The New York Stock Exchange is what most people associate with ICE.

It is not why I own the stock.

What interests me is the collection of networks behind it: exchanges, clearing houses, energy benchmarks, fixed-income data, mortgage software, and now MarketAxess.

ICE keeps building infrastructure around markets that are difficult to operate. Once customers organize liquidity, data, workflows, and risk management around those networks, they become hard to displace.

The other thing I like is the mix. Roughly half the business is recurring, while the transaction side can benefit when customers need to trade, hedge, clear, or manage more risk.

ICE reports three operating segments:

In the second quarter of 2026, ICE generated $2.7 billion of net revenue.

Exchanges produced $1.46 billion, Fixed Income and Data Services produced $645 million, and Mortgage Technology produced $557 million.

The economics are very different across the three businesses.

The exchange segment had a 75% adjusted operating margin.

Fixed Income and Data Services was at 46%.

Mortgage Technology was at 43% on an adjusted basis.

The exchange business is still the earnings engine.

But the company ICE has been building for the last decade is broader than an exchange operator.

That matters because each piece responds differently to the market environment.

An exchange is useful because other people already use it.

That sounds obvious, but it is the foundation of the moat.

A trader wants the venue with deep liquidity.

Liquidity attracts more traders.

More traders improve price discovery.

Better price discovery makes the benchmark more useful.

The benchmark then becomes harder to displace.

ICE has spent more than two decades building this dynamic across energy, financial futures, equities, options, and clearing.

Its Brent crude contract sits at the center of a global oil network that ICE says includes more than 800 related crude and refined oil products.

That is more important to me than simply saying ICE owns an exchange.

The valuable asset is the network built around the benchmark.

At the end of 2025, total futures and options open interest on ICE was roughly 103 million contracts, up 17% from the prior year.

By the second quarter of 2026, ICE said open interest across its markets was up another 20% year over year.

Open interest is not revenue by itself.

But it is evidence that customers are continuing to build positions and manage risk inside ICE's markets.

That creates future trading, clearing, data, and connectivity opportunities.

ICE is unusual because volatility can be good for business. When energy prices move sharply, rates change, currencies shift, or geopolitical risk rises, customers have more reason to hedge.

I am not buying ICE because I think I can predict volatility. I like that periods that hurt many businesses can create more activity on ICE’s networks.

The first quarter of 2026 was a good example.

Against a volatile macroeconomic backdrop, ICE reported record quarterly net revenue of $3.0 billion and adjusted earnings per share of $2.35.

But a company that depended entirely on trading activity would still be more cyclical than I want.

That is why the recurring side of ICE matters so much.

In 2025, ICE generated $5.06 billion of recurring revenue and $4.88 billion of transaction revenue.

That is almost a 50/50 split.

In the second quarter of 2026, recurring revenue increased 8% to $1.35 billion.

The mix is even more recurring in the newer businesses.

Fixed Income and Data Services generated $531 million of recurring revenue in Q2 against only $114 million of transaction revenue.

Mortgage Technology generated $406 million of recurring revenue against $151 million of transaction revenue.

I think this is one of the most important changes in ICE's history.

Founder and CEO Jeff Sprecher has said publicly that he would rather own a predictable, compounding subscription business than a purely episodic transaction business.

That is exactly what ICE has been moving toward.

The company started as an electronic energy market.

Today, recurring subscriptions, data, listings, software, and connectivity make the business much less dependent on any single quarter's trading volume.

The transaction engine gives ICE upside when markets become active.

The recurring engine gives it visibility when they do not.

That is why management calls the model "all-weather."

The phrase is marketing language.

The revenue mix gives it substance.

ICE's Fixed Income and Data Services segment is easy to overlook because it does not have the brand recognition of the NYSE.

I think it may have one of the best long-term runways inside the company.

The business includes fixed-income pricing, reference data, analytics, indices, execution, credit-default-swap clearing, and network technology.

In Q2 2026, fixed-income data and analytics revenue grew 9%.

Data and network technology grew 11%.

The segment generated $645 million of revenue, with more than 80% coming from recurring sources.

At the end of 2025, annual subscription value in the segment was nearly $2.0 billion, up 8.3%.

The attraction is similar to what I like about S&P Global.

Customers build data, pricing, analytics, and workflows around the product.

Once that happens, ICE becomes part of how the institution operates rather than another piece of software that can be swapped out casually.

I would not call the switching costs absolute.

But I think the combination of proprietary data, workflow integration, distribution, and customer habit creates a strong position.

ICE agreed in July to acquire MarketAxess for roughly $6 billion in cash.

This is now central to the investment thesis.

MarketAxess operates one of the leading electronic institutional bond trading networks.

ICE already has fixed-income pricing, reference data, indices, analytics, and a retail and wealth bond trading franchise.

MarketAxess adds the institutional execution network.

I see the logic.

Fixed income is still far less electronic than equities.

ICE is effectively trying to connect data, price discovery, execution, indices, and post-trade tools into one ecosystem.

That is the same playbook Sprecher has used for years: find an inefficient market, add technology and data, connect participants, then build more services around the network.

The announced economics also look reasonable on the surface.

ICE said the enterprise value is approximately $5.7 billion, or about 10.6 times MarketAxess trailing earnings before interest, taxes, depreciation, and amortization after full run-rate cost savings.

Management expects $100 million of annual expense savings and says the deal should increase adjusted earnings per share in the first full year after closing.

But I do not give management automatic credit simply because a deal is accretive.

The transaction is financed with debt.

ICE expects gross leverage around 3.4 times after the deal and plans to bring that back to 3.0 times or below within 18 to 24 months.

That is manageable if the cash flows perform as expected.

It also reduces room for error.

The MarketAxess deal makes the fixed-income opportunity more interesting and the balance-sheet discussion more important at the same time.

Mortgage Technology is the part of ICE I have the most mixed feelings about.

The strategic idea is compelling.

ICE wants to digitize the mortgage process from origination through closing, servicing, and data.

Encompass is a system of record for loan origination.

The servicing software ICE acquired with Black Knight helps manage loans from boarding through final payment or default.

Many of those servicing contracts run five to seven years.

That creates embedded workflows and recurring revenue.

In the second quarter, Mortgage Technology produced $557 million of revenue, up 5%.

Recurring revenue was $406 million.

The adjusted operating margin was 43%.

The issue is what ICE paid to build this franchise.

ICE acquired Black Knight in 2023 for approximately $11.8 billion of consideration before required divestitures.

It took on meaningful debt and created a large amount of goodwill and intangible assets.

The company was still carrying $19.8 billion of debt at the end of June 2026, before the MarketAxess acquisition closes.

I don’t think Black Knight was obviously a mistake. But after paying roughly $11.8 billion and adding meaningful balance-sheet complexity, ICE now has to prove that the mortgage assets can earn an attractive return over time. Revenue growth by itself is not enough.

Jeff Sprecher is still ICE's founder, chair, and CEO.

He originally built the company around electronic energy trading and has spent more than two decades expanding through acquisitions and internal development.

The NYSE, Interactive Data, Ellie Mae, Black Knight, and now MarketAxess all came through that capital-allocation strategy.

This is not a company where I can analyze the existing assets and ignore management's next move.

Sprecher is a builder.

That has created enormous value historically.

It also means acquisition discipline matters more here than it does at a company that simply returns every excess dollar to shareholders.

I view that as both a strength and a risk.

This is where I care less about whether an acquisition is immediately accretive and more about whether the acquired asset becomes more valuable inside the existing network.

The skill is not buying businesses.

The skill is buying infrastructure that becomes more valuable inside ICE than it was on its own.

MarketAxess will be another test.

ICE has now reported record annual revenue for 20 consecutive years.

In 2025:

The first half of 2026 remained strong.

Adjusted earnings per share reached $4.25, up from $3.54 the prior year.

Adjusted free cash flow reached $2.6 billion, up from $2.0 billion.

ICE also reduced its diluted weighted average share count from 576 million to 568 million during the first half.

Through June, the company repurchased $1.2 billion of stock and paid $591 million in dividends.

The board subsequently increased the repurchase authorization to $4 billion.

ICE clearly generates plenty of cash.

The question is how much of that cash should go toward acquisitions, debt reduction, repurchases, and dividends.

With MarketAxess coming, I would put debt reduction ahead of aggressively increasing buybacks.

ICE closed October 5 at $152.15 per share.

On a trailing adjusted basis, I calculate roughly $7.66 of earnings per share using full-year 2025 results and replacing the first half of 2025 with the first half of 2026.

That puts the stock at roughly 20 times trailing adjusted earnings.

The expected 2026 dividend is $2.08 per share, or roughly a 1.3% yield at the current price.

Twenty times earnings is not cheap in isolation.

For a business with exchange network effects, recurring data revenue, high margins, and a long history of compounding, it is a valuation I take seriously.

I would frame the return in a few simple cases.

If adjusted earnings per share grow 6% annually for five years and the stock ends at 18 times earnings, the implied share price is around $185.

That is only about a 4% annual price return before dividends.

If earnings grow 10% and the multiple remains around 20 times, the implied value is roughly $247.

That is close to a 10.2% annual price return before dividends.

If earnings grow 12% and the market pays 22 times earnings, the implied value approaches $297, or roughly a 14.3% annual price return before dividends.

Those are not forecasts.

They show why the current price is interesting without making the stock obviously cheap.

The return still depends on continued earnings growth.

And adjusted earnings deserve scrutiny because ICE has repeatedly acquired businesses and adds back acquisition-related intangible amortization.

I understand why management presents the adjustment.

I do not treat acquisition costs as though acquisitions are unrelated to the business model.

That is especially relevant for a serial acquirer.

The risks I am watching most closely are:

Market activity normalizes. A prolonged period of low volatility could reduce trading and clearing revenue across parts of the exchange business.

Energy benchmarks lose relevance. ICE's energy franchise is extremely valuable because market participants already organize around its contracts. Regulation, competing benchmarks, or structural changes in energy markets could weaken that position.

MarketAxess integration disappoints. ICE is paying cash, adding debt, and underwriting cost savings and strategic benefits. If the execution network does not combine well with ICE's data assets, the deal could earn an inadequate return.

Mortgage Technology never earns enough on the capital invested. The assets may be sticky while still producing returns below what ICE paid for Black Knight and related businesses.

Debt limits flexibility. ICE had $19.8 billion of debt before the MarketAxess deal. The acquisition increases leverage again.

Regulation changes market economics. Exchanges, clearing houses, energy markets, fixed-income trading, and mortgage technology are all regulated businesses.

Management becomes too acquisition-driven. I want ICE buying assets because the combined network becomes more valuable, not because a larger company looks better.

I would become more cautious if:

A quieter quarter in trading would not bother me.

A weaker network would.

What interests me about ICE is not that it owns the New York Stock Exchange.

It is that ICE repeatedly finds markets where participants need trusted infrastructure.

Energy producers need hedging markets.

Investors need price discovery.

Clearing members need risk management.

Bond investors need pricing, data, and execution.

Mortgage lenders and servicers need software that sits inside the workflow.

ICE gets paid for connecting those participants.

The business also has a useful internal balance.

When markets become volatile, transaction revenue can benefit.

When they are quieter, recurring data, software, listings, and connectivity revenue keep coming in.

That is a difficult combination to find.

The biggest question for me is capital allocation.

ICE has built an exceptional collection of financial infrastructure by acquiring aggressively and integrating well.

That history deserves respect.

It does not remove the need to judge Black Knight, MarketAxess, debt, and repurchases on the economics of each decision.

At roughly 20 times trailing adjusted earnings, ICE is interesting to me.

The quality of the exchange and data franchises is not the hard part.

What I have to decide is whether MarketAxess, debt reduction, and future repurchases can create as much value as the last round of capital allocation.

Disclosure: I own shares of Intercontinental Exchange as of the date of publication. This research reflects my personal views and is not investment advice.