Anthropic IPO | Groq 2.0 & NeoClouds | Jensen Huang's Exit | SpaceX & Grok Bot
Anthropic IPO | Groq 2.0 & NeoClouds | Jensen Huang's Exit | SpaceX & Grok Bot
YouTube33 min 38 sec
Watch on YouTube
Note: AI-generated summary based on third-party content. Not financial advice. Read more.
Quick Insights
  • Consider AI data-center and inference-compute providers for long-term exposure to rising AI demand, but prioritize businesses that can turn contracts into operating capacity and revenue.
  • Treat CoreWeave, Nebius, and Crusoe backlogs as potential demand signals—not proof of future earnings—and scrutinize debt, construction timelines, and execution risk.
  • For Anthropic, wait for post-IPO price action rather than buying immediately; the speakers expect possible volatility in the first 30–60 days, with lockup releases potentially adding share supply.
Detailed Analysis

Anthropic (Private; IPO discussed)

  • The speakers considered a reported $1.4 trillion IPO valuation fair and expected strong demand. They cited a reported six-times oversubscribed offering of about $90 billion and said pricing might come in above the stated range.
  • They pointed to continued buying in private-market secondary shares as evidence of investor appetite.
  • They expect potential volatility in the first 30–60 days after listing. The discussion also highlighted staggered lockup releases and rapid index inclusion as factors that could affect trading and institutional demand.
  • One speaker said they would wait rather than buy immediately, looking for a possible post-IPO dip.

Takeaways

  • Strong reported demand does not eliminate valuation or trading risk. For investors considering a listing, the speakers’ own approach suggests watching post-IPO price action and lockup-related supply before deciding whether to enter.
  • Index additions and early share releases could influence demand and supply, but neither guarantees a sustained share price.

AI NeoClouds and Data-Center Compute

  • The speakers were strongly bullish on AI infrastructure providers that build data centers and sell compute, particularly for AI inference.
  • Their thesis is that demand is growing faster than available capacity. They cited large backlogs and contracts at CoreWeave and Nebius, including some customer prepayments, and argued that providers need to build capacity before they can recognize the associated revenue.
  • They defended the sector’s heavy borrowing and cash use as a way to fund data-center construction and avoid diluting shareholders. They also acknowledged that large spending needs and debt make the businesses difficult for some investors to underwrite.
  • One speaker said the market could face a pullback or bad quarters, but still viewed the buildout as early and expected compute demand to support the companies over time.

Takeaways

  • The opportunity described is exposure to the infrastructure behind AI, rather than only to AI applications. The key question is whether contracted demand turns into operating capacity and revenue on schedule.
  • Debt, capital spending, and execution are material risks raised in the discussion. Backlogs and prepayments are encouraging signals, but they do not by themselves establish that a provider’s economics will work.

CoreWeave (Private; NeoCloud)

  • CoreWeave was presented as a major AI compute provider, with a cited backlog of about $104 billion.
  • The speakers argued that it needs to keep building infrastructure to serve contracted demand, and said debt-funded expansion can make sense if the underlying business economics work.

Takeaways

  • Investors should assess whether CoreWeave can build capacity and convert its backlog into revenue while managing its financing needs. The speakers’ bullish view depends on sustained demand and successful execution.

Nebius (Private; NeoCloud)

  • Nebius was cited as an example of a provider with significant customer commitments. The speakers described customers signing contracts and, in some cases, paying in advance for future compute capacity.
  • They said some contracted compute may not be delivered until 2028, underscoring the long lead time between commitments and service delivery.

Takeaways

  • Prepayments and long-term contracts may improve visibility into demand, but delivery depends on building the required infrastructure. Capacity, timing, and financing are central points to monitor.

Crusoe (Private; NeoCloud)

  • The speakers cited a reported $120 billion backlog for Crusoe and described the sector as racing to bring compute capacity online.

Takeaways

  • The backlog illustrates the scale of demand discussed, but investors should distinguish contracted demand from capacity that is already built and producing revenue.

Groq 2.0 (Private; NeoCloud)

  • The speakers described Groq 2.0 as a separate, data-center-focused business, similar to Lambda Labs, that sells compute to third parties and is focused on inference.
  • They cited an $8 billion valuation and compared it with Together AI, which they said was valued around $8–8.4 billion.
  • One speaker said they might consider investing in Groq 2.0 in the future, after a remaining payment related to the earlier shareholder buyback was completed.
  • The transcript uses “Groq” and “Grok” inconsistently when discussing chips and the related transaction. The speakers describe the new data-center business as distinct from the legacy company and refer to it buying chips from NVIDIA.

Takeaways

  • The speakers see Groq 2.0 as a potential way to participate in AI compute growth. Before evaluating the stated valuation, investors would need to examine its revenue, customer usage, capital requirements, and the details of the separation and remaining payment.

Together AI (Private; AI Compute)

  • Together AI was cited at a valuation of about $8–8.4 billion. One speaker called that valuation “very attractive” in comparison with other companies in the space.

Takeaways

  • That was an individual opinion, not a detailed valuation analysis. The transcript provides no revenue, profitability, or financing figures to support a comparison, so investors would need more information before assessing the claim.

Lambda Labs (Private; AI Compute)

  • Lambda Labs was used as a comparison for Groq 2.0: both were described as businesses that build or operate data centers and sell compute to customers.

Takeaways

  • The transcript offers Lambda as a business-model comparison, but gives no valuation or financial figures for it. The broader sector’s opportunity and risks are discussed under AI NeoClouds and Data-Center Compute.

SpaceX (Private)

  • The speakers were enthusiastic about SpaceX’s data-center and compute ambitions, in addition to its space and Starlink businesses.
  • They highlighted the company’s successful rocket-capture demonstration as evidence of technical execution.
  • One speaker said SpaceX was targeting $500 billion in revenue by the end of the following year, while the other suggested a possible valuation of $500 billion to $1 trillion by then. These were presented as forward-looking claims in the conversation, not established outcomes.
  • They also cited a claimed less-than-one-year return on investment for data-center infrastructure and expected GrokBot subscriptions to contribute to the business.
  • The discussion touched on a possible future role for NVIDIA CEO Jensen Huang at SpaceX, but this was speculative.

Takeaways

  • The investment case described depends on SpaceX executing across several ambitious businesses, including compute and Starlink. The revenue, valuation, and return figures are speaker claims and should be treated as uncertain projections.
  • The transcript does not provide detailed financial data to assess the data-center economics or the expected contribution from subscriptions.

NVIDIA (NVDA)

  • NVIDIA was discussed as the supplier of chips used by AI infrastructure businesses. The speakers said NVIDIA earnings were due that day, but did not provide an earnings forecast or price target.
  • The conversation included claims that CEO Jensen Huang was leaving NVIDIA and speculation about whether he might return, start a company, or move to another business. The speakers said a new company founded by him could likely attract funding and talent, but the discussion was conjecture rather than a concrete investment recommendation.
  • They also discussed the possibility that a departure or new venture could affect NVIDIA’s stock, particularly if employees or investors followed him.

Takeaways

  • NVIDIA’s role as a chip supplier links it to the broader AI-infrastructure theme discussed in the episode. The transcript does not offer a specific valuation view or price target for the stock.
  • The discussion raises leadership and key-person uncertainty as a consideration, but its scenarios about Huang’s next move are speculative. No actual investment recommendation was made.

GrokBot / Grok (Product; No Standalone Ticker Mentioned)

  • The speakers praised GrokBot as a tool for creating websites and products, conducting research, organizing information, and automating recurring tasks.
  • One speaker said they used it for CRM and research workflows and cited a $200 monthly cost for their setup. They argued that wider personal and workplace use could increase demand for compute.
  • The conversation mentioned potential subscription growth but gave no subscriber numbers or revenue figures.

Takeaways

  • The speakers’ enthusiasm supports the broader thesis that AI use could drive demand for compute, but the transcript does not quantify GrokBot’s commercial traction.
  • GrokBot is discussed as a product, not as a separately investable security. Any investment exposure would depend on the ownership and financial structure of the company behind it, which the transcript does not detail.

Tesla (TSLA) and TSMC (TSM)

  • Tesla was mentioned as a possible place Jensen Huang could work or develop a new chip-related venture, with the speakers noting that Tesla already makes AI chips.
  • TSMC was briefly raised as another hypothetical destination for Huang.
  • Neither possibility was presented as confirmed, and the speakers made no specific recommendation about either stock.

Takeaways

  • These were speculative career scenarios, not investment theses. The transcript provides no company-specific financial analysis, valuation view, or price target for Tesla or TSMC.
Ask about this postAnswers are grounded in this post's content.
Video Description
Anthropic is public: priced at $83 above its $70–$78 range, and closed its first day at $97.17, up 17%. Recorded the day before pricing, this episode breaks down why the hosts saw a ~$1.4 trillion target as fair, how a book reported at roughly six times oversubscribed set up the debut, and why a staggered, multi-release lockup schedule and fast-track index inclusion make the institutional plumbing as important as the headline valuation. From there the conversation moves to the compute layer underneath it all. Groq, now a NeoCloud operating GroqCloud after Nvidia's roughly $20 billion licensing deal and acqui-hire, is reportedly seeking a Series B at an $8 billion pre-money valuation. Our house view is that for CoreWeave, Crusoe, Nebius and the next wave of NeoClouds, debt-funded buildout against signed backlog is the business model, not a red flag. CoreWeave's backlog reached $132.5 billion at Q3, Crusoe reports more than $140 billion in contracted value, and Nebius says customer prepayments covered over 100% of the capex on deals it signed in Q3. Aaron Ross and Aaron Dillon break down Anthropic's IPO, Groq 2.0, the NeoCloud debt debate, Jensen Huang's departure from Nvidia, and SpaceX's transformation into an AI compute company. In this episode: Anthropic's IPO: why ~$1.4T looked fair, pricing above the range, and the ~6x order book The new IPO playbook: staggered lockup releases, early-release windows and fast index inclusion Groq 2.0: what remained after Nvidia's ~$20B licensing deal, and the reported $8B Series B Groq vs. Together AI ($8.3B post-money, July 2026) as NeoCloud comps "If they're profitable, they're doing it wrong": why NeoClouds borrow to build Backlogs and prepayments at CoreWeave, Crusoe and Nebius, and whether markets need a bad quarter to reset Jensen Huang steps down after 33 years: Nvidia's next chapter and what Jensen might do next SpaceX's first Starship upper-stage tower catch, and Musk's $500B annualized revenue target for December 2027 Grok Bot in practice: how Dillon runs an always-on AI agent as a "25-year-old employee" Why the best investments look obvious in hindsight Chapters 00:00 Intro: Anthropic, Groq 2.0 & SpaceX 00:33 Anthropic IPO: is ~$1.4T fair? 01:15 Pricing above the range, ~6x oversubscribed 02:25 The new lockup playbook & fast index inclusion 03:11 Institutional support & post-IPO volatility 03:58 Groq 2.0: from chipmaker to NeoCloud 06:12 Groq vs. Together AI valuations 07:02 Why NeoClouds should be burning cash 08:13 Signed contracts, prepayments & the backlog 09:59 CoreWeave, Crusoe & the race to stand up compute 10:48 Do markets need a couple of bad quarters? 11:54 Jensen Huang steps down from Nvidia 15:06 The Sam Altman playbook: could Jensen return? 19:43 Where does Jensen go next? SpaceX, Tesla & Terafab 23:48 A Jensen pre-IPO round? 24:28 SpaceX catches Starship's upper stage 25:18 SpaceX is a data center company 25:42 Grok Bot and always-on AI agents 28:28 How Dillon runs "Jake," his Grok Bot 30:41 One agent for work, one for home: hence the compute 31:10 The best investments are the obvious ones 33:13 SpaceX wrap: sub-one-year payback on compute 33:52 Close This content is for informational and educational purposes only and does not constitute investment advice, an offer to sell, or a solicitation of an offer to buy any security. Opinions expressed are those of the speakers as of the recording date and may change without notice. Private company securities are illiquid, speculative, and involve a high degree of risk, including the potential loss of the entire investment, and are generally available only to accredited investors. The hosts or their affiliates may hold positions in, or have business relationships with, companies discussed. Valuations and figures cited are based on public reporting and have not been independently verified. Short / SEO Variant Anthropic's IPO, Groq 2.0, Jensen Huang and SpaceX on The Cap Table — Pre-IPO Podcast. Aaron Ross and Aaron Dillon break down Anthropic's $83 IPO pricing and record debut, lockup schedules and index inclusion, Groq's NeoCloud pivot and reported $8B Series B, the CoreWeave, Crusoe and Nebius backlogs, Jensen Huang's resignation from Nvidia, SpaceX's first Starship ship catch, and how Grok Bot works as an always-on AI agent. Plus: why AI is still in its first inning. Informational and educational purposes only. Not investment advice or a solicitation to buy or sell any security.
About The Cap Table — Pre IPO Podcast
The Cap Table — Pre IPO Podcast

The Cap Table — Pre IPO Podcast

By @thecaptablepodcast

The Cap Table is a weekly podcast hosted by Aaron Ross and Aaron Dillon, breaking down the most important private and Pre-IPO companies before they hit the public markets. Interested in investing in Pre-IPO stocks? Let's talk. Aaron.ross@rosspreipo.com Aaron.dillon@agdillon.com