from the $8.1 billion raise in 2024. If the IPO price is high, G42 will exercise this option, leading to greater dilution for existing Cerebras shareholders.
Cerebras IPO Outlook and Timing
Copied to clipboard!
(01:10:38)
  • Key Takeaway: Cerebras needs to broaden its customer footprint beyond G42 to demonstrate durable revenue before attempting an IPO, as it currently offers one of the few pure-play AI stories available to institutional investors.
  • Summary: The speaker suggests Cerebras must announce broader strategic partnerships, perhaps with U.S. hyperscalers, to mitigate the concentration risk before going public. Cerebras is currently a pure-play AI story, unlike Google or Meta, making it attractive to institutional investors seeking AI exposure. Success hinges on demonstrating revenue durability outside of the Middle East customer base.