KI-Hype: Ist das der Anfang vom Ende? (Anthropic IPO)

Summary

The video examines Anthropic’s reported IPO disclosures and what they might imply for AI-related stocks and the wider market. The auto-generated subtitles appear to have lost some decimal points or units in several figures, so the financial amounts below should be treated cautiously.

Anthropic’s reported financial picture

• IPO valuation: The subtitles render the target as $1.5–2 trillion, but this appears likely to be a decimal or unit transcription error. The intended figure may be $150–200 billion.
• Revenue: Revenue reportedly grew by more than 10× year over year, with another roughly tenfold increase shown in comparisons between half-year periods. First-half 2026 revenue is given as $16 billion. A further tenfold increase would imply $160 billion in first-half 2027, but the presenter stresses that this is hypothetical, not a forecast.
• Profitability: The subtitles report a first-half loss of $80 billion, including a $72 billion fair-value change on convertible notes. Those amounts appear especially likely to have lost decimal points in transcription. The presenter treats the convertible-note charge as a non-operating accounting effect and says operating expenses still exceeded revenue during the half-year.
• Recent operating trend: The presenter says revenue has grown faster than computing-power costs. The transcript reports a ratio improving from 0.4 in Q1 2025 to 2.08× in the latest quarter, and an adjusted operating margin of 19% in that quarter. The ratio is described inconsistently in the captions, so its exact interpretation is uncertain.
• Cash flow and investment: First-half 2026 operating cash flow is reported at nearly $4 billion. After accounting for $2.2 billion of share-based compensation, the presenter estimates roughly $1.5–2 billion of positive operating cash flow. Investment spending is given as $17 billion. The subtitles also report about $110 billion raised during the half-year, another figure that may have a decimal or unit issue.
• Cash and compute commitments: Cash is reported as $112 billion, likely another potentially misrendered figure. The company is said to have secured $600–800 billion of computing-power commitments over time, including agreements involving Microsoft, Google, Amazon, and xAI. The presenter estimates that commitments of around $600 billion spread across seven years would require roughly $80–90 billion of annual revenue to cover; at twice that level, around $160 billion would be needed. These figures are estimates, and the commitments’ terms and durations vary.

Investment considerations and risks

• Revenue growth alone does not prove a durable business model. The presenter illustrates how a company could grow sales by selling a product below cost, then lose customers when it raises prices.
• Anthropic relies on substantial computing capacity. Long-term commitments could support continued growth, but create risk if revenue falls short or growth slows.
• The presenter thinks short-term profitability is plausible given improving operating and cash-flow indicators, but cautions that the evidence covers only a short period.
• The presenter believes broader AI use is likely to keep growing and suggests that a 10–30% price increase, if it did not bring comparable additional costs or customer losses, could improve profitability.
• Competition is a central long-term concern. OpenAI, Meta, Alphabet, and Chinese open-weight models are among the alternatives. The presenter worries that models could become more interchangeable, triggering price competition and margin pressure.
• Other risks discussed include overbuilding data-center capacity, slower-than-expected AI adoption, more efficient models requiring less computing power, AI moving onto phones and laptops instead of large data centers, and excessive stock valuations.
• The presenter does not view Anthropic’s possible failure as necessarily a systemic threat to the whole AI market: demand for computing infrastructure could continue through other AI providers. However, a broader slowdown in AI demand could affect infrastructure suppliers.
• The presenter says they are not investing in the IPO, citing a personal principle, and notes that IPOs have statistically tended to underperform slightly.

Business-model framework

The presenter assesses Anthropic using five factors:

• Recurring revenue and switching costs: Recurring revenues exist, but switching between models appears relatively easy; rated weak.
• Network effects: User feedback may help improve models, but the presenter considers the effect limited. Major gains may instead come from more computing power or technical improvements.
• Economies of scale: Strong. Significant scale and capital are needed to compete.
• Proprietary technology: Low to medium, in the presenter’s view; competing providers can produce similar models.
• Brand: Medium to somewhat above average, though OpenAI and ChatGPT also have strong brands.

The presenter’s resulting score is 13 out of 25, characterized as a middling business model that may face limits on pricing power and margins.

Assets, companies, and sectors mentioned

• Companies and models: Anthropic, OpenAI, ChatGPT, Google/Alphabet, Amazon, Microsoft, Meta, xAI, Berkshire Hathaway, NVIDIA, TSMC, and ASML.
• Other investments and sectors: ETFs; AI models and software; cloud and hyperscalers; semiconductors and chipmaking equipment; data centers, construction, cooling, cabling, machinery, and energy.
• No stock ticker symbols are stated in the subtitles.

Disclosures and sources

The channel’s disclosure says the video reflects the presenter’s personal opinion, is not investment advice, and that investments carry risk and should be independently checked. The subtitles also mention three warnings about catastrophic or existential risks from AI but do not detail them.

Presenter/source: Jannes Lorenzen (Aktienrebell). Florian Heinemann of startup investor Project A is cited for views on AI competition and companies adopting Chinese open-weight models.