Full-Year Guidance Withdrawn and Q3 Bookings Guided to the First Decline in Company History: the Problem Moved From Sign-Ups to Monetization

Results vs. Consensus

Q2 2026 Scorecard

A note on the widely-circulated "revenue miss." Several aggregators paired a $1.59B "revenue" consensus with a $1.56B "actual." Both figures are bookings; the $1.56B matches reported bookings of $1,557M exactly, while GAAP revenue was $1,469M. Headlines reading "revenue misses at $1.47B" are comparing a GAAP revenue print against a bookings estimate. We treat this as a bookings miss of 2%, not a revenue miss of 8%.

Year-over-Year Comparison

Quarter-over-Quarter Comparison

Income Statement Detail

Revenue and bookings assessment. Bookings of $1,557M grew 8.3% and landed $7M above the floor of the $1,550-1,610M guide given on May 1. Clearing a floor is not the same as meeting a plan, and management was explicit that it did not meet theirs. The decomposition is the important part. DAUs of 123M grew 10% and hours of 29B grew 5%, both against a 2025 comparison base inflated by that summer's viral titles, and both roughly in line with what the company expected internally. What missed was monetization: average bookings per DAU of $12.66 fell 2% year over year and 3% sequentially from $13.11. Constant-currency bookings grew 7% against the 8% reported, so a point of the growth was foreign exchange rather than user spend.

Margins assessment. The operating margin improved 1,420bps to (15.6)% and Adjusted EBITDA reached $152M against $18M, but the drivers deserve separation. Two cost lines fell as a share of revenue for reasons tied to the deferral mechanic rather than to operating discipline: developer exchange fees at 24.7% of revenue (down 460bps) and personnel excluding stock compensation at 19% (down 330bps). Measured against bookings, which is the honest denominator for a cost incurred in the period, both moved the other way: DevEx rose 130bps to 23% of bookings and personnel rose 120bps to 18%. Infrastructure and trust and safety rose on both bases, to 24.7% of revenue on the income statement and, on the company's narrower cash definition, 450bps to 15% of bookings. Adjusted EBITDA also now excludes $34M of legal settlement expense related to state youth-safety matters, $91M year to date, an add-back that did not exist a year ago.

EPS and cash assessment. The $(0.26) loss beat the $(0.33) Street by $0.07 and narrowed from $(0.41), helped by $59M of interest income on a $6.1B cash and investment balance. Free cash flow of $294M grew 66%, though the prior-year base was depressed by a $30M creator payout that slipped out of Q1 2025, which flatters the comparison. Capital expenditure was only $24M in the quarter against a Q3 guide of $170M, so the cash generation on display in H1 reflects a spending program that had not yet started. Weighted-average shares rose 4.7% year over year despite $380M of buybacks in the quarter, so the repurchase program is not yet offsetting dilution.

Key Operating Metrics

The six-quarter series is the clearest way to see what happened. Roblox's 2025 was extraordinary and the 2026 comparisons run against it; but the sequential path, which is comparison-free, has now declined for two consecutive quarters on every engagement and monetization measure.

Average bookings per DAU is computed as quarterly bookings divided by average DAUs for the quarter; the Q2 2026 figure of $12.66 matches the company's disclosure.

Assessment. Daily actives have declined for three consecutive quarters from the 151.5M peak, hours for three, and payers for two. Bookings per DAU is now the lowest of the six quarters shown other than Q1 2025. Management's framing, which is defensible, is that the 2025 peak was a viral anomaly and the platform is normalizing to a healthier baseline. The difficulty with that framing is that Q2 2026's absolute DAU count of 123M is now below Q4 2025's 144M and below Q1 2026's 132M, and the company is guiding to a sequential DAU increase in Q3 driven substantially by seasonality and by Russia coming back online rather than by the acquisition engine repairing itself.

Geographic and Cohort Performance

The regional split remains the most bifurcated part of the story, and the bifurcation widened. The developed markets where age-check penetration is highest have effectively stopped growing engagement; the emerging markets where it is lowest are still compounding at rates that would be the whole story at any other company.

US & Canada: engagement growth has effectively stopped, and it is now also the monetization problem

US and Canada DAUs grew 6% and hours grew 1%, decelerating from +17% and +21% respectively in Q1. This is the region that carried the age-check rollout first and carries the highest penetration, and it is now the region management named as the source of the monetization shortfall. Roblox attributed the weakness to a mix shift away from the high-monetizing 2025-vintage viral titles toward new and evergreen experiences that monetize less per hour, compounded by its own recommendation-algorithm change, with the impact concentrated in the under-13 cohort.

"The bookings shortfall reflects a decline in per hour monetization most notably with younger cohorts in the U.S. and Canada."
— Q2 2026 shareholder letter

Assessment: This is the single most consequential change from last quarter. In Q1 the diagnosis was that the platform's on-platform economics were healthy and only new-user acquisition was impaired, which is a repairable plumbing problem. Q2 says the users who are already there are spending less per hour in the platform's most mature and most profitable market. That is a demand-side observation, not a funnel observation, and it invalidates the "the engine is fine, only the intake is clogged" frame that supported the Hold.

Japan and India: still the offset, but decelerating from extraordinary levels

Japan DAUs grew 67% and India 64%, with hours at +61% and +59%. Both decelerated from Q1's +96% and +84%. Payers outside the US and Canada grew 23%, ahead of the 15% total. Roblox was also reinstated in Russia during June, which management characterised as a roughly three-percentage-point headwind to year-over-year DAU and hours growth because the unblocking came late in the quarter.

Assessment: The international engine is intact and remains the reason this is not a broken business. But two things temper it. The growth rates are decelerating on a base that is still small relative to US and Canada monetization, and these are the markets where average revenue per user is structurally lowest, so international DAU growth converts to bookings at a materially worse rate than the US engagement it is offsetting. Adding a Japanese or Indian daily active does not replace the bookings of a lost US under-13 spender.

The over-18 cohort: still the best part of the story, and it decelerated too

Globally, 27% of age-checked daily actives are over 18, with 35% under 13 and 38% between 13 and 17. In the US, over-18 users are approximately one third of age-checked DAUs, they grew 32% on DAUs and 27% on hours, and the 18-34 band within that grew 42% on DAUs and 37% on hours. US over-18 users monetize more than 50% higher than under-18 users.

"Today in the U.S., our 18-and-up cohort represents nearly 1/3 of our now age-checked daily active users and our U.S. 18-34 cohort daily active users and hours are growing at approximately 40% year-on-year."
— David Baszucki, CEO and Co-Founder

The comparison against last quarter is the part that did not get airtime. In Q1, the company disclosed that US over-18 DAUs and hours grew "over 40%" and the 18-34 band grew "over 50%." Those figures are now +32%/+27% and +42%/+37%.

Assessment: The aging-up thesis remains directionally validated and is the strongest structural pillar left standing. But the flagship cohort decelerated by roughly ten percentage points in a single quarter, and the company's own footnote is unusually candid that these are estimates: the demographic distribution of age-checked users is extrapolated onto non-age-checked users, and age-checked metrics are "not directly comparable to historical self-reported data." The pillar that the bull case now rests most heavily on is also the one measured with the least precision.

Content diversity: the genuinely good news

Experiences outside the top 10 grew hours 25% and Robux spend more than 20%. The top 10 accounted for roughly 20% of hours, down from about 30% three years ago. New titles including Animal Hospital, Grow a Garden 2 and Kick a Lucky Block moved from release to top experiences quickly.

Assessment: This is real and it is the argument against reading Q2 as a structural break. The multi-hit ecosystem thesis, which has been the first bull pillar since initiation, is confirmed again: the platform is producing hits without depending on any one of them. The uncomfortable corollary is that this same diversification is mechanically part of the monetization problem, because the long-tail and evergreen content that is gaining share monetizes at a lower rate per hour than the viral titles it is replacing. Content health and near-term bookings are, for now, in direct tension.

Key Topics & Management Commentary

Overall Management Tone: Considerably more forward-leaning and less diagnostic than last quarter. Where Q1 was a methodical walk through a mechanism and a fix list, Q2 spent the majority of prepared remarks on a multi-year product vision covering 2D gaming, photorealistic rendering, prompt-based creation and an integrated communications stack, with the deterioration in the current quarter's monetization handled briskly and the withdrawal of annual guidance presented as a governance improvement rather than a loss of visibility. Management was least convincing where it was most pressed: on when the retention-for-monetization trade pays back, and on whether 2026 remains a growth year, both of which drew explicit refusals to quantify.

1. The monetization break, and why it is a different problem from last quarter's

For three quarters the bear case on Roblox has been about the top of the funnel. Age checks gated chat, reduced communications compressed word of mouth, weaker word of mouth and app-store ratings reduced organic sign-ups. That mechanism was mechanical, well-articulated and fixable, and it is the reason we held rather than downgraded further in May. Q2 reports that the funnel actually improved and something else broke.

"New user sign-up activity improved throughout the quarter, benefiting from seasonality and the reinstatement of Roblox in Russia. Retention of existing users remained stable, and engagement hours were in line with our expectations. Monetization, however, as measured in bookings per hour, was below our forecast, particularly with under-13 cohorts, and was the primary factor impacting our bookings performance."
— Naveen Chopra, Chief Financial Officer

Management identified two causes of unequal character. The first is exogenous mix: 2025's viral titles monetized at exceptional rates per hour and the content replacing them does not. The second is endogenous: Roblox changed its own recommendation algorithm and the change reduced monetization more than it expected.

Assessment: The Q1 diagnosis was reassuring precisely because it isolated the damage to acquisition while certifying that on-platform economics were intact. That certification no longer holds. A company can fix a sign-up funnel with product changes on a known timetable. It cannot as easily make its existing users spend more per hour, particularly when the reason they spend less is that the content they now prefer is content the company deliberately promoted.

2. The self-inflicted half: the RFY algorithm change

In April, Roblox began deploying an updated "Recommended For You" algorithm that evaluates player behaviour over a 28-day window rather than 7 days, adding play-through rate, first-play bounce rate, spend days and Robux spent per user as signals. The intent is to surface retentive content instead of content optimized for immediate extraction. The company believes the retention gain compounds and eventually exceeds the monetization it costs.

"We do believe that based on the curves that we've seen with, call it, a few weeks of data for each of these experiments, that those lines cross, meaning the benefit of the incremental retention will more than outweigh the short-term hit to bookings. We're not going to put a specific timeline on that right now."
— Naveen Chopra, Chief Financial Officer

Assessment: We think the strategic logic is correct and would defend it in isolation. Optimizing a discovery surface for 28-day retention rather than 7-day spend is what a platform with a twenty-year horizon should do. The problem is evidentiary and it is not small: the crossover is asserted from "a few weeks of data" in A/B tests, no date is attached to it, and the company has simultaneously removed the annual guidance that would have let investors check the claim against a number. Management is asking the market to accept an unquantified future benefit in exchange for a quantified present cost, immediately after a quarter in which its own forecast of that cost was wrong.

3. Withdrawing full-year guidance

Roblox withdrew 2026 guidance entirely. The framing was that it is accelerating a previously announced transition to quarterly-only guidance by one quarter.

"In light of these changes, we are accelerating our previously announced transition to quarterly-only guidance by one quarter, issuing guidance for Q3, but not the full year. Given our long-term focus we do not believe annual guidance is a helpful tool for investors."
— Q2 2026 shareholder letter

The transition to quarterly guidance was announced at the Q4 2025 call and was to begin in 2027. Pulling it forward by one quarter has the effect of removing the annual number in the specific quarter in which that number would have had to be cut for the second time in ninety days.

Assessment: We do not dispute that annual guidance can be a poor tool for a business with this much variability, and the company said so before this quarter went wrong, which earns it some benefit of the doubt. But the timing means investors lose the anchor at exactly the moment they need it, and the stated rationale ("we do not believe annual guidance is a helpful tool") is materially harder to accept when the alternative on offer was a large downward revision. On May 1 the company defended a $7.5B annual bookings number. On July 30 it declined to defend any annual number at all.

4. The first negative bookings guide in company history

Q3 bookings are guided to $1,576-1,653M, down 14% to 18% year over year against Q3 2025's $1,922M. Roblox has never previously guided to a bookings decline. Revenue is still guided up 4-10%, purely because the 27-month deferral tail is still recognising 2025's elevated bookings.

Assessment: The comparison base genuinely is punishing, and a decline against a quarter that grew 70% is not evidence of collapse. But the guide is also roughly 14% below where the Street sat, which means the deceleration is worse than the comparison alone required. The divergence between a bookings line guided to −16% at the midpoint and a revenue line guided to +7% is the clearest possible illustration of why the reported income statement will keep looking better than the business for another two to three quarters, and why it will then look considerably worse.

5. Free cash flow goes to approximately zero

Q3 operating cash flow is guided to $110-175M and free cash flow to $(60)M-$5M, against $294M of free cash flow in Q2 and $890M in the first half. Capital expenditure steps up to $170M in the quarter from $24M in Q2, and the guide includes roughly $40M of year-over-year working capital headwind from creator payout timing.

Assessment: Cash-flow compounding has been the fourth bull pillar since initiation and the one that made the equity defensible through the de-rating. FY2025 free cash flow was $1,353M and grew 111%. On the guided path, 2026 free cash flow lands near $930M, below the withdrawn $1,050-1,275M guide and down year over year. The pillar is not broken permanently, since the capex is deliberate and the balance sheet carries $6.1B of cash and investments against $1.0B of debt, but for the next several quarters the company converts approximately none of its bookings to free cash while its bookings decline.

6. The AI investment program widens as the top line rolls over

Roblox used the call to announce or expand an unusually broad slate: Build, a prompt-based mobile game-creation tab now in public alpha in New Zealand; Moments, a short-form video tab moving to the homepage and expanding to influencer video; Roblox Reality, targeting photorealistic multiplayer, accelerated by the acquisition of Morpheus AI; 2D and single-player game support; and an integrated voice, video and text communications stack. Management sized the cost indirectly.

"If you look at our margin guidance for Q3, and you look at the margin compression relative to the year-ago period, I would say that about half of that is just fixed cost deleveraging coming off the back of the bookings reduction. The other half is related to these investments in AI."
— Naveen Chopra, Chief Financial Officer

Chopra also flagged a structural change in the character of the spend: historically infrastructure scaled with users and hours, whereas model-training costs are fixed.

Assessment: Converting infrastructure from a variable cost that flexes with engagement into a fixed cost that does not, in the same quarter that engagement and bookings turn down, is the least forgiving possible sequencing. We are not arguing the investments are wrong; Build and Moments are coherent answers to a discovery problem and Roblox Reality is a credible long-dated option. We are arguing that the company has removed its annual accountability metric, guided cash generation to zero, and widened a fixed-cost program simultaneously, and that the combination leaves shareholders with no near-term instrument for checking whether any of it is working.

7. Age check reaches 57%, and the communications rebuild is still incomplete

Age-check penetration reached 57% of global daily actives, with the US and Australia above 70% against a 90% long-term goal, and under-18 penetration at 60% globally and 75% in the US. Roblox Kids and Roblox Select launched in June with engagement and bookings impact in line with expectations, and the catalogue available to younger users has grown to roughly 30,000 games, up nearly 50% since launch. Global Chat reached an average of 30 million daily messages in June and Quick Words is running above five million per day.

"We have started to see improvement in on-platform communications, though we still have more work to do to drive on-platform communications above pre-age check levels."
— Q2 2026 shareholder letter

Assessment: This is the cleanest execution in the quarter and the strongest rebuttal to the view that safety compliance is an open-ended drag. The rollout is on schedule, the Kids and Select launch did not repeat the Q1 disruption, and the communications products are shipping. The candour that on-platform communications remain below pre-age-check levels is welcome and also confirms the Q1 word-of-mouth mechanism has not yet fully reversed. Safety is no longer the binding constraint on this thesis. Monetization is.

8. Capital allocation: a buyback that does not yet offset dilution

Roblox repurchased 8.2 million shares for approximately $380M in the quarter under the $3B authorisation announced on May 19, of which $1B is intended over twelve months. Fully diluted shares stood at 752 million at June 30, up 2% year over year, and weighted-average shares used for EPS rose 4.7%.

Q: "What impact, if any, does the stock performance have on your willingness to maybe lean in further on the buyback relative to alternative uses of capital?"
— Chris Scholl, UBS

A: "The way we have structured our buyback, which is largely to offset dilution from employee equity grants, does result in us effectively buying more stock when the stock price goes down and buying less when the stock price goes up. There is some sort of natural adjustment built into the mechanism. We do have the ability to accelerate that or supplement it."
— Naveen Chopra, Chief Financial Officer

Assessment: The programme is explicitly framed as dilution management rather than value capture, and on the current run rate it is not achieving even that: share count rose despite $380M of repurchases. With $5.1B of net cash and the stock at a two-year low, the "ability to accelerate" is real optionality that management notably declined to commit to. We would treat any material acceleration as a genuine signal; the current pace is not one.

9. A new non-GAAP add-back on the company's core regulatory exposure

Adjusted EBITDA now excludes legal settlement expenses of $34M in the quarter and $91M year to date, described as relating to settlements and settlement negotiations with certain states regarding youth-related consumer protection and digital safety matters. The company argues these arise outside the ordinary course, have limited precedent and are unpredictable.

Assessment: Without this add-back, Q2 Adjusted EBITDA would be $118M rather than $152M, and the first-half figure $160M rather than $251M. We are uneasy about the characterisation. Youth safety is not a peripheral legal matter for Roblox; it is the central regulatory question the entire business faces, it drove the product changes that caused the last two quarters of disruption, and coverage notes an anticipated EU Very Large Online Platform designation alongside a securities class action with an August 7 lead-plaintiff deadline. Costs arising from the defining risk of a business are ordinary course by any reasonable reading, and excluding them makes the trend in adjusted profitability look better than it is at the exact moment that trend matters.

Guidance & Outlook

What was withdrawn

Implied full-year bookings. First-half bookings were $3,288M. At the Q3 guide midpoint of $1,614M, nine-month bookings reach approximately $4,903M. Against 2025's four reported quarters of $1,207M, $1,438M, $1,922M and $2,222M, which sum to $6,788M, the fourth quarter is the swing factor. If Q4 declines at the same rate as the Q3 guide midpoint, full-year bookings land at approximately $6,769M, or roughly 0.3% below 2025. Across the full Q3 guide range applied to Q4, the band is approximately $6,725M to $6,814M, or −0.9% to +0.4%.

Implied quarter-over-quarter ramp. Achieving even flat full-year bookings requires Q4 to come in near $1,886M, a 15% decline against Q4 2025. Achieving the withdrawn +8% guide floor would require Q4 bookings of roughly $2,428M, up 9% year over year, one quarter after a guided 16% decline. That gap is why the annual guide could not survive the quarter.

Street at. Consensus was near $1.87B for Q3 bookings against a $1,614M midpoint, a 14% gap. Estimates will now be rebuilt without an annual company anchor for the first time, which we expect to widen dispersion and keep the shares volatile into the Q3 print.

Guidance style. Historically conservative and reliably beaten through 2025, when the company raised the full-year bookings guide by roughly $600M in Q2 and again by roughly $671M in Q3. That pattern inverted in 2026. Q2 bookings landed $7M above a floor the company set ninety days earlier and below its own internal plan, and management now describes Q4 as having a range of outcomes too wide to guide. We would treat the Q3 range as genuinely uncertain rather than sandbagged, because the company has just demonstrated that it does not currently forecast its own monetization accurately.

Analyst Q&A Highlights

What investors should watch to know the engagement headwinds have peaked

The opening question asked for observable signposts that would mark the trough, and separately whether the guided sequential DAU increase in Q3 reflects genuine improvement or merely seasonality. The answer separated the two cleanly and, in doing so, corrected the premise of the question in a way that reframed the whole quarter.

Q: "I guess between here and there, what are you watching and what should investors be watching to help determine when the engagement headwinds from those changes have peaked? Connected to that, when we think about the expectation of user growth sequentially in the Q3, is that just a function of seasonality, or are there any improvements in the trajectory as you've exited the Q2?"
— Matthew Cost, Morgan Stanley

A: "I just would clarify, you talked about near-term engagement friction. That's really not what we're seeing. We're actually pretty happy with the engagement trends that we've seen... The weakness that we saw that was inconsistent with what we expected when we put our original plans together for the quarter was really around monetization. That is going to continue as we continue to evolve some of what we're doing around discovery and recommendations. The DAU trends themselves, seasonality is definitely a big factor there. We're going to get a full quarter benefit from having Russia back online."
— Naveen Chopra, Chief Financial Officer

Assessment: The correction is the most important exchange on the call. Management explicitly declined the engagement framing and relocated the problem to monetization, which it then said "is going to continue." It also conceded that the guided sequential DAU improvement rests on seasonality and a geographic reinstatement rather than on the acquisition engine recovering. Both halves of the answer are more negative than the question assumed.

When the retention-for-monetization trade pays back

A recurring line of questioning pressed for the timeline and evidence behind the claim that longer retention eventually outweighs lower hourly monetization. Management described the shape of the curves in detail while declining to date the crossover.

Q: "In the shareholder letter you mentioned, you were comfortable based on internal testing that longer retention would ultimately overcome a reduction in hourly monetization. Could you elaborate a little bit more on that or maybe what you had seen when you had run those sort of comparisons?"
— Clark Lampen, BTIG

A: "When we sort of A/B test them relative to people on, call it, older discovery algorithm, is that very quickly there is a change in user retention... However, there is a pretty immediate hit to dollars per hour... We do believe that based on the curves that we've seen with, call it, a few weeks of data for each of these experiments, that those lines cross... We're not going to put a specific timeline on that right now, in part because we're continuing to evolve those algorithms."
— Naveen Chopra, Chief Financial Officer

Assessment: The asymmetry in the answer is the tell. The cost is described as immediate and observed; the benefit is described as a crossover inferred from a few weeks of experimental data with no date attached. Management also disclosed that its current focus is on reducing the near-term bookings impact while keeping the retention benefit, which is an acknowledgment that the trade as currently implemented is more expensive than intended.

Whether 2026 remains a growth year

With the annual guide withdrawn, the most direct question of the call sought any guardrail at all for the fourth quarter, and asked for a simple commitment that full-year bookings would exceed 2025. Management declined to provide either.

Q: "Number one, is 2026 still a growth year? Would you be comfortable telling investors that 2026 bookings will be higher than 2025 bookings in total? ... What's getting better in the Q4 versus the Q3?"
— Omar Dessouky, Bank of America

A: "We're not going to provide guidance for the Q4 right now simply because of all of the moving pieces... The reality is, it's very early days for many of those new initiatives. There's no model that we can look at that's going to say here's exactly what they're going to do over the next three to six months... we can't give you any more specific guidance than what I've shared."
— Naveen Chopra, Chief Financial Officer

Assessment: A company confident that its full year will grow says so when asked this plainly, because the statement costs nothing if true. The refusal is consistent with our arithmetic, which puts full-year bookings within roughly a point of flat in either direction. This exchange, more than the guide itself, is what converted a bad print into a 27% decline.

Why the under-13 cohort absorbed a disproportionate share of the damage

Questioning probed whether the concentration of monetization weakness among the youngest users reflects something structural about how that cohort behaves, or is simply an artifact of which titles happened to be popular a year ago. Management argued firmly for the latter.

Q: "Can you please talk a bit about the differences in monetization patterns between under 13 players and other cohorts that may have driven an outsized impact this quarter? Was it specifically related to the viral hits from last year, or is it a general rule that these younger users are maybe more volatile in their monetization patterns?"
— Andrew Marok, Raymond James

A: "I think it's more the former... those games did appeal in particular to younger users. So when we have seen this mix shift from those games getting replaced, if I call it games with more normal levels of monetization, that shows up most significantly with these younger users. I don't think there is anything sort of structurally with respect to changes in younger user behavior."
— Naveen Chopra, Chief Financial Officer

Assessment: If correct, this is the most reassuring thing said on the call, because a mix effect anniversaries out while a behavioural shift does not. The claim is plausible and consistent with the 2025 hit profile. It is weakened by the fact that the company's own algorithm change is the second driver of the same weakness in the same cohort, and that change does not anniversary out; it persists until the crossover management would not date.

The shape of the cost structure through the transition

Questioning sought to understand how the financial profile changes over the next twelve to twenty-four months given the simultaneous push into over-18 content, AI tooling and higher creator payouts. The answer separated the long-term margin algorithm, which management said is unchanged, from a near-term step-up in fixed infrastructure.

Q: "Could you talk to us about any changes you might expect in the cost structure? How either from either a DevEx standpoint, from a tech standpoint, a platform standpoint, how might the financial profile of the business look in success 12-24 months out relative to the prior 12-18 months?"
— Ken Gawrelski, Wells Fargo

A: "Our view of the long-term margin potential of the business really has not changed... in call it the near to medium term, the investments that we're making in a lot of AI-powered product enhancements, I do expect to result in some elevated levels of infrastructure expense. It's important to realize that unlike our infra expense historically, this is more of a fixed cost, because historically, infrastructure investments have been a function of hours and users on our platform. These infrastructure costs are largely being driven by model training."
— Naveen Chopra, Chief Financial Officer

Assessment: The disclosure that infrastructure is converting from variable to fixed is the most consequential financial detail volunteered on the call and received little attention. It means the operating leverage that has defined the model works in both directions now: the cost base no longer contracts when engagement does. Every margin scenario from here is more sensitive to bookings than the historical relationship implies.

Competition for time and attention beyond gaming

Questioning widened the frame to ask how Roblox positions against both the gaming landscape and the broader contest for young users' time, including the online-versus-offline debate. The response did not engage the competitive question and instead restated the platform vision.

Q: "How do you think about the competitive landscape continuing to evolve, both within gaming, but also among younger users where there's sort of conflicting narratives about being online versus offline and gaming versus non-gaming in terms of time spent?"
— Eric Sheridan, Goldman Sachs

A: "I really think the broader definition is a bigger thing than Roblox, and it is our vision that the world needs more play... I do think what is gaming is changing. I think we're going to enter an era just like 10 years ago, where most of us didn't know how to edit video, and now all young people, essentially, many of them know how to edit video on their phone."
— David Baszucki, CEO and Co-Founder

Assessment: A question about competitive positioning was answered with a thesis about category expansion. No competitor, no share data and no discussion of the specific title risk facing the platform this year. The vision is coherent on its own terms, but the exchange left the actual question unaddressed, and it is a question that matters more in a quarter when hours per user declined.

The economics of the photorealism program

Questioning sought to size the financial commitment behind the newly acquired real-time video-model technology and the associated infrastructure headwind. The response reframed to the aggregate AI program and, separately, disclosed the intended monetization model.

Q: "Just on Morpheus. Could you help us think about either the magnitude of financial commitment or potentially help size the IPS headwind that we might see this year?"
— Clark Lampen, BTIG

A: "The cost of infra here will initially be offset by subscription. Games will run both in regular Roblox mode as well as in super upsampled Morpheus-Video photo reality mode, we will initially expect to charge subscription to access that."
— David Baszucki, CEO and Co-Founder

Assessment: The specific sizing was not given; the answer redirected to the roughly half-of-Q3-margin-compression figure covering all AI investment. The genuinely new information is that Roblox intends to charge a subscription for photorealistic mode, which is the first concrete monetization mechanism attached to the program and a modest offset to the fixed-cost concern. It remains an unshipped product with unproven demand at an unstated price.

What They're NOT Saying

1. Whether 2026 is a growth year: asked in the plainest possible terms and declined. A company expecting growth confirms it, because the confirmation is costless if true. Our arithmetic puts full-year bookings within roughly a point of flat.
2. When the retention-monetization crossover occurs: explicitly refused. The entire justification for accepting lower bookings rests on a payback with no date, inferred from a few weeks of A/B data.
3. Any quantification of the Morpheus AI purchase: no price, no earn-out, no expected contribution. Cash used for business combinations was $6M in the quarter and goodwill rose $20M, so either the transaction is small or it closed after the quarter end. Neither was clarified.
4. App-store ratings and organic sign-up conversion: the specific metric chain management itself introduced in Q1 as the cause of the guide cut. Q2 says sign-up activity "improved throughout the quarter" but discloses no rating or conversion data, so the fix cannot be verified.
5. Bookings per hour in absolute terms: repeatedly described as below forecast, never quantified. Average bookings per DAU is disclosed; the per-hour metric management says was the primary driver is not.
6. Q4 2026 in any form: not even directional guardrails, against a Q4 2025 base of $2,222M that is the hardest comparison in the company's history.
7. Whether the buyback will be accelerated: management confirmed it has the ability and declined to commit, with the stock at a two-year low and $5.1B of net cash on the balance sheet.
8. The comparability of age-checked cohort data: disclosed only in a footnote. Over-18 growth and monetization premia are extrapolated from age-checked users onto non-age-checked users and are "not directly comparable to historical self-reported data," which materially qualifies the pillar the bull case now leans on hardest.
9. The scale of legal and regulatory exposure: $91M of settlement expense year to date is added back to Adjusted EBITDA with no disclosure of the number of states, the range of possible outcomes, or the expected 2026 total.

Market Reaction

• Pre-print setup: closed at $48.67 on July 30 entering the after-the-close release. Down 39.9% year to date against the S&P 500's +8.7%, down 61.0% over twelve months, and down 10.5% over the trailing thirty days. The 52-week closing range entering the print was $41.31 to $141.56. Options were pricing an implied move of roughly 14%.
• After-hours, July 30: approximately −14% on the initial release, deepening through the call.
• Next-day session, July 31: opened at $38.88, a 20.1% gap down, traded a range of $33.88 to $40.00, and closed at $35.60, down 26.9% or $13.07. The intraday low marked a 30.4% decline.
• Volume: 60.9 million shares against a 9.4 million thirty-day average, 6.5 times normal.
• Context: the S&P 500 rose 0.7% on the session. The close was below the prior 52-week closing low of $41.31.

The realized move was roughly 1.9 times the options-implied 14%, which tells you the market was positioned for a disappointing quarter and not for the withdrawal of the annual guide. The path matters as much as the magnitude. The press release, the Q3 numbers and the guidance withdrawal were all public before the 4:30 PM ET call, and the initial reaction was approximately −14%, roughly in line with what options had priced. The move then roughly doubled through the call and the following morning. What was added in between was not data. It was management declining to say whether 2026 would grow, declining to date the monetization crossover, and declining to offer any Q4 framework, followed by a wave of sell-side downgrades at the open.

The mechanical driver of the severity is the loss of an anchor. Roblox is a stock that trades on bookings, and the market has just lost both its forward number and its confidence in the company's ability to forecast that number, in the same release in which the near-term direction turned negative. When a growth multiple's denominator becomes unknowable, the multiple compresses to whatever the most pessimistic credible scenario supports. That is what a 27% single-session decline in a $27B company looks like.

Street Perspective

Debate: Is this normalization from a viral peak, or the beginning of platform lifecycle decline?

Bull view: 2025 was an anomaly driven by a handful of titles that monetized at exceptional rates, and the comparisons are simply brutal. Daily actives still grew 10%, hours grew 5%, payers grew 15%, the long tail grew hours 25%, and the top ten now account for 20% of hours rather than 30%. A platform producing new hits with less concentration is healthier, not sicker. The comparison base normalizes from Q2 2027 and growth resumes.

Bear view: Daily actives have fallen for three consecutive quarters, hours per user are declining, payers are down 26% from the Q4 2025 peak, and the company has now guided to an outright bookings decline while withdrawing its annual forecast. The bear camp reads the diversification favourably for content health and unfavourably for economics, because the content gaining share monetizes worse. At least one desk has explicitly raised the possibility that the platform is entering lifecycle decline.

Our take: The bull view is closer to right on the underlying platform and the bear view is closer to right on the next four quarters. The content diversification is genuine and the international engine is real. But normalization and decline are indistinguishable in real time, and the company has removed the disclosure that would let an investor tell them apart. We do not think this is lifecycle decline; we think it is at least three more quarters before anyone can prove that it is not.

Debate: Was withdrawing full-year guidance good governance or lost control?

Bull view: the transition to quarterly-only guidance was announced at the Q4 2025 call, before any of this went wrong, and pulling it forward one quarter is a minor schedule change. Annual guidance genuinely is a poor instrument for a business with this much variability, and forcing management to defend a twelve-month number encourages exactly the short-term monetization optimization the discovery algorithm change is designed to unwind.

Bear view: the timing is not incidental. The annual number disappeared in the specific quarter it would have had to be cut for the second time in ninety days, after being cut fourteen points in May. Management then declined to provide even a directional Q4 guardrail or confirm that the year would grow. The Street lost its anchor at the moment of maximum uncertainty and must now rebuild estimates with no company reference point.

Our take: the bear reading is the more honest one, even though the bull's premise is factually correct. The prior announcement earns management some benefit of the doubt on intent, and we would have supported the change in a stable quarter. Executed in this one, and paired with the refusal to answer the growth-year question, it functions as a reduction in accountability rather than an improvement in disclosure quality. The market priced it that way and we think the market was right.

Debate: At $35.60, is the damage priced?

Bull view: the stock is down 61% over twelve months and roughly 75% from its 52-week closing high, at a two-year low, with $5.1B of net cash, an authorized $3B buyback, roughly 3.2 times enterprise value to our full-year bookings estimate, and free cash flow that is depressed by deliberate investment rather than by operating failure. Sentiment is capitulatory, five downgrades landed on the same morning, and the comparison base turns favourable within twelve months.

Bear view: a $26.8B market capitalization on approximately $6.8B of flat bookings, roughly breakeven free cash flow in the second half, and no annual guidance is not a value proposition. The next two prints show year-over-year bookings declines. Reported revenue growth, currently the most flattering line, decays mechanically from +36% toward zero as the deferral tail unwinds, so the income statement gets worse from here even if bookings stabilize. There is no catalyst before the Q3 print in late October and no annual anchor to re-rate against.

Our take: the bear view has the better of it over our twelve-month horizon, which is the horizon the rating addresses. We accept that the shares are cheap against a 2027 recovery and that the balance sheet removes any solvency question. But the de-rating is not obviously complete: the multiple is still a growth multiple, applied to bookings that are currently shrinking, with the crossover that would restore growth explicitly undated by management. The asymmetry that supported a Hold at $45 has not improved at $35.60, because the denominator fell further than the price did.

Model Update Needed

Valuation framework. On the 752 million fully diluted shares disclosed at June 30, the $35.60 close equates to a market capitalization of approximately $26.8B. Against $6.1B of cash and investments and $1.0B of long-term debt, enterprise value is approximately $21.7B, or 3.2 times our $6.77B full-year 2026 bookings estimate and 3.0 times our $7.3B 2027 estimate. On the market-capitalization basis used in prior notes, the stock trades at approximately 4.0 times 2026 bookings, against 4.5 times at the $45.13 level following the Q1 print. The multiple has compressed less than the price because the bookings estimate fell alongside it.

Valuation impact. Our base case of $29-34 implies 18% downside to 4% downside from $35.60. The bull case of $46-53 implies 29% to 49% upside and requires visible evidence of the monetization crossover in the Q3 or Q4 prints together with a 2027 framework above 15% growth. The bear case of $20-25 implies 44% to 30% downside and requires 2027 bookings to be flat again with free cash flow near zero. The distribution is wide, which is the point: with no annual guidance and an undated crossover, the range of defensible outcomes has widened materially, and the midpoint sits below the current price.

Thesis Scorecard Post-Earnings

Against the triggers we set in May

Overall: Thesis weakened. Three of the four bull pillars remain intact in substance, and the safety bear point genuinely improved, which is why this is not a capitulation call on the business. But the pillar that funded patience, cash-flow compounding, is guided to approximately zero for a quarter; the problem migrated from a repairable acquisition-funnel issue to a demand-side monetization issue that management will not date; and the annual disclosure that would let an investor track either has been withdrawn. The explicit downgrade trigger we published in May fired.

Action: Downgrade to Underperform from Hold. Existing holders should reduce into strength rather than sell into the gap, recognizing that a 27% single-session decline to a two-year low frequently produces a technical bounce that offers better exit levels than the print itself. New capital should wait: the earliest evidence that would change this view is the Q3 print in late October, which needs to show sequential bookings improvement that is not attributable to seasonality, a quantified crossover in the discovery trade, and ideally the restoration of some annual framework. We would revisit toward Hold on any two of those, and toward Outperform on all three plus a 2027 framework above 15%. Below roughly $25 the risk/reward inverts regardless, because at that level the market would be pricing a permanent impairment that the platform metrics do not support.