Vector Accelerates as the Amortisation Burden Collapses

Unity 2Q 2026: Vector Accelerates as the Amortisation Burden Collapses

A reactive read of Unity Software's second-quarter print, and a follow-on to our May note on the first quarter. Every line management guided in May came in above the top of its range, the balance sheet flipped to net cash, and the commitment to GAAP profitability moved forward a quarter. The operating result and the accounting result are both real, and they are doing different work.

Executive Summary

Unity ($U) reported 2Q 2026 on 6 August, and cleared the top of every range it guided in May. Total revenue of $546.5m grew 23.9% year-over-year. Strategic Grow, the line carrying Unity Vector, printed $329.0m for growth of 63.4% against a guided 50% to 52%, beating the top of its range by $23.0m. Adjusted EBITDA of $160.2m beat by $25.2m at a 29.3% margin. Free cash flow was $202.0m and the GAAP net loss narrowed to $22.7m from $107.4m a year ago.

In May we set out five things to test in this print. Three came back clean. Strategic Grow beat its guide rather than merely meeting it. The Supersonic divestiture closed on 4 August, to Tripledot Studios, ending an exit that had been at the financial-advisor stage in May. Stock-based compensation fell to 13.8% of revenue, below the 15% target management had named as one of three drivers of GAAP profitability, and CFO Jarrod Yahes moved the profitability commitment itself forward from 4Q26 to 3Q26. Create stabilised at the top of its guided range but guided down again for the third quarter. The print adds no new evidence on Vector's competitive position.

Three takeaways:

1. The Vector beat is organic, and the segment line understates it. Management guided Vector to 12% to 13% sequential growth and delivered nearly 23%, against four prior quarters at roughly 15%. Yahes disclosed that only $3m of that came from ironSource customers migrating after the network shut, so the beat is not explained by revenue moving between Unity's own networks.
2. Acquisition accounting lowered the GAAP hurdle, and the operating result is now clearing it. Amortisation of acquired intangibles fell from $117.4m to $77.5m, and intangible assets net from $650.5m in December to $163.1m in June, which structurally reduced Unity's GAAP break-even point. On our arithmetic the third-quarter guide crosses into profit across its whole range with the second-quarter reconciliation held constant, and would imply a loss above $30m at the first quarter's burden. Further declines in amortisation and restructuring set the size of the profit rather than whether it appears.
3. Greater China is now 21.0% of revenue and contributed more year-over-year growth than the United States. It grew 37.0% to $114.5m, a sixth consecutive quarter of sequential growth, and supplied 29.3% of the revenue increase against the US at 25.6%. Strip it out and Unity grew 20.9% rather than 23.9%. Yahes cited it twice, both times behind Create. Across eight analyst questions, nobody asked about it.

Financial Highlights

Headline results:

2Q26 against the guide issued in May:

3Q26 guidance:

Two features of this guide are worth separating. The Grow acceleration is genuine, with Yahes assuming 19% to 21% sequential Vector growth against the 12% to 13% assumed going into the second quarter. The headline rates are flattered by the denominator. Non-strategic revenue falls from $60.1m to $20m and is gone by the fourth quarter, so total revenue growth of roughly 20% sits under strategic growth of 44% to 47% for reasons unconnected to the operating business. The two bases converge at the end of this year, and the comparison gets harder.

A note on quality of earnings. The margin expansion is in the right places. Adjusted cost of revenue held at 17.1% of revenue, unchanged year-over-year, so gross economics scaled rather than improved. Adjusted research and development rose 17.2% in dollars to $161.1m, which is a company still spending on product. The leverage came from adjusted sales and marketing, down 2.2% in dollars and 460bps as a share of revenue, and adjusted general and administrative, down 8.6% in dollars. Adjusted gross margin was 83%, within the 82.5% to 84.0% band it has held for nine quarters. The savings came from lines that can stay down, rather than from deferring product spend.

The cash flow needs one adjustment. Second-quarter free cash flow of $202.0m is 126% of adjusted EBITDA, which no business sustains. Working capital contributed $61.7m in the quarter, $63.1m of it from publisher payables, the float created when Unity collects from advertisers before settling with publishers. That float grows while Vector grows. The six-month figures strip most of the timing out: free cash flow of $268.4m against adjusted EBITDA of $298.5m is 90% conversion, with working capital a net $19.9m drag across the half. Ninety percent is the better current anchor, though cash taxes, the post-Supersonic mix and the behaviour of publisher float as Vector scales could all move it.

Cash and equivalents of $2,352.0m against $2,237.4m of carrying-value convertible notes puts Unity in a net cash position of roughly $115m, against net debt of about $90m three months ago. The $557.2m of 2026 notes will be repaid in cash at their November maturity from the balance sheet, leaving about $1,795m of cash and $1,680.2m of notes. Goodwill of $3,155.6m was not touched by the first-quarter impairment, its only movement since December being the $10.7m reclassified into the Supersonic disposal group, and at 47.4% of total assets it exceeds book equity of $3,040.2m. That is a monitoring item rather than a thesis.

The Amortisation Burden and Unity's GAAP Break-Even

The GAAP loss from operations improved from $351.4m in the first quarter to $32.2m in the second, a swing of $319.2m. Two acquisition-accounting charges account for almost all of it.

The $21.9m improvement in adjusted EBITDA was more than offset, on the GAAP bridge, by $24.5m of additional restructuring expense. The $318.6m came from the impairment not repeating and from acquired-intangible amortisation falling from $117.4m to $77.5m. Both follow from the same decision: writing down the ironSource and Supersonic asset base in the first quarter removed assets that were generating amortisation.

That explains why the second quarter looks so much better than the first. It does not, on its own, explain why management moved the profitability date. The impairment was taken in the first quarter, and the commitment to GAAP profitability by 4Q26 was made in May, after it. Whatever moved the date from the fourth quarter to the third was disclosed in August, not in March.

The two facts connect at the break-even point rather than at the announcement. At $117.4m a quarter, acquisition amortisation left even the current third-quarter EBITDA guide well below GAAP break-even. At $77.5m, and falling, the hurdle is one the operating business can reach. On the guide, it already does. Holding the second-quarter EBITDA reconciliation constant, along with below-the-line interest and tax, the third-quarter adjusted EBITDA guide of $185m to $190m implies GAAP net income of $2.1m to $7.1m. The crossing does not require another amortisation step-down or a restructuring normalisation. Those determine the size of the profit: restructuring returning to the $7m quarterly run-rate Unity carried through 2025 would add roughly $24m, and a further amortisation decline to $58m would add roughly $19m.

The counterfactual sets out what the accounting contributed. Run the same third-quarter guide with amortisation still at the first quarter's $117.4m and everything else unchanged, and Unity posts a GAAP loss of $32.8m to $37.8m. The pull-forward is therefore an operating result meeting a hurdle that acquisition accounting had already lowered, rather than an accounting event in itself. Both halves are load-bearing.

None of this is ordinary scheduled runoff. Unity accelerated the amortisation before impairing and shortening what was left, across three decisions the filings date precisely. Effective 1 July 2025, it cut remaining useful lives on certain intangibles, primarily developed technology from the 2021 Wētā FX acquisition, from four-to-seven years to one-to-three, disclosing the effect as roughly $39m of additional amortisation. The accounts show it: the charge jumped $39.1m between 2Q25 and 3Q25, from $86.2m to $125.3m. The first-quarter impairment then removed $271m of carrying value outright, $225m of it developed technology. Effective 1 April 2026, Unity shortened lives again on the ironSource and Supersonic assets, from two-to-three years to one-to-two. The $125.3m peak was manufactured by the first revision, and the $77.5m now reported is what survived the write-down.

The endpoint is better bounded than it was without being fixed. Assets reset to one-to-two years on 1 April 2026 amortise out between 2Q27 and 2Q28, and management guided in May to acquisition amortisation below $25m for FY27, which together imply the bulk of the remaining $163.1m runs off across the second half of this year with a tail into 2027. The quarterly filings carry no year-by-year schedule, so until the FY26 annual report provides one, the profile is bounded by disclosed useful lives rather than calculated.

Vector, and What the Segment Line Does Not Show

Bromberg guided Vector to 12% to 13% sequential growth for the quarter and delivered close to 23%, against four prior quarters at roughly 15%. He put the product at "substantially over $1bn in annual run rate," reached two quarters earlier than the company expected, six quarters after the product did not exist.

Two disclosures narrow the range of explanations. Yahes confirmed that only $3m of second-quarter Vector growth came from ironSource customers moving across after that network shut on 30 April, against a $50.3m sequential rise in Strategic Grow, so the migration is about 6% of the increase. Unity did not shut one network and book the revenue in another. And Vector grew close to 23% sequentially while Strategic Grow, which contains it, grew 18.0%, so the remainder grew more slowly than the segment.

That remainder can be sized. On the first-quarter call management put Vector at 80% of the strategic Grow line, implying roughly $223m in 1Q26 and a non-Vector remainder near $56m. Growing Vector at the stated rate gives $272m to $274m in the second quarter, 83% of the segment, and a remainder of $55m to $57m. Within the rounding the legacy remainder is flat, and the implied run-rate of $1,088m to $1,097m reconciles with Bromberg's "substantially over $1bn." Effectively all of Strategic Grow's growth is one product.

Unity's disclosure of that product has moved denominators every quarter: 56% of Grow revenue in 4Q25, up from 49% two quarters earlier; 80% of strategic Grow revenue in 1Q26, a different base; and in 2Q26 no share figure at all, only a growth rate and a run-rate. A reader cannot build a consistent Vector series from what has been disclosed, which is worth registering while that growth rate is the central input to the equity case.

The strategic claim underneath the numbers is runtime data. Unity began feeding signals from the roughly 3bn monthly players of Unity-built games into the Vector models late in the quarter, delivering on the schedule management set out in May. Bromberg described the connection as "a deep and sustainable competitive advantage" and set it against the structural position of rivals: "If you're Meta, you own your apps. If you're us, you have access to the 3bn people playing a made-with-Unity game."

He then declined to size it. Asked by Morgan Stanley whether runtime was a material contributor, Bromberg said he did not know "if there's any analytical value for anyone to single out the impact of runtime data," attributing performance to product enhancements, data quality and model improvement together. That is defensible for a self-learning system where the inputs are not separable. It sits awkwardly against Unity's own precedent: on the first-quarter call management quantified the day-28 ROAS product precisely, citing an 80% incremental campaign lift and a 37% ROAS improvement over the seven-day benchmark. The company quantifies product performance when it chooses to. The 23% arrived in a quarter containing more than 20 Vector updates and the full rollout of that same day-28 product, whose campaign spend tripled sequentially with over 25% of advertisers adopting it. Those are the drivers Unity has attributed. The runtime contribution is asserted rather than measured.

The test that would settle it. Runtime data and ordinary model iteration produce the same thing in the revenue line, so separating them needs evidence Unity has not been asked for, and each candidate has a precedent in what it already reports. A performance lift attributable to runtime signals, stated the way the day-28 lift was. Vector's share of strategic Grow on a stable denominator, so the series can be tracked rather than re-based. Spend per advertiser or retention, given management's claim that the market is fully penetrated and growth comes from optimisation rather than new logos. And the natural experiment Bromberg set up himself when he told analysts Unity will not ship 20 enhancements every quarter: if Vector's growth holds through a quarter of slower cadence, the advantage is in the data. If it tracks the cadence, it is in the shipping.

Greater China Is Now a Fifth of Revenue

The supplemental disclosures carry a geographic split the release does not discuss.

Greater China has grown sequentially for six consecutive quarters, from $65.2m in 1Q25 to $114.5m now, and its share of revenue has moved from 13.8% two years ago to 21.0%. Of the $105.5m of year-over-year revenue growth, it supplied $31.0m, or 29.3%, against $27.1m from the United States. EMEA, still the largest single region, grew 14.9% year-over-year and declined 0.3% sequentially, the only region to fall quarter-on-quarter.

Two definitions matter before anything is read into this, and the 1Q26 10-Q supplies both. Unity allocates revenue by the invoice address of the customer, not by where a game is played or where its revenue is ultimately earned, and its Greater China line includes Hong Kong and Taiwan alongside the mainland. Both loosen the geopolitical reading: this measures where Unity bills, which for a developer with global distribution says little about where the underlying economics sit.

What survives the qualification is the growth dependence, and it is straightforward to size. Excluding Greater China, Unity grew 20.9% year-over-year against the 23.9% it reported, so the region's outperformance is worth roughly 300bps of consolidated growth. Had it grown 10% rather than 37%, the quarter prints 18.8%; flat, and it prints 16.9%. Yahes named it twice, as a driver of the Create result alongside price increases and minimum annual commitments, and again in the Create guide, which puts it behind the weaker of the two segments. Unity's own risk disclosures list adverse changes in the US-China relationship among the factors that could cause results to differ materially. Neither the release nor the prepared remarks quantified the exposure beyond the geographic table, and none of the eight analyst questions addressed it. This is a growth dependence worth monitoring rather than something that changes the case on Unity today.

Create Remains ARPU-Led, and the Catalyst Sits in 2027

Strategic Create delivered 13.6% year-over-year growth adjusted for the prior-year term licence, the top of the 11% to 14% guided range and the answer to one of the questions we set in May. The sequential series is less encouraging. Strategic Create has printed $148.1m, $161.8m, $153.7m and $157.5m across the last four quarters, a range of $14m and no trend. The third-quarter guide of $159m to $163m implies 1% to 4% sequential growth and steps the year-over-year rate down to 7% to 10%.

Management attributes the growth to ARPU, "supported by price increases and minimum annual customer commitments, as well as strong growth in China." Two of those three are monetisation mechanisms: price increases and contractual minimums raise revenue per customer without adding one. Greater China is a growth driver of undisclosed composition, and could be higher spend per customer, more customers, more seats, mix, or some combination. So Create is ARPU-led on the drivers management has identified, with the customer-versus-ARPU split of its fastest-growing region left open. Raising monetisation per customer is a legitimate way to grow a subscription business and Unity has executed it, but price has a ceiling that seat growth does not, and nothing in these disclosures shows which lever is doing how much.

The product answer is Unity 7, announced at Unite Seoul two weeks before the call: a re-architecture in which developers and coding agents work side by side, with the MCP made free, the API and command line opened, and no upgrade break from Unity 6. Bromberg was explicit that the strategic point is not the editor: "We don't care how people access the Unity UI, the application itself. The application layer is just the surface piece." The intent is to widen the funnel of people building interactive content and configure Vector, commerce and live-service products automatically for whatever they build. Unity Commerce reached general availability on 30 June, free to developers, with Bromberg naming visibility into purchase behaviour as a benefit to the ad models ahead of the direct economics.

The timing is the constraint. Unity 7 enters beta in 4Q26 and reaches full release in 1Q27, so the full-release catalyst sits in 2027 and there is little room for it to alter the 2026 Create trajectory materially. Until then Create grows on the levers above, against a guide that decelerates.

Transcript and Management Commentary

Bromberg framed the whole business as one flywheel and asked investors to stop reading the segments separately, returning to it in three separate answers: cheaper game creation drives more games, more games make discovery harder, and harder discovery drives the ad business. The construction is coherent, and it is also untestable in any single quarter, which is what makes the unquantified runtime contribution material.

Yahes put the margin trajectory on a five-quarter record and extended it. Adjusted EBITDA margin has expanded every quarter since 1Q25, averaging roughly 200bps, and the third-quarter guide of 33% is a step of nearly 400bps he attributed to operating leverage "amplified by cost reductions enabled by our strategic actions." Stock-based compensation fell 25% year-over-year to 13.8% of revenue, its lowest level as a public company, against 35% of trailing revenue two years ago.

On advertiser acquisition, management described a market that is already fully penetrated. Asked by Bank of America, Bromberg said "the vast majority of folks already do know who we are and are already spending with us to one degree or another," with growth coming from optimisation and account management rather than new-logo sales. That is favourable for Vector's competitive position and unfavourable for anyone modelling growth from customer count, since it makes the growth rate a function of continued performance gains rather than distribution.

Two transactions carry figures management did not say out loud. Unity's investment in AppsFlyer, made alongside Meta ($META), Alphabet ($GOOGL) and Moloco, is a binding obligation to acquire $100m of shares by the end of 2027, disclosed in the commitments note rather than on the call. Yahes framed it as preserving "competition, choice, and innovation in the mobile attribution and measurement ecosystem." Four parties competing for the same advertising budgets are jointly funding the layer that measures whether that spending works, and Unity's share is $100m. Supersonic went to Tripledot for $40m of cash against a disposal group carried at roughly $36m net, with no impairment recognised on the held-for-sale measurement. Neither figure appeared in the release or the prepared remarks.

What to Watch

• Vector's sequential rate against the 19% to 21% assumption in the third-quarter guide, and whether management ever puts a number on the runtime contribution. Four quarters at roughly 15%, then 23%, now a guide above the old run-rate.
• The margin of safety on the third-quarter GAAP crossing rather than the crossing itself, which the guide clears on unchanged reconciling items. A year-by-year remaining-amortisation schedule, absent from the quarterly filings, would settle how far the burden falls from $77.5m and when.
• Create's sequential progression, and any disclosure of what drives Greater China. Range-bound across four quarters, guided to decelerate to 7% to 10%, with Unity 7's full release outside 2026. Whether the Greater China contribution is customers or spend per customer remains unanswered.
• Greater China at 21.0% of revenue, growing 37% and worth roughly 300bps of consolidated growth, cited by the CFO behind the weaker segment, named in Unity's risk factors, unaddressed on the call.
• The scale-up in hosting commitments. Non-cancelable purchase commitments, substantially all data-centre hosting, rose from $727.8m at March to $1,420.8m at June, with $823.2m of the total falling in 2027 and 2028. Read the increase carefully: the June footnote scopes the disclosure to commitments with original terms of more than one year, a qualifier the March footnote does not carry, so the two are not certainly like-for-like, and the movement could reflect new contracts, extensions, minimum-spend terms or presentation as easily as incremental capacity. What is not in doubt is the level, and that a company guiding to 33% EBITDA margins now carries $1.4bn of committed hosting against $2,357.1m of cash. Neither the figure nor the increase came up on the call.

What would change our view. Vector reverting to mid-teens growth for two consecutive quarters while its share of strategic Grow stalls, which would recast this quarter as a step rather than a trajectory. Runtime remaining unquantified into 2027 while growth decelerates, resolving the moat question against management by default. Create failing to reaccelerate two quarters after Unity 7's full release. Adjusted EBITDA margin stalling once restructuring normalises, which would identify the expansion as a cost programme rather than operating leverage. Any of the four would matter more than a single quarter's revenue miss.

Closing

Unity's second quarter beat every line it guided in May, by $23m on Strategic Grow and $25m on adjusted EBITDA, and disclosed enough to rule out the ironSource migration as the explanation. Adjusted EBITDA margin has expanded for five straight quarters, cash conversion across the half was 90%, and the balance sheet is in a net cash position for the first time. On the operating numbers, management's description of the quarter as the best in the company's history as a public company holds.

The GAAP milestone is best read alongside that rather than as a separate achievement. Writing off the ironSource and Supersonic intangibles in the first quarter cut the amortisation burden from $117.4m to $77.5m a quarter and lowered the adjusted EBITDA needed to reach break-even. The second-quarter beat and the third-quarter guide are now enough to clear it, and on our arithmetic they clear it across the full guide range. Further declines in amortisation and restructuring decide how large the first GAAP profit is. What the disclosures do not yet show is where the amortisation burden settles after FY27, and with it the GAAP margin the business carries once acquisition accounting stops flattering the comparison.

Marvin Labs' deep research agent can interrogate the full 2Q26 transcript, supplemental disclosures and historical filings on demand at marvin-labs.com.