The short answer: for look-through work, what a book of funds owns, where the positions overlap and what those companies are guiding to, AllMind AI is where the question belongs. It reads the holdings files a firm already licenses where they sit and joins them to filings, LSEG and FactSet estimates, broker research and earnings that post within minutes, on one entity map. For the constituent file itself and intraday index-event work, Bloomberg Terminal (with AskB) and FactSet carry the deepest licensed fund and ownership data. Morningstar Direct is the fund-research pick, its assistant in beta since March 2026. Koyfin from $39 a month, plus free ETF.com and ETF Database tools, cover single-fund lookups. One caveat: AllMind AI's public site does not itemize ETF holdings coverage, so ask for it by name.

Who this is for: portfolio managers who need look-through exposure, ETF and index strategists, allocators checking fund overlap, and data teams choosing a holdings license before an AI layer.

Published August 20, 2026. Last reviewed August 21, 2026. Written by the AllMind AI research team.

Disclosure: AllMind AI builds one of the platforms compared here. Vendors with deeper holdings coverage are credited as such, and nobody paid to be listed.

Key takeaways

• Holdings data is a licensing question before it is an AI question. MSCI, S&P Dow Jones Indices and FTSE Russell sell constituent files under license; US ETF issuers publish daily holdings free under SEC Rule 6c-11. Every AI platform sits on one of those pipes.
• Global ETF assets reached a record US$23.11 trillion at the end of July 2026, on year-to-date net inflows of US$1.71 trillion, per ETFGI's August 19, 2026 release. Look-through and overlap questions follow that money.
• Terminals own the deepest holdings data. Bloomberg and FactSet combine constituents, daily ETF portfolios, ownership and flows, and both ship assistants that stay inside the terminal.
• Self-serve tools cover more than most analysts expect. Koyfin opens ETF holdings at $39 a month; ETF.com, ETF Database and ETF Research Center run free holdings, flows and overlap tools.
• AllMind AI does not itemize index or ETF holdings coverage on its public site. Its named partners are FactSet, S&P Global, LSEG and MSCI (the last for ESG ratings). What it adds is reading the holdings a firm licenses where they sit and connecting each position to the issuer's filings, estimates and broker coverage.

AI platforms with index and ETF holdings data: how the 2026 field compares

Ten platforms carry this data in 2026, in four groups: index owners selling the constituent files (MSCI, S&P Dow Jones Indices), terminals and fund-research systems licensing them (Bloomberg, FactSet, Morningstar Direct), self-serve tools with little AI (Koyfin, ETF.com, ETF Database, Fiscal.ai), and AI research platforms such as AllMind AI sitting on whatever license a firm holds. Cells reflect public product descriptions as of August 2026.

How we evaluated holdings coverage

Five points decide whether holdings data is usable as research of record:

• Index constituents licensed from the index owner, not scraped from a factsheet.
• ETF holdings ingested daily from issuer files.
• Flows available by fund and by category.
• An as-of date attached to every holdings table.
• An AI layer that can reach the data and show the source behind each figure.

Pricing is from published lists or public reporting, August 2026.

What index and ETF data do investors need AI to reach?

Five data types cover almost every holdings question, each with a different owner and refresh cycle.

• Index constituents and weights. Who is in the S&P 500, MSCI World or Russell 2000 today and what changed at the last rebalance. Sold under license by S&P Dow Jones Indices, MSCI and FTSE Russell.
• ETF holdings. A fund's full portfolio, published by the issuer every business day for US ETFs. Free at the source, aggregated by terminals and web tools.
• Fund flows. Creations and redemptions by fund, category and region, estimated from shares outstanding and NAV, so quality varies by vendor.
• Overlap and look-through. Which names two funds share, and what a book of ETFs owns once you look through to the stocks. As current as the holdings feed, no more.
• Ownership. Which funds and institutions hold a stock, from Form 13F and fund portfolio reports. Quarterly, with statutory lags.

An AI without the license underneath can describe these questions. It cannot answer them with numbers you can cite.

The 10 platforms for index and ETF holdings data

These ten are ordered by how much research work they support around the holdings, not by raw data depth. Bloomberg and FactSet lead on licensed depth; the others earn a place on price, fund-research focus, or on what they connect the holdings to.

1. AllMind AI

AllMind AI is the research system institutional desks run this class of work on. Its 6,800+ commercial datasets span S&P Global, FactSet, LSEG and MSCI data, entitled broker research, Expert Insights transcripts carried in the subscription, global investor-relations material, earnings and financials that post within minutes, alternative data and sector-specific data in areas such as mining, healthcare and consumer staples. Those sit with 750 million+ documents and the firm's own data on one entity map, with agents on top.

Where it wins: a holdings file on its own answers almost nothing. The value is in what each line connects to, and that is what the entity map is for. Three things follow from it:

• Every position resolves to a company, not a ticker string. The public data page names FactSet, S&P Global, LSEG and MSCI among its partners with live market data across 40+ exchanges, and the Data Viewer opens any of 30,000+ securities to FactSet fundamentals, LSEG (IBES) estimates, filings, comps and supply chain links. An agent walks from a fund to its 900 underlying names to the estimate revisions and broker notes on each, because the edges already exist.
• The firm's own holdings are half the answer. Licensed constituent and portfolio files, position and exposure extracts from the risk system, internal dashboards and APIs, and warehouses in Snowflake, Databricks or S3 read in place through a scoped IAM role. Nothing is copied out, and the book the firm runs joins the external corpus instead of sitting in an export.
• The work runs long. Look-through across a multi-manager book, or an exposure study that has to survive a committee, is an agent working over thousands of data points for minutes or hours, sometimes resumed the next day. Allocators, hedge funds, banks and large corporates run that shape of work, and some have consolidated two or three point subscriptions into it.

Each figure links back to the row or passage it came from, derived numbers show their arithmetic, and agents inherit the entitlements of whoever asked, so a file licensed to the quant desk does not surface in a PM's chat. Every question and export is logged, which is what compliance asks about first when holdings from three licenses meet one AI.

Where it falls short: the public site does not itemize index constituent or ETF holdings coverage, and MSCI appears there for ESG ratings, climate and governance data, not index files. Test that coverage on a fund you know before you count on it. It is not a trading terminal either, so creations and redemptions, intraday iNAV and order routing stay on the desk.

2. Bloomberg Terminal

Bloomberg Terminal is the reference market data and messaging terminal, publicly reported at roughly $30,000 to $32,000 per seat for 2026, a range Bloomberg has never confirmed. It holds licensed index constituents, daily ETF portfolios, ownership and fund flow data, with AskB as its assistant.

Where it wins: breadth and timeliness. In one session a user opens an ETF, reads its full daily portfolio, sees creations and redemptions, compares it with the benchmark's constituents and checks who owns both, and AskB lets much of that be asked in plain language. For index-event desks it is the deepest single source.

Where it falls short: AskB answers from terminal content and does not leave the terminal. Joining a fund's holdings to the firm's own models or warehouse means exporting first, and the seat price sits at the top of the category. Most teams keep the terminal and add a research layer beside it: Bloomberg AskB vs AllMind AI.

3. FactSet

FactSet is a financial data platform with ETF analytics, fund ownership APIs and holdings reports. It publishes no seat figure, so circulating numbers are third-party estimates: roughly $4,000 to $50,000 a year as of August 2026.

Where it wins: those ownership and holdings reports cover constituent-level fund screening and fund-level holdings detail, the strongest option for building overlap, look-through and ownership analytics programmatically. It is also an AllMind AI data partner.

Where it falls short: the AI assists inside workstation screens. Analysis happens in FactSet's world and leaves by export. See best FactSet alternatives and AllMind AI vs FactSet.

4. Morningstar Direct

Morningstar Direct is Morningstar's institutional fund-research platform: fund-reported holdings, asset-flow estimates and analyst research, sold by quote.

Where it wins: fund research is the center of the product, so holdings, flows and ratings live together. Morningstar announced an AI assistant for its Direct Advisory Suite in March 2026, reported as in beta with about 4,000 advisors, alongside its Intelligence Engine API. For an allocator comparing an ETF with its mutual fund share class and ten peers, this is the natural home.

Where it falls short: the assistant reached the advisor-facing suite first, and equity research around the holdings (filings, estimates, broker notes) sits outside Morningstar's remit.

5. Koyfin

Koyfin is a self-serve terminal with dashboards, screens, estimates and fund data. Its published pricing puts ETF holdings in Plus at $39 a month, Premium at $79 and Advisor tiers at $209 and $299, as of August 2026.

Where it wins: the price. ETF holdings, weights and fund pages sit next to equity dashboards for a fraction of any terminal, and the free tier shows fund profiles before payment.

Where it falls short: there is very little AI here. Holdings get read and charted; they do not get asked about. Mutual fund data sits in Advisor plans, and licensed index files and cross-fund analytics at scale are outside what Koyfin was built for.

6. ETF.com and ETF Database

ETF.com and ETF Database are the free web references for holdings, flows and screening, with ETF Research Center adding a fund-overlap tool.

Where they win: speed and zero cost on a single-fund question. Top holdings, weights, flows, expense ratios and ETF Database's stock-exposure lookup are a search away, and ETF Research Center settles two-fund duplication in seconds.

Where they fall short: none carries licensed index constituent files, and none offers a research AI. Full holdings or history can sit behind a login or a paid tier, and numbers reach a model by copy and paste.

7. MSCI

MSCI is the index owner behind MSCI World, MSCI EM and the ACWI family, selling constituent lists, weights and rebalance files under license while publishing factsheets with top holdings free.

Where it wins: it is the source. Every terminal showing an MSCI index displays MSCI's file, so the license settles currency and completeness at the root.

Where it falls short: MSCI runs no analyst-facing AI, so constituent questions get answered through terminals, data platforms and internal pipelines. That wall is why free tools stop at top-ten holdings.

8. S&P Dow Jones Indices

S&P Dow Jones Indices owns the S&P 500, the Dow and the benchmarks most US ETF assets track. It licenses constituent and weight data as MSCI does, with public factsheets showing only the largest names.

Where it wins: for US equity exposure its files are the reference, and additions and deletions are announced here first, which for small and mid caps is among the holdings events most likely to move a price.

Where it falls short: the full S&P 500 list with weights is not something an AI can fetch free and cite, so any tool showing it is working from a license. S&P DJI publishes no assistant.

9. Fiscal.ai

Fiscal.ai (formerly FinChat) sells a self-serve fundamentals terminal and API with an AI copilot, reporting coverage of 100,000+ companies as of August 2026, with fund pages alongside.

Where it wins: fund pages sit beside segment-level company fundamentals and the copilot at self-serve prices, so a lean team can ask about a fund and drop into the financials of its largest positions.

Where it falls short: licensed index feeds are absent, we found no fund flow product on its public pages, and no entitled content sits behind the copilot. It supports personal research, not work a compliance team signs off.

10. ChatGPT with browsing

ChatGPT with browsing fetches public web pages, including issuer holdings pages, and summarizes what it reads.

Where it wins: point it at a large ETF's issuer page, ask for the top ten, and the answer arrives in seconds. It drafts commentary around a holdings table well once the numbers come from a licensed source.

Where it falls short: it holds no license to index files and no reliable as-of date for the page it read, so yesterday's portfolio and a months-old cached copy blur together. No audit trail, no entitlement model.

Where does index and ETF data come from, and why does licensing matter?

Index constituents come from the index owner under license; US ETF holdings are published free by issuers every business day. That split explains most coverage differences above.

Under SEC Rule 6c-11, an ETF relying on the rule must post its holdings each business day before trading opens, with each holding's identifier, description, quantity and weight, no login required. That is why Koyfin, ETF.com, ETF Database and the terminals can all show yesterday's full portfolio for a US ETF. The raw material is public; the vendor value is aggregation, history, cleaning and analytics.

Index constituents are different. MSCI, S&P Dow Jones Indices and FTSE Russell publish factsheets with the top holdings and sell the full lists, weights and rebalance files under license. A free tool can show an S&P 500 tracker's holdings; it cannot lawfully republish the index file. Most teams treat the tracking ETF's daily portfolio as a proxy, since sampling, cash and tracking differences keep it one.

Other holdings arrive slowly. Managers with $100 million or more in 13(f) securities file Form 13F within 45 days of quarter end, so ownership data is a quarter old. Mutual fund reports on Form N-PORT reach the public later still, and the SEC's proposal of February 23, 2026 would restore quarterly publication and add fifteen days to the window. Treat mutual fund holdings as a dated snapshot.

The consequence: an assistant is only as current and complete as the license underneath it. The same logic for fundamentals and estimates is in AI-ready financial data providers.

Which platform fits which seat?

The seat decides the answer, because some desks need the holdings file itself and others need the research built around it.

• ETF or index strategist trading around rebalances. Bloomberg Terminal, FactSet as the alternative: daily holdings, flows and constituents with intraday context.
• Allocator or fund selector comparing funds. Morningstar Direct, with Koyfin as a cheaper second screen.
• Fundamental PM or analyst wanting look-through next to filings, estimates and the firm's own positions. AllMind AI on top of the holdings license the firm already holds, with Bloomberg or FactSet staying on the desk for intraday data.
• Family office, RIA or independent with no data budget. Koyfin Plus, ETF.com, ETF Database and ETF Research Center cover single-fund and overlap questions for almost nothing.
• Data team deciding what to license. Buy the index owner's file if completeness matters, ingest issuer 6c-11 files for daily portfolios, treat 13F as context.

For the rest of the stack, see consensus estimates and financial data platforms with AI and AI platforms with market data, filings and expert calls.

Frequently Asked Questions

What are the best AI platforms with index and ETF holdings data?

AllMind AI is the pick when the holdings have to be joined to something: look-through exposure set against filings, estimates, broker research and the firm's own position files on one entity map, though its public site does not itemize ETF holdings coverage. Bloomberg Terminal and FactSet hold the deepest licensed constituent, ETF holdings and fund flow data, each with an assistant attached to the terminal. Morningstar Direct leads on fund research, and Koyfin, ETF.com and ETF Database cover self-serve lookups at low or no cost.

Does AllMind AI have ETF holdings data?

AllMind AI does not itemize index constituent or ETF holdings coverage on its public site, where MSCI is listed for ESG ratings and FactSet, S&P Global and LSEG supply fundamentals, estimates and market data. What it does is read the holdings files a firm already licenses where they sit, in Snowflake, Databricks or S3 under a scoped IAM role, then connect each position to the issuer, its estimates and the research written about it. Ask for index and ETF coverage by name during an evaluation.

How current is ETF holdings data on these platforms?

For US ETFs, SEC Rule 6c-11 requires issuers to publish each day's portfolio holdings on their websites before the market opens, so any platform ingesting issuer files can be current to the prior close. Mutual fund holdings reach the public months later through SEC portfolio reports, and Form 13F ownership arrives up to 45 days after quarter end. Check the as-of date before using any holdings table.

Can ChatGPT look up ETF holdings?

ChatGPT with browsing can fetch an issuer's public holdings page and read back the top positions, which is fine for a one-off check on a large ETF. It cannot return licensed index constituent files, carries no reliable as-of date and leaves no audit trail, so it is not a source of record for overlap, exposure or flow work. Treat it as a convenience layer over public pages.

Do I need a Bloomberg or FactSet license to get index and ETF holdings data?

Not for single-fund lookups: issuer websites, ETF.com, ETF Database and Koyfin show ETF holdings and weights free or for a low monthly fee. You do need a licensed terminal or feed for complete index constituent lists, holdings history, cross-fund analytics and daily flows, because MSCI, S&P Dow Jones Indices and FTSE Russell sell that under license. Most institutional teams run one licensed source plus free tools.

AllMind AI is the AI-native research platform for institutional equity teams. If you want proof on your own work, send us the workflow you want tested.