Unity Software Inc. (NYSE: U)

1. What this business actually is

Unity sells the answer to three problems a game developer has in sequence. Building a game that runs on phones, PCs, consoles and headsets without writing it several times. Finding players once it is built. Making money from those players. The first is Create Solutions, the second and third are Grow Solutions, and the company's entire strategic argument is that owning both is worth more than owning either.

In FY2025 that produced $1,850m of revenue: Grow $1,228m (66%) and Create $621m (34%). Grow is the growth; Create is the reason Grow has an advantage. Every game shipped on the Unity engine is a candidate integration point for the ad SDK, which means Unity can source advertising supply it does not have to buy at the market price — and it sees conversion outcomes inside games it also tooled.

2. How one unit moves through the business

Create. The Unity Editor plus the runtime that ships inside the finished game, sold per seat. The free Personal tier is available only below $200,000 of revenue or funds raised in the prior twelve months; above that Pro is mandatory at $2,310 per seat per year, and companies above $25m of revenue must be on Enterprise at negotiated pricing (unity.com, September 2026). Revenue is recognised over the contract term because Unity treats the software and its updates as a single, interdependent performance obligation. At 31 December 2025, $494m of contracted revenue on multi-year terms had not yet been recognised, with $225m — 46% — expected within twelve months. That is the most predictable revenue Unity has, and it is a third of the company.

Grow. An advertiser, usually another game studio, bids to acquire a player. Unity's auction picks the winning bid and serves the impression inside a publisher's app. The publisher earns a payout; Unity keeps the difference and books only that difference as revenue. Cash follows the same shape: $431.5m of publisher payables against $643.6m of receivables at 31 December 2025 — the float is real, but receivables are larger, so growth in Grow consumes working capital rather than releasing it.

Unity discloses neither the gross flow of advertiser money nor its take rate, which is why the unit price above has to be inferred. It is the single most important undisclosed number in this business.

What has been cut, and why the economics differ

Since the fourth quarter of 2023 Unity has been shrinking to what it calls the Strategic Portfolio — the engine, its consumption services, and monetisation. Professional services were narrowed, independent development of professional artistry tools stopped, the ironSource Ads Network was sunset on 30 April 2026, and the Supersonic game-publishing business was sold to Tripledot on 4 August 2026 for $40m in cash. Every exit removed revenue that carried people, hosting or publishing risk rather than software margin — which is why gross margin rose from 66% in FY2023 to 74% in FY2025 while revenue fell and then flattened.

3. The industry, and where Unity sits in it

Engines sell avoided engineering cost to customers who treat them as an expense line. Ad networks sell matching — the ability to predict which player is worth what to which advertiser — and are paid out of marketing budgets. The second industry is where the money is.

The largest reliably collected toll in mobile is the app store's: Apple takes 30%, or 15% under its Small Business Program. Google Play moves the other way from 30 June 2026 in the US, EEA and UK, to roughly 20% plus a 5% billing fee on new-install transactions. In-app advertising is untaxed by the store, and on Apple's own commissioned estimate it reached $151bn in 2025 against $149bn of App Store digital-goods billings — advertising is now the larger pool, and it is the pool Unity fishes in.

The benchmark for what that pool pays is AppLovin, which reported FY2025 revenue of $5,481m, up 70%, net income of $3,334m and an 82% adjusted EBITDA margin. Unity's adjusted gross margin of 83% is in the same territory. Its adjusted EBITDA margin of 22% is not. The gap is not unit economics — it is the size of the cost base sitting on top of them, which is the single most important thing to understand about this company's profitability.

Which barriers actually bind

Engine share is genuinely contested. GDC's 2026 survey of more than 2,300 industry professionals puts Unreal at 42% as primary engine, Unity at 30% and Godot at 11%. That survey is self-selected and skews to PC and console production, so it does not measure Unity's mobile position — Unity claims 70% of the top 1,000 mobile games, but the claim is its own and carries a January 2025 footnote. No independent, methodology-stated measurement of mobile engine share was found.

Two barriers bind. Mid-project switching cost: an engine choice is locked once production starts, so revenue erosion from a share shift arrives years after the shift. And data scale in ad targeting, where the evidence is unusually clean — when Apple's App Tracking Transparency removed deterministic identifiers, trackable US traffic fell from 73% to 18%, average impression prices fell 23%, and trackable impressions began commanding a 51% premium (Skiera et al., hosted by the FTC). Privacy shocks did not shrink the ad pool; they concentrated it toward whoever held the most first-party conversion signal.

Two barriers are weaker than they sound. The asset store and plugin ecosystem are cited as lock-in, yet Godot has reached 11% primary-engine share with neither royalties nor a comparable ecosystem. And the installed base is not self-renewing: the GDC series shows primary-engine share moving materially within a few years.

Unity is the only company that owns both a major engine and a large mobile ad network — AppLovin, Moloco, Meta and Google can outbid it for demand, but none owns the authoring layer where games begin. The honest test is whether the pairing shows in the numbers, and so far it shows on one side only: in Q2 2026 strategic Grow revenue grew 63% while strategic Create grew 5%. The ads business is winning on model quality, not on any bundled advantage flowing from the engine. The supply asset is real; the cross-sell is not yet demonstrated.

The kind of company that wins in mobile ad matching is the one with the most first-party conversion signal at the lowest supply cost. Unity is structurally that kind of company — it has only recently built the modelling to exploit it, and the evidence that it now can is one year old.

4. The growth engine

Unity's reported growth and its underlying growth are two different numbers, and the difference is deliberate. In Q2 2026 total revenue rose 24% while strategic revenue rose 38%; the 14-point gap is the ironSource network being switched off and Supersonic being sold. Both numbers are true. Only one describes the business Unity will still own in 2027.

Source: Unity Q2 2026 results release, 6 August 2026. The same quarter's 10-Q reports the older Create/Grow Solutions cut, which ties to the same total but is not the same split. Growth today is entirely organic — Unity has made no acquisition since ironSource closed in November 2022, and its only 2026 commitment is $100m of AppsFlyer shares by the end of 2027. The 2022–2023 revenue line was largely bought; the 2026 line is not.

5. Margin, cash and what management does with it

Almost none of Unity's cost of revenue varies with an incremental impression, which is why gross margin expands sharply when revenue recovers. But the operating cost base is where this company is actually decided: headcount fell from 7,703 at the end of 2022 to 4,412 at the end of 2025, a 43% reduction, and operating expenses fell from $2,286m in FY2023 to $1,851m in FY2025 while gross profit fell only $82m. Unity did not become more profitable by selling more. It became more profitable by employing fewer people.

Sources: FY2020–FY2025 10-Ks; Q2 2026 revenue of $546.5m annualised. FY2022 includes ironSource for under two months and FY2023 for a full year, so the FY2022–FY2023 step is largely acquired rather than grown. Revenue is back at the prior peak on a deliberately smaller portfolio.

Sources: FY2020–FY2025 10-Ks; Q2 2026 10-Q. The Q2 2026 gross margin is shown for the quarter alone: the six-month figure is distorted by $227m of impairment charges booked in cost of revenue in Q1 2026. Adjusted EBITDA on the current definition is disclosed only from FY2024.

Sources: FY2022–FY2025 10-Ks. Comparability: the portfolio reset began in Q4 2023, and the definition of "non-strategic" was widened again in Q1 2026 to include the ironSource network and Supersonic, with prior periods restated. FY2022 operating cash flow was flattered by roughly $200m of prepaid licence fees received from Weta FX.

Cash conversion is strong and improving — FY2025 free cash flow of $403.9m alongside a GAAP net loss of $401.5m, and $268.4m in the first half of 2026 against $134.0m a year earlier. Capital expenditure of $19.0m is about 1% of revenue. But the business consumes a great deal of equity: stock compensation of $380.2m in FY2025 is 21% of revenue, and shares outstanding rose 48% from 292.6m at the end of 2021 to 432.9m at the end of 2025. Adjusted EPS of $0.86 is struck on 474.8m diluted shares against the 420.9m used for the GAAP loss per share.

Where the cash has gone, ranked

Acquisitions dominated, then undoing them. Unity paid roughly $2.92bn in stock for ironSource in November 2022, $1.53bn for Weta Digital's assets in December 2021 and $0.33bn for Parsec — about $4.8bn, nearly all in equity issued at high prices. It then repurchased $1.5bn of its own stock in 2022 and about $0.25bn in 2023 under a $2.5bn authorisation part-funded by a concurrent $1.0bn private placement, leaving $750m unused when the programme lapsed. There is no repurchase programme today and no dividend has ever been paid.

The recent behaviour is different in kind: buying back debt at a discount. Unity repurchased $480m face of its 2026 notes for $415m in March 2024 and $688m face for $642m in Q1 2025, funded by issuing $690m of zero-coupon 2030 notes. The pattern is a company that bought growth with expensive equity at a market peak, then spent three years slowing the share count and retiring the debt taken on to support it. The Weta assets bought for $1.53bn have since had their remaining useful lives cut from four-to-seven years to one-to-three — an accounting admission that they will not be used as planned.

Post-dating the FY2025 accounts: Supersonic sold on 4 August 2026 for $40m; a binding obligation to buy $100m of AppsFlyer shares by the end of 2027; and $558m of zero-coupon notes maturing in November 2026, which Unity intends to settle in cash against $2.36bn on hand.

6. Cyclicality, and where the business sits in its own cycle

Unity is exposed to two cycles at once. Grow depends on advertiser budgets, held under revenue-share arrangements with no minimums. Create depends on the number of employed game developers — a population that has been shrinking, with 28% of respondents to GDC's 2026 survey laid off in the preceding two years. Neither exposure is hedged by contract.

Source: FY2025 10-K, revenue disaggregated by customer invoice address.

Revenue peaked at $2,187m in FY2023, troughed at $1,813m in FY2024, and Q2 2026 annualises to $2,186m — back at the prior peak. Margins are not back at prior levels; they are above them. Adjusted EBITDA margin was 29% in Q2 2026 against 21–22% in FY2024 and FY2025, and both gross margin and free cash flow are at records. Unity is at a cyclical and structural high on profitability while its industry's volume backdrop is flat — the configuration in which reported margins are least likely to be a fair guide to the next few years.

The industry's downside case would reach Unity through supply, not demand. If engine share drifts — Godot at 11% and free is the live example — fewer new games embed Unity's SDK, the ad network's owned supply footprint erodes with a multi-year lag, and its first-party conversion signal weakens exactly where the ATT evidence says signal is worth most. That is a margin mechanism, not just a volume one, because Grow's economics depend on not having to buy supply at the market price.

7. Risks the cycle does not capture

That matters because retention is the one thing an outside investor could previously test independently of the revenue line. Without it, a deteriorating expansion rate becomes invisible until it shows up in revenue itself — by which point it is a year old.

The moat and the growth are in different halves of the company. Create is what makes Unity hard to replace, and it grew 5% in Q2 2026 and 1.2% in FY2025. Grow is where the growth is, and it competes with Meta, Google, Apple and AppLovin on modelling quality, where scale compounds. If Create's seat base stagnates while Godot takes indie share, the ad network's structural supply advantage decays quietly over several years — long after the point at which it could be arrested.

Growth rests on a single model migration with no contractual buffer. Grow revenue is earned under revenue-share arrangements with no minimums, and customers can move budget between networks within days. Unity's own risk disclosure makes the point: should customers "lose confidence in the value or effectiveness of our Grow Solutions, consumption of these offerings could decline" (FY2025 10-K).

The equity itself is a recurring cost. Stock compensation of $380m in FY2025 equals 21% of revenue and 93% of adjusted EBITDA. The share count has risen 48% since 2021 with no repurchase programme in place, and the August 2026 award of 880,000 price-vesting units to the chief executive — vesting on 30-day average prices of $50, $60 and $75 — sets management's incentive on the share price rather than on the operating metrics tracked here.

The Unity China redemption right is a cash claim ahead of shareholders. Investors who bought 20.5% of Unity China for $197m in 2022 can require Unity to repurchase their interest at a floor of ¥1.9bn, with a redemption date of August 2027; $252.6m of redeemable non-controlling interests sat on the balance sheet at 31 December 2025. It is outside the conventional net-debt calculation and falls due alongside the 2027 convertible notes.

Unity's largest competitors are also its distribution. Apple and Google set the store rules, own the identifiers that determine targeting quality, and sell competing ad products. Google's October 2025 retirement of Privacy Sandbox removed a symmetric threat on Android — a downside withdrawn, not an upside created — while the permitted US commission on Apple link-outs remains unresolved before the Supreme Court.

What the sources could not answer

8. What to take away