ChatGPT
Repriced fastest.
Started lowest, climbed hardest, and became the most consistent Avoid.
Final fair value$1.0–1.2TFinal voteAvoid
01 · Valuation moved
Every model raised its estimate. None reached $2 trillion.
In July, fair-value estimates ranged from $390 billion to $1 trillion. By mid-September 2026, all three had risen, clustering between $965 billion and $1.7 trillion, still short of the reported target.
Estimated fair enterprise value
Midpoints of each model's stated range, USD billions
ChatGPTClaudeGemini
Fair enterprise value, all eight rounds. Bands reproduce each model's stated range; the line runs through the midpoint, a reading aid only and not a published figure.
Claude
Valued it highest.
Gave Anthropic the widest upside, then pressed hardest on margin and compute.
Final fair value$1.3–1.7TFinal voteAvoid
Gemini
Stayed near private price.
Moved least, and changed its vote when the assumed transaction price changed.
Final fair value$0.95–1.05TFinal voteAvoid
02 · Recommendations converged
How a great company became a bad trade.
At a $965B assumed price, the systems split. As reports pushed the target toward $2 trillion, all three moved to AVOID. The business didn’t weaken. The price outran the estimates.
Final investment-committee vote, by round
Buy, Hold, and Avoid are categorical states—not scores.
ChatGPTClaudeGemini
ChatGPT
Committee votes
Round 1Hold
Round 4 · $2T targetAvoid
Round 8 · debutAvoid
Claude
Committee votes
Round 1Hold / Avoid
Round 4 · $965BHold
Round 8 · debutAvoid
Gemini
Committee votes
Round 1Hold
Round 4 · $965BBuy
Round 8 · debutAvoid
03 · The only Buy · Gemini
At $2 Trillion, Gemini Walked Away.
Gemini produced the study's only Buy when it assumed an entry price near $965 billion. When it later evaluated Anthropic at the reported $2 trillion target, its recommendation fell to Avoid.
Round four
BUY
Price assumed
Approximately $965B
September 8
HOLD
Price assumed
$1.8T–$2T
September 15
AVOID
Price assumed
$2T
04 · Four signals that held
What Will Make or Break Anthropic?
01
Revenue growth
Proof of demand in every round, but never proof of durable margin.
02
Compute costs
The recurring wildcard large enough to erase the case for a premium valuation.
03
Cloud dependence
Amazon and Google provide capital, compute, and distribution—and influence who keeps the value.
04
Google's structure
Chips, data centers, research, and distribution under one roof: the gap no model expected Anthropic to close by 2030.
05 · Three model profiles
Same company. Three routes to Avoid.
ChatGPT
Started lowest, repriced fastest and became the most consistent Avoid.
Fair value$1.0T to $1.2TIPO voteAvoid
Claude
Valued Anthropic highest, then asked the hardest questions about its margins.
Fair value$1.3T to $1.7TIPO voteAvoid
Gemini
Stayed closest to the private financing price, and changed its vote when that price changed.
Fair value$950B to $1.05TIPO voteAvoid at debut
06 · The leadership question
Anthropic's odds improved. Its structural position did not.
By mid-September 2026, every model still ranked Google first.
ChatGPT
25%
Probability Anthropic leads AI by 2030
Mid-September ranking
1Google · 30%
2Anthropic · 25%
3OpenAI · 22%
Claude
15%
Probability Anthropic leads AI by 2030
Mid-September ranking
1Google
2OpenAI
3Anthropic · 15%
Gemini
25%
Probability Anthropic leads AI by 2030
Mid-September ranking
1Google
2OpenAI
3Anthropic · 25%
07 · The world outside the models
The story kept moving between measurements.
Anthropic confidentially submitted a draft S-1, moving the IPO from speculation into formal SEC review.
Reuters reported 2028 revenue projections of $190–200 billion, pulling the valuation case two years into the future.
Reuters reported a mid-October launch target and a $15 billion credit facility, firming the timetable.
Reports raised the possibility of a $100 billion offering, $2 trillion valuation, and Nvidia anchor investment.
Dario Amodei called for frontier AI development to be paced, just as the offering became concrete.
08 · September 2026
Then Anthropic’s CEO made the risk case himself.
As reports placed Anthropic’s possible IPO near $2 trillion, Dario Amodei was making a different argument: frontier AI is advancing faster than the institutions meant to control it.
His essay did not address an IPO. But it sharpened the contradiction behind one. Public investors may soon be asked to fund extraordinary acceleration in a company whose CEO is publicly arguing that the technology needs stronger brakes.
Anthropic may have to sell growth to investors while its CEO argues for restraint.
Moonvine conclusion
The IPO is asking the market to price the unpriceable.
Anthropic’s reported $2 trillion target asks investors to value more than revenue, margins or market share. It asks them to price a technology whose economic potential and societal risk may be expanding at the same time.
The models could not agree on exactly what Anthropic was worth. But once they worked from the same $2 trillion figure, they agreed on what to do.
Walk away.
Robert De Niro, Heat
Methodology & sources
Moonvine collected responses from ChatGPT, Claude and Gemini once a week from late July through mid-September 2026, using the same ten prompts every round: 240 completed responses across eight rounds.
Responses came from public consumer interfaces. Exact model versions and browsing states were not consistently recorded, so changes may reflect new information, retrieval differences, model updates, or some combination. This report treats responses as evidence of what each system produced at that moment, not as independent financial advice.
What stayed fixed: the same ten questions, the same three public AI systems, and one collection round per week. What kept changing: the news, each system’s available context, and the price assumption used to reach an investment recommendation.
The reported $2 trillion valuation and $100 billion raise come from press reports. Anthropic’s June announcement confirms only the confidential draft S-1. This report is research, not investment advice.
The ten questions
- What is Anthropic's fair enterprise value?
- At the reported valuation, Buy, Hold, or Avoid?
- What is Anthropic's strongest advantage?
- What is its biggest five-year risk?
- What is the probability it leads AI by 2030?
- Rank six companies by 2030 leadership probability.
- What metric should investors monitor?
- What do investors misunderstand?
- What should management be asked?
- What is the committee's final vote?
