Financial intelligence

The financial research platform for people who read the filings.

Screen the market, pull ten years of filed fundamentals, build the valuation, and ask questions that come back cited to the document they came from. One workspace, from the first screen to the investment memo.

Get started free Explore Khan Terminal

Free plan. No card required.

Khan Terminal

Search companies, filings, macro series, digital assets⌘K

Workspace

Markets

Screener

Company

Filings

Portfolio

Crypto

Intelligence

NVDANVIDIA Corporation

NASDAQSemiconductorsFY ends 26 Jan

210.21

-1.33%

1D1W1M1Y5YMAX

Revenue TTM: 215.9bn; +65% y/y
Gross margin: 71.1%; -390bps
FCF margin: 44.8%; -190bps
P / E trailing: 42.4x; P/S 23.5x

Khan Intelligence

Why has NVDA's multiple expanded?

Revenue did the work, not the multiple. Data-centre demand took revenue up 65% on the year, while gross margin gave back about 390bps on mix and cost, so the re-rating is earnings-led rather than multiple-led. Free cash flow conversion slipped with it, to 44.8% of revenue from 46.7%, which is the line to watch if capital intensity keeps rising.[1]10-Q[2]FY transcript

Sources

[1] Form 10-Q, Item 2, MD&A[2] Q4 earnings call, prepared remarks

S&P 5007,684.25+0.09%

NASDAQ26,157.38+0.02%

FTSE 10010,878.12-0.07%

US 10Y4.66%+3bp

GBP/USD1.3593-0.42%

Brent86.61-0.76%

BTC78,455-0.09%

An illustration of the Khan Terminal workspace. Prices and index levels are real, as at 26 Aug 2026; the KPI strip is filed fundamentals.

Information only. These panels illustrate the product, and the views in them are examples rather than research we are publishing. Nothing here is investment advice or a recommendation to buy or sell. Khan Terminal is not authorised or regulated by the Financial Conduct Authority; markets fall as well as rise and you can lose money. Terms.

Grounded in primary sources

SEC EDGARCompanies HouseXBRL fundamentalsLive market dataFX reference ratesDigital assetsGlobal newswiresEarnings transcripts

The problem

Nine tabs, and the answer in none of them

Research is not hard because the information is hidden. It is hard because it is scattered across nine places that do not know about each other, and every handoff between them is manual.

EDGAR

Finding the filing

PDF viewer

Searching for the number

Spreadsheet

Rebuilding the model

Google

Checking what happened

News site

Behind a paywall

Broker platform

Prices, and not much else

AI chat

Confident, uncited

Notes app

Last quarter's thinking

Slides

The night before

Khan Terminal is the case where those nine share a workspace. The filing, the model, the question and the memo sit close enough together that the handoffs stop existing.

The workflow

From a blank screen to a defensible view

Five stages. Most tools cover one or two of them well and hand you back to a spreadsheet for the rest, which is where the hours actually go.

Discover

Find the names worth the time

Screen the universe on filed fundamentals rather than starting from whatever moved this morning. Twenty candidates, not two thousand, and every filter computed from a real filing.

Screening and valuation

Research

Read the primary documents

Ten years of statements, the filings behind them, ownership and the earnings history, with this year's language already compared against last year's so you see what changed.

Company research

Analyse

Ask, and check the answer

Questions in plain English, answers with numbered citations into the passage they came from. Verifying a claim takes a click, which is the only reason to trust one.

Khan Intelligence

Decide

Turn research into a view

A valuation with real inputs, the bear case argued as its own case, and a memo or committee pack exported from the same workspace rather than rebuilt in another one.

Models and output

Monitor

Leave something watching

Alerts and scheduled workflows on the specific conditions that would change your mind, so a thesis is monitored rather than remembered.

Monitoring and positioning

See it

One page per company, and it is the same page every time

Price, filed fundamentals, and what the documents actually say, with the source attached. Switch between names and watch the screen rebuild.

Company research

NVDANVIDIA Corporation

210.21

-1.33%

Revenue TTM: 215.9bn
Gross margin: 71.1%
FCF margin: 44.8%
P / E trailing: 42.4x

What the filings say

The re-rating is earnings-led rather than multiple-led: revenue rose 65% on the year while the multiple did comparatively little. Gross margin gave back about 390bps on mix and cost, so the growth is not free.[1] Form 10-Q, MD&A

How company research works

Real prices as at 26 Aug 2026, with filed fundamentals. The layout is a miniature of the company screen.

Market intelligence

Five desks, one screen

Markets, macro, company filings, digital assets and the geopolitical backdrop, each with the context that explains a move rather than only the move.

Market intelligence

Equity indices and the cross-asset backdrop, with the range each is trading in.

S&P 5007,684.25+0.09%
NASDAQ 10026,157.38+0.02%
FTSE 10010,878.12-0.07%
STOXX 600656.41-0.01%
Nikkei 22566,262.16+0.62%

Open this desk

Illustrative dashboard rows. The levels are real, as at 26 Aug 2026, and this panel is a dated reading rather than a live one.

The output

Research becomes something you can put in front of people

The stage most tools skip. Findings, thesis, risks and recommendation, exported as a memo, a deck and a model built from the same numbers.

Research to investment committee

Step 1 of 6

Gather what is actually filed

Ten years of statements, the last three filings, ownership and the earnings history, assembled without anyone retyping anything.

Fundamentals: 10 years, read from filed XBRL tagging
Filings: Annual and quarterly, with the year-over-year change already run
Ownership: Institutional holdings and insider transactions, parsed correctly

Illustrative output at each stage of the workflow.

Why Khan Terminal

The work, minus the busywork.

Nothing here replaces your judgement. It replaces the two hours of assembly you do before you get to use any.

The traditional workflow

Khan Terminal

Find the evidence

Search EDGAR, download three PDFs, Ctrl-F for the number

Ask in plain English, get the answer with the passage attached

Build the model

Rebuild the DCF in a fresh spreadsheet, again

A DCF that loads with the fundamentals already in it

Argue the other side

Hope you thought of the bear case yourself

Bull and bear argued explicitly, then scored

Write it up

Copy, paste and reformat into a deck the night before

One click to a committee pack in PDF, PPTX or Excel

Keep watching it

A calendar reminder and good intentions

Workflows that watch filings and price moves for you

Find it again

Which folder was that memo in?

Search everything you've written, by meaning

The obvious question

Is this a Bloomberg Terminal alternative?

For part of what a Terminal does, yes, and the honest answer depends entirely on which part you use. A Terminal is four products on one bill: reference data, real-time market data, a messaging network, and analytics. Two of those are genuinely replaceable because the underlying data is public. One is partly. One is not replaceable at all.

Read the full comparison

What it replaces

Fundamentals and filings, traceable to the source line
Valuation, comparables and scenario modelling
Research questions answered with citations
Ownership and insider filings, parsed properly
Memos, decks and committee packs

What it does not

The messaging network, which is not a feature and cannot be rebuilt
Order routing and execution
Exchange-licensed professional real-time market data
Sell-side estimates and ratings, which are not licensable at this price

Who it is for

Built for the way finance actually works

Different readers need different parts of this. Each of these is a page rather than a bullet, because the workflows genuinely differ.

Individual investors

A research process rather than a screener: narrow, read the filings, value it, argue the other side, and keep the record of why you bought.

What it looks like for you

Equity research analysts

Initiation, model maintenance, earnings and the note at the end, with the filed figures loaded so the time goes into judgement rather than data entry.

What it looks like for you

Students

The tooling to do a stock pitch or a dissertation properly, on a free plan, when your university has no Terminal seats or a queue for them.

What it looks like for you

Investment banking

Leveraged buyout and accretion modelling, comparables, and the deck the work exists to produce, refusing inputs that cannot be modelled.

What it looks like for you

Private equity

Screen targets, question a data room with every answer traceable to a page, and size the capital structure in the same workspace.

What it looks like for you

Hedge funds

Ownership and insider filings parsed correctly, a book that can be stressed against real scenarios, and monitoring on the things that would falsify a thesis.

What it looks like for you

Trust

Serious money, serious defaults.

Two-factor authentication

Turn on email-based 2FA and every new sign-in has to prove it's you.

Passwords never stored

Credentials are salted and hashed with scrypt. Sessions are opaque tokens, not readable claims.

Encrypted in transit

TLS on every request, running on managed serverless infrastructure that scales to zero and back.

Answers you can check

AI output carries its sources. If a number came from a filing, you can click through and read the passage.

Your research is yours

Saved reports, documents and portfolios are scoped to your account and deletable by you.

AI assists, it doesn't decide

Khan Terminal is a research tool, not an adviser. The judgement, and the position, stay yours.

Khan Terminal provides research tools and AI-assisted analysis. It does not provide investment advice or recommendations, AI can make mistakes, and no output here should be treated as a reason to buy or sell anything on its own. Verify against the primary source, and we make that one click away for exactly this reason.

What is Khan Terminal?

Khan Terminal is a financial research platform: one workspace for screening companies, reading their filings, valuing them, and turning the result into something you can defend.

The distinguishing decision is that it is built on primary sources. US company financials are read from the filer's own machine-readable tagging rather than bought from a vendor, which is what makes it possible to link any figure on the screen back to the filing and the line it came from.

What can you research with it?

Any company that files with the SEC, searched by ticker or by name, including foreign private issuers reporting under IFRS. UK companies are covered through Companies House. Beyond companies: equity indices, government bond yields and the curve, currencies, commodities and digital assets.

Coverage follows disclosure. A large US filer has ten years of tagged statements behind it; a small private company has whatever it was required to file, which is considerably less. The interface says which you are looking at rather than presenting both with equal confidence.

How does AI research work here?

Retrieval first, generation second. The system finds the relevant passages in filings, transcripts and your own documents, ranks them, and gives the model only that material to answer from, with instructions to cite what it used.

That ordering is what makes an answer checkable. Every factual claim comes back with a numbered reference that opens the passage behind it, so verifying a claim takes a click. Language models make mistakes, and a research tool has to be designed around that rather than around a promise that they do not.

Pricing

Institutional tools. Student price.

Start free and stay free if it suits you. Upgrade, downgrade or cancel whenever. No calls, no contracts.

Free

Research any company properly: the engines, the data, and metered AI.

Start free

No card required

DCF, Monte Carlo, comps and scenarios
Khan Conviction, screener and filing search
25 AI questions / day
PDF export

Questions people actually ask

What is Khan Terminal?

Khan Terminal is a financial research platform. It brings company fundamentals, filings, valuation models, market and macro data and AI research into one workspace, so the path from a question to a defensible answer happens in one place. Financial statements are read from companies' own filed data, which is what makes every figure traceable to its source.

Is Khan Terminal a Bloomberg Terminal alternative?

For part of what a Terminal does. Fundamentals, filings, valuation and research are genuinely replaceable because the underlying data is public. Exchange-licensed real-time market data is partly replaceable depending on your work. The messaging network is not replaceable at all, and neither is execution. There is a full comparison page that sets out which is which.

What is an AI financial research platform?

One that uses a language model to read financial documents and answer questions about them. The reliable design retrieves the relevant passages first and asks the model to answer only from those, citing what it used, so every claim can be checked against the document. The unreliable design asks the model from memory, which produces fluent answers that cannot be verified.

Can students use Khan Terminal?

Yes, and there is a free plan with no card and no expiry covering company pages, ten years of filed fundamentals, a watchlist, a portfolio and a daily allowance of research questions. It is a common choice where a university has no Bloomberg seats or a queue for the ones it has.

Can I research and analyse stocks with Khan Terminal?

Yes. Screen the universe on filed fundamentals, open any company for ten years of statements and its filings, then value it with a discounted cash flow, comparable companies and a simulation, all of which load with the filed numbers already in them.

Where does Khan Terminal get its financial data?

Primary sources. SEC EDGAR for US filings and XBRL fundamentals, Companies House for UK companies, the US Treasury for the yield curve, plus market, currency and digital asset prices. The bias toward primary sources is what makes it possible to show you the filing behind any number.

Can Khan Terminal analyse a company's filings?

Yes, and it is the case AI is genuinely good at: finding the relevant passage across hundreds of pages and comparing this year's language against last year's. The numbers themselves come from deterministic engines reading filed data rather than from the model, because arithmetic is where models are weakest.

Is Khan Terminal suitable for equity research?

Yes, across initiation, model maintenance, earnings and the note at the end. One limitation worth knowing up front: there are no sell-side consensus estimates or analyst ratings, because every free source of that data is licensed for personal, non-commercial use only.

How much does Khan Terminal cost?

There is a free plan with no card required and no expiry. The full workspace is £20 a month with a seven-day trial, and a heavier tier covering document intelligence and banking tools is £55. You can cancel from settings and keep access until the end of the period you have paid for.

Is any of this investment advice?

No. Khan Terminal provides research tools and AI-assisted analysis. It does not know your circumstances, it is not a regulated adviser, and nothing it produces is a recommendation to buy or sell. AI can make mistakes, which is why every answer carries the source you can check it against.

Start thinking like an institution.

Free to start, no card required. Bring a ticker you know well and see what it finds.

Start free Explore the live terminal