$500 Billion of Orders, a $104 Billion Loss — Anthropic's $83 Debut

Anthropic priced its initial public offering at $83 a share on Wednesday, above the $70-to-$78 range it pitched investors eight days earlier. The order book was reported at roughly $400 billion from institutions and close to $100 billion from individuals — none of it binding. The same company lost $104 billion in the first nine months of 2026 and carries nearly $900 billion of compute commitments stretching toward 2045. That is the tension: a debut priced on demand nobody is obliged to honor, for a business whose costs already run in the hundreds of billions. This piece asks whether the $1.5 trillion tag is a rational price for scarce AI capacity or the last, largest print of a cycle the market is starting to doubt.

Figure 1 — The four numbers behind the $83 print: price, valuation, revenue, loss. Sources: Bloomberg and CNBC via Investopedia, 18 Nov 2026; Anthropic S-1/A.

The price, not the business, is the story

The deal priced after the US close on 18 November and begins trading on Nasdaq on 19 November under ANT. At $83, the 825 million shares originally registered would raise about $68.5 billion, though the offering was upsized — one of the largest listings on record, the second AI mega-IPO after $SPCX(SpaceX)$ in June. The price sat 6.4% above the top of the range, a deliberate flex of confidence in a market that has spent two weeks punishing anything tied to AI capital spending.

The facts cut both ways. Nine-month revenue reached $33.24 billion, up roughly 1,200% year on year, and the annualised run rate hit $81 billion by 31 October, $10 billion higher a month earlier — real traction, driven largely by Claude Code. Against it sits a $104 billion nine-month net loss, mostly a non-cash charge from fair-value changes in convertible notes and derivatives. In Q3 the company swung from $974 million of operating income in Q2 to a $567 million operating loss, and posted a $23.3 billion net loss.

Figure 2 — The same private company marked at $965B in May, about $2T in September, $1.5T at the IPO price. Sources: Investopedia, 18 Nov 2026; AI Affairs citing Guardian and CNBC, 18 Nov 2026.

The valuation that cannot sit still

Here is the number that does not add up. Private investors marked $ANT(Anthropic)$ at $965 billion in May, analysts near $2 trillion in September, the prospectus range at $1.3 trillion to $1.4 trillion, and the final price at about $1.5 trillion. Four marks, one company, inside six months — and the business did not change by those magnitudes. What changed was the mood toward AI, and the price is that mood.

The mood is fragile. The ten biggest US IPOs of 2026 are down about 20% on average from their offer prices, per a Financial Times analysis cited by Investopedia. Morningstar's fair value was $92 a share, above the $83 print but not dramatically. The deal is priced above its own range, below that fair-value estimate, into a class that has lost a fifth of its value — a narrow band with little room for a disappointing first print.

Figure 3 — The 6.4% premium to the range top against the 20% average decline in 2026's biggest IPOs. Sources: FT analysis via Investopedia, 18 Nov 2026; Financial News, 19 Nov 2026.

The demand is enormous — and non-binding

The $500 billion headline is the most quoted number of the deal and the least firm. Institutional indications of interest were about $400 billion and retail interest neared $100 billion. Indications of interest signal attention, not capital; they are routinely oversized because they cost nothing to place. Treating them as demand at $1.5 trillion is like treating a show of hands as a purchase order.

What is firm is the customer base. By 30 September, 11,900 enterprise customers were spending at least $100,000 a year, against 1,500 at end-2025 — the genuinely impressive part of the filing, and a better foundation than the order book. It also concentrates risk: a group including $Alphabet(GOOGL)$ , $Microsoft(MSFT)$ , $NVIDIA(NVDA)$ and $Advanced Micro Devices(AMD)$ agreed to invest $26 billion, while Amazon and Alphabet each already hold stakes worth more than $100 billion. When investors are also suppliers and buyers, circularity becomes a valuation input.

The compute check nobody can unwrite

The cost structure is where bull and bear separate. Anthropic spent nearly $9 billion on computing capacity in the first half of 2026 and another $5 billion in Q3. That spend bought the revenue growth above. The commitments ahead are an order of magnitude larger: more than $70 billion over the next twelve months, roughly $450 billion by 2030 and a further $450 billion by 2045, close to $900 billion in total.

The prospectus says the quiet part aloud: if partner financing becomes unavailable or more costly, contracted capacity "may be delayed, repriced, reduced, or not delivered." The Federal Reserve has flagged AI-related debt and opaque funding structures as financial-system risks. A company that loses $104 billion and has signed nearly $900 billion of compute checks is not simply buying growth — it is assuming capital markets keep funding the build for two decades.

Figure 4 — Compute spend already booked versus commitments still to come, in billions of dollars. Sources: Anthropic prospectus via The Independent and AI Affairs, 18 Nov 2026.

Bull case

  1. $ANT(Anthropic)$ is the only listed pure play on frontier-model demand; $81 billion of annualised revenue can justify a premium to slower software peers.

  2. The $26 billion strategic investment from $MSFT(Microsoft)$, $GOOGL(Alphabet)$, $NVDA(Nvidia)$ and $AMD(AMD)$ aligns the chip supply chain with the equity story.

  3. Morningstar’s $92 fair value sits above the $83 price, implying a margin of safety for buyers.

  4. Pricing above range despite a weak AI tape shows genuine scarcity still commands a premium.

  5. 11,900 enterprise customers above $100,000 of annual spend is a recurring base a one-quarter loss cannot erase.

Bear case

  1. Roughly $500 billion of cited demand is non-binding and cannot support the opening print.

  2. A $104 billion nine-month loss and a Q3 swing into operating loss show the economics are not self-funding.

  3. Nearly $900 billion of compute commitments to 2045 depend on partner financing the company itself warns could fail.

  4. The mark moved from $965 billion to $2 trillion and back to $1.5 trillion in six months — a price in search of a business.

  5. The 2026 IPO class is down 20% on average, and the debut lands days after the Fed signalled further hikes.

Figure 5 — Where the deal was pitched, priced and independently valued. Sources: Anthropic S-1/A; Bloomberg and CNBC; Morningstar via Investopedia, 18 Nov 2026.

What the estimates are really saying

There is no formal guidance, so the relevant revision is the price itself. The range was set on 10 November at $70 to $78; demand pushed it to $83 on an upsized share count. The market is asked to accept that a company repriced itself upward in eight days on interest carrying no obligation. The honest read: $83 is a clearing price for a scarce asset, not a statement of $1.5 trillion of intrinsic value. Buyers should anchor to the sourced $92 fair value and $81 billion run rate, not an order book that evaporates when the bell rings.

Figure 6 — The two axes that confirm or break the $1.5T case: demand conversion and unit economics. Sources: Anthropic S-1/A; CNBC; FT via Investopedia, 18 Nov 2026.

Strategic Slam

  • $ANT(Anthropic)$ — a scarce asset at a fair-but-full price; the position is only as good as the first week's volume and the Q4 revenue print.

  • $NVDA(Nvidia)$ and $AMD(AMD)$ — named IPO investors and compute suppliers, so the debut is a sentiment read on AI infrastructure, not a change in fundamentals.

  • $MSFT(Microsoft)$ and $GOOGL(Alphabet)$ — strategic holders with mark-to-market exposure, immaterial to their earnings power.

Catalyst Calendar

  • 19 November — first trading day under ANT; the opening print is the first binding price.

  • December — first month as a public company and the next read on the annualised revenue run rate.

  • Early 2027 — first quarterly report; the test is whether operating losses narrow as revenue scales.

  • Ongoing — any partner financing stress on the roughly $900 billion of compute commitments.

Figure 7 — Four dates that turn the debut into a stress test. Sources: CNBC, Investopedia, Trading Economics, 17–19 Nov 2026.

Verdict

The $83 price is defensible; the $500 billion justification is not. Anthropic is a real business with $81 billion of annualised revenue and 11,900 serious enterprise customers, and it deserves a public listing. But a $1.5 trillion valuation set above range on non-binding orders, while the company books a $104 billion loss and underwrites nearly $900 billion of compute checks, is priced for a world where financing never tightens. With the Fed signalling hikes and the 2026 IPO class down 20%, that is a world the market is refusing to grant. The stance is neutral on the asset and sceptical of the narrative: own the revenue, not the order book.

Key Takeaways

  1. Anthropic priced at $83, above its $70–78 range, valuing it at about $1.5 trillion.

  2. Roughly $500 billion of orders were reported, but none were binding.

  3. Nine-month revenue was $33.24 billion, up about 1,200%; the loss was $104 billion, mostly non-cash.

  4. Compute commitments reach nearly $900 billion through 2045 and depend on partner financing.

  5. The 2026 IPO class is down 20% on average — the debut’s real benchmark.

If the opening week holds above $83 on real volume, was the premium justified — or just the last buyer pricing in a market already doubting the AI build-out?

Tickers

$ANT(Anthropic)$ $NVDA(Nvidia)$ $MSFT(Microsoft)$ $GOOGL(Alphabet)$ $AMD(AMD)$

Sources

Bloomberg and CNBC via Investopedia, "Anthropic Has Priced Its IPO," 18 Nov 2026; AI Affairs citing the Guardian and CNBC, 18 Nov 2026; Financial News, 19 Nov 2026; TRT World and The Independent, 18 Nov 2026; Financial Times analysis via Investopedia, 18 Nov 2026; Anthropic Form S-1/A and prospectus supplements, SEC EDGAR, Nov 2026; Trading Economics, 18 Nov 2026.

Disclaimer

This is not investment advice. The author holds no position in any security mentioned and receives no compensation from any company discussed. Figures are cited to the named sources; forward-looking statements are the sources' views, not guarantees. Do your own research.

# 💰Stocks to watch today?(19 November)

Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.

Report

Comment

  • Top
  • Latest
empty
No comments yet