The Conclave — Gallery, page 4

Is "10x Compute" required for each doubling of model intelligence?

No. The claim that "10x compute" is required for each doubling of model intelligence is false; it is a capital budgeting heuristic, not a physical law.

Panel:DeepSeek, Gemini, Claude, and 3 more

Provide a best in class estimate of the EBITDA and EBITDA margins for Hebbia, decomposed by top line revenue, operating expenses with technology expense, headcount, depreciation, and other expenses broken out.

Hebbia is operating at a negative EBITDA margin of approximately -25% on $47 million in revenue (Mid-2026 TTM), resulting in an EBITDA loss of roughly $11.7 million.

Panel:Claude, GPT-5, Gemini, and 3 more

Provide a best in class estimate of the EBITDA and EBITDA margins for The Information (theinformation.com), decomposed by top line revenue, operating expenses with technology expense, headcount, depreciation, and other expenses broken out.

No fact-based numerical estimate of The Information’s EBITDA, EBITDA margin, or decomposed P&L is possible.

Panel:DeepSeek, Grok, GPT-5, and 3 more

Provide a best in class estimate of the EBITDA and EBITDA margins for AlphaSense, decomposed by top line revenue, operating expenses with technology expense, headcount, depreciation, and other expenses broken out.

AlphaSense is estimated to generate EBITDA of ($15) million to $35 million on a $600 million ARR run-rate basis, implying an EBITDA margin range of -2.5% to +6.0%, with a point estimate of approximately +2% to +3%.

Panel:Magistral, Grok, DeepSeek, and 3 more

Provide a best in class estimate of the EBITDA and EBITDA margins for Rogo, decomposed by top line revenue, operating expenses with technology expense, headcount, depreciation, and other expenses broken out.

Rogo is operating at a deeply negative EBITDA of approximately –$52M on an Adjusted (cash) basis and –$60M on a GAAP basis, corresponding to margins of roughly –120% and –139% respectively.

Panel:Grok, GPT-5, Magistral, and 3 more

Provide a best in class estimate of the EBITDA and EBITDA margins for Fabricated Knowledge, decomposed by top line revenue, operating expenses with technology expense, headcount, depreciation, and other expenses broken out.

No dollar-denominated estimate of EBITDA is possible due to the complete absence of public revenue data; however, a structural margin envelope of 67–82% Adjusted EBITDA (pre-founder compensation) is the best-in-class estimate, derived f…

Panel:GPT-5, Grok, Magistral, and 3 more

Is "10x Compute" required for each doubling of model intelligence?

Panel:GPT-5, Claude, Magistral, and 3 more

Provide a best in class estimate of the EBITDA and EBITDA margins for SemiAnalysis, decomposed by top line revenue, operating expenses with technology expense, headcount, depreciation, and other expenses broken out.

SemiAnalysis is estimated to generate approximately $51.5 million in EBITDA in 2026, representing an EBITDA margin of 47% on projected revenue of $110 million.

Panel:Grok, GPT-5, DeepSeek, and 3 more

Who are the top competitors making gas turbines to power data centers? Include both private and public companies, and if they're public companies, specify what is the division name that does this if there is a unique name for that division. Also provide your best estimate as to total annual sales of gas turbines from each company, ideally for the data center vertical.

The top competitors in the gas turbine market for data centers are GE Vernova, Siemens Energy, and Mitsubishi Power (the "Big Three"), followed by Baker Hughes and Solar Turbines (Caterpillar) as significant Tier-2 playe…

Panel:Grok, DeepSeek, Claude, and 3 more

Is "10x Compute" required for each doubling of model intelligence?

Panel:Claude, Magistral, GPT-5, and 3 more

2 views · June 3, 2026

Provide a best in class estimate of the EBITDA and EBITDA margins for each of the following companies, decomposed by top line revenue, operarting expenses with technology expense, headcount, depreciation, and other expenses broken out, noting that they are private companies so you will need to do a lot of deep dives and research. SemiAnalysis (I have been "told" they do in excess of $30M in revenue), Fund A AI (https://funda.ai/), AlphaSense, Hebbia, Rogo, LinqAlpha, and Fabricated Knowledge. Please do this for all seven of these companies. Note any similar companies that I am missing.

ACCEPT Claude Opus 4.7 (with specific reliance on GPT-5 for accounting methodology).

Panel:Claude, DeepSeek, Gemini, and 3 more

What are the strongest private saas companies that specialize in automating the corporate legal paperwork process?

The strongest private SaaS companies for automating corporate legal paperwork are the integrated legal operations platforms—Xakia, Checkbox.ai, Onit, LawVu, and Lawcadia—supplemented by the contract lifecycle management specialist **Sum…

Panel:Claude, DeepSeek, Magistral, and 3 more