Thesis: Magnite's Q2 print converted the CTV mix shift into cash — revenue +11.2% YoY, operating margin 16.2%, $186.6M operating cash in the quarter — but a +20% gap already paid for it; the setup is the retest of the pre-print shelf, not the chase.

Attribution note: the qualitative recovery framing in circulation comes from Seeking Alpha opinion pieces by two named contributors (Gytis Zizys, David Zanoni) — attributed, not independently verified. Every number below is from the SEC filings; where a claim rests only on that coverage, it says so.

The Story Right Now

Magnite filed its Q2 2026 10-Q and an Item 2.02 results 8-K on 2026-08-05, after the session. The tape gapped the next morning: from a 2026-08-05 settled close of $20.67 to $24.92 on the 2026-08-06 intraday snapshot, +20.6% on the day, on 2.2x average volume. The 2026-08-05 settled row had the name already trending strong-up — 50-day above the 200-day, +36.1% over the 200-day, RSI 62.1, still 22.4% under its 52-week high. One session later the same name reads RSI 77.8 and 6.5% under that high. The move was one print wide.

The print deserved most of it. Revenue was $192.823M for the June quarter against $173.332M a year earlier — +11.2% YoY, the first double-digit growth quarter since Q3 2025 and a sharp reversal of the +5.5% Q1 posted three months earlier. What matters more is what fell through: operating income $31.222M on 16.2% operating margin, against $21.959M and 12.7% a year ago, on a cost of revenue that fell year over year ($62.038M vs $64.953M) while revenue rose. That is the mix shift working — CTV and higher-take-rate biddable inventory replacing lower-margin open-web volume — and it is the first quarter where it is legible in the operating line rather than in a non-GAAP bridge. Diluted EPS $0.13 against $0.08.

The cash line is the part the recovery essays never had. First-half operating cash flow was $65.862M against $21.089M in the first half of 2025; because Q1 2026 burned $120.766M on the seasonal working-capital reversal, the June quarter alone generated roughly $186.6M of operating cash. Cash and equivalents went from $184.648M at March 31 to $332.615M at June 30. Against $350.400M of long-term debt, that is a net debt position of about $17.8M — effectively a clean balance sheet for a company that carried a 20% operating margin quarter.

The reason to be careful is on the other side of the tape. Two things now sit inside the price: an 11% growth rate re-rated to roughly 4.8x trailing sales and 21x trailing earnings at $24.92, versus 4.0x and 17.7x at the prior settled close; and the entire Google-remedy optionality, which is not decided and will not be for a long time. Meanwhile the ad-tech cohort is splitting violently — Trade Desk sits 80.6% under its 52-week high and Applovin fell 19.4% in the same session Magnite rose 20.6%. Sell-side infrastructure is being repriced up while demand-side and open-web performance names are being repriced down. That divergence is either the market correctly identifying where the take rate is defensible, or a rotation that reverses on the next cohort print. Magnite is on the right side of it today; being on the right side of a two-day rotation is not a thesis.

What the company is doing, and says it will do

Magnite describes itself in its 10-K as "the world's largest independent omni-channel sell-side advertising platform" and "the largest independent programmatic CTV marketplace." Revenue is earned as a percentage of the ad spend running through the platform, with fixed-CPM arrangements for ad serving and full-spend recognition on insertion-order campaigns — which is why the top line is a take rate, not gross billings, and why accounts receivable stood at $1.384B on June 30 against a $192.8M revenue quarter. Gross advertising spend transits the balance sheet; only the fee reaches the income statement. The company reports Contribution ex-TAC as its own non-GAAP top-line measure alongside GAAP revenue.

The stated growth vector is unambiguous: the filing says Magnite believes "CTV will be a significant driver of our revenue growth for the foreseeable future," and it names the mechanism — a shift from reserve/programmatic-guaranteed toward biddable transactions, which the company says "carry a higher take-rate compared to reserve auctions." The product build follows that: a next-generation SpringServe CTV platform announced April 2025 combining the streaming SSP and ad server into one interface, and ClearLine on the buy side giving buyers direct access to premium CTV inventory with curation. Polygon's company description puts the channel mix at 43% CTV, 40% mobile applications, remainder web — a vendor figure, not a filed one, and it should be read as approximate.

The filed risk factors name the two structural threats better than any outside bear does. On AI: "AI is changing the way in which users access information and content on the open internet, in particular with respect to search referral traffic, which has and is expected to continue to decline," creating challenges for the open-web publishers whose inventory Magnite monetizes, and potentially "reallocating budgets to new formats and approaches, including walled-gardens that we cannot access." On concentration: because of consolidation among CTV sellers, "the loss of a CTV client may result in a significant decrease in the amount of CTV inventory available through our platform."

And one filed fact that reframes the Google catalyst entirely: on 2025-09-16 Magnite filed its own suit against Google LLC in the Eastern District of Virginia seeking damages, following the court's ruling that Google engaged in unlawful anticompetitive practices in ad tech. The 10-K is blunt about the two-sidedness — Google is "both a major partner and competitor," "a significant portion of our revenue is generated through our relationship with Google," the action is "in its early stages," damages "may not be commensurate with our expectations," and there is explicit exposure to "retaliatory actions by Google." Magnite is not a bystander to the Google remedy; it is a claimant with counterparty risk.

What the street and socials say

The available coverage is stale relative to the event. All five items in the current news pull are Seeking Alpha pieces built on the Q1 2026 print — the Q1 earnings-call transcript summary (non-GAAP EPS $0.13 beating by $0.02, revenue $164.4M beating by $5.34M, 2026 adjusted EBITDA margin target raised to at least 35.5%, mid-30% free-cash-flow growth projected) plus four opinion pieces from two contributors arguing the CTV-momentum and post-drawdown recovery case. The Q1 revenue figure checks out against the filing ($164.371M). The EBITDA-margin target and FCF-growth projection are that coverage's characterization of the call, not a figure verified against a filed document here. Nothing in the pull covers the Q2 print that actually moved the stock, and the item dates are inferred rather than published.

The social layer is empty of signal. The sweep returned five YouTube items and nothing from X — retail explainer channels, a CEO interview, a "does it have upside" segment — scored on view counts in the hundreds. There is no positioning read here, and none should be invented from it. What positioning evidence exists is mechanical: short interest of 13.606M shares at the 2026-07-15 settlement, 5.35 days to cover, down from 14.713M two weeks earlier; and a run of Form 4 and Rule 144 filings on 2026-07-07, 07-09, 07-10, 07-14, 07-15 and 07-17, consistent with the leading-indicator read of zero insider buys against sustained insider selling worth roughly $52.1M net. Whether those sales are 10b5-1 programmatic is not established from what is on hand.

Setup

The arithmetic is the whole argument. Buying the middle of the zone at $20.50 risks 12.7% to the stop for 29.3% to the target — 2.3:1. Chasing at $24.92 risks 28.2% for 6.3% — 0.22:1. Same thesis, same levels, and a ten-fold difference in what the trade is worth, decided entirely by whether you pay for the gap. RSI at 77.8 on 2.2x volume is not an entry; it is a print being absorbed. The desk's position is that this name is worth owning below $21.50 and worth nothing above $24 until the next quarter re-anchors the multiple.

Bull case

1. The margin story is now on the operating line, not in a bridge. Cost of revenue fell year over year ($62.038M vs $64.953M) while revenue rose 11.2% — gross margin 67.8% against 62.5%, operating margin 16.2% against 12.7%. That is the CTV/biddable mix doing what the filing said it would do, and it compounds if the mix keeps shifting.
2. Cash generation is no longer seasonal noise. First-half operating cash flow tripled year over year ($65.862M vs $21.089M), and the June quarter alone threw off roughly $186.6M. FY2025 free cash flow was $165.6M on $236.2M operating cash less $70.5M capex. A ~$3.6B company generating that, with net debt near zero, has real optionality on buybacks.
3. Independent SSP scarcity is increasing. Integral Ad Science left the public market in a $1.9B all-cash take-private by Novacap completed 2025-12-23. Every independent ad-tech asset that goes private raises the strategic floor under the ones that remain, and Magnite is the largest independent CTV marketplace by its own filed claim.
4. Magnite holds a claim, not just an exposure, in the Google case. Its own damages suit was filed 2025-09-16 in E.D. Va. following the liability ruling. Any recovery is upside that no forward multiple currently carries.
5. The cohort is repricing sell-side infrastructure up. On the 2026-08-06 intraday snapshot Magnite is +20.6% and PubMatic flat, while Trade Desk is −6.2%, Applovin −19.4%, and Criteo and Taboola sit in downtrends. If the market is sorting on defensible take rate, Magnite is on the correct side.

Bear case

1. Eleven percent is not a growth multiple's worth of growth. Trailing-twelve-month revenue is $742.0M; at $24.92 that is 4.8x sales and 21.4x trailing earnings, against 4.0x and 17.7x at the prior settled close. The re-rate happened in one session, ahead of any second confirming quarter.
2. The prior-year comparison flatters. Q1 2026 grew 5.5% and Q1 2025 posted a −0.9% operating margin. Half of the year-over-year expansion is a weak base, not new capability; the two-quarter run rate (first-half revenue +8.5%) is the honest number.
3. Trailing earnings are contaminated by a tax item. Q4 2025 net income of $123.1M included an income-tax benefit of $75.0M — the trailing P/E is not an operating multiple. Strip it and the earnings-based valuation is materially worse.
4. Insiders are selling into it. Zero buys against sustained selling worth roughly $52.1M net, with Form 4 and Rule 144 filings clustered through July. That is not proof of anything, but it is not the pattern that accompanies management conviction in a re-rate.
5. The open-web half of the business is structurally shrinking. The company's own filing says search referral traffic "has and is expected to continue to decline," which erodes the publisher base underneath the non-CTV inventory. The CTV growth has to outrun a shrinking denominator indefinitely, not just for a quarter.

Catalysts

Risks

• Open-web secular decline. The non-CTV inventory sits on publishers whose economics are deteriorating. This is not cyclical and does not reverse with the ad cycle.
• AI answer engines removing the referral layer. Magnite's own filed risk factor states search referral traffic "has and is expected to continue to decline" and warns of budgets reallocating "to new formats and approaches, including walled-gardens that we cannot access." An SSP monetizes page views it does not control; the AI interface removes the page view.
• Ad-cycle sensitivity. The filing's first-listed risk is that results "are highly dependent on the overall demand for advertising." A 16.2% operating margin on a take-rate model is high-beta to spend in both directions — Q1's 4.7% operating margin, three months earlier, is the same business in a weaker quarter.
• The remedy going nowhere. Behavioral-only relief, blunted further by years of appeals, is the base case implied by the judge's stated lean. Any part of the current multiple that prices structural separation of AdX is paying for an outcome the court has signalled against.
• CTV seller concentration. The filing names it directly: consolidation among CTV sellers means one client loss removes a disproportionate share of inventory.
• The gap itself. RSI 77.8 on a one-session 20.6% move with no second confirming quarter is a structure that fills more often than it extends.

Financials

All figures GAAP, USD, quarterly periods as labelled. Q2 2026 = the three months ended 2026-06-30, from the 10-Q filed 2026-08-05. Revenue is Magnite's take-rate top line (a percentage of ad spend, with fixed CPM for ad serving and full-spend recognition on insertion-order campaigns), not gross advertising spend — the company reports Contribution ex-TAC as its own non-GAAP top line alongside it, and gross spend transits the balance sheet as receivables ($1.384B at 2026-06-30 against a $192.8M revenue quarter).

Growth and margin percentages computed from the filed dollar figures; not restated from any provider's own ratio field.

Basis notes. Q4 2024 and Q4 2025 rows come from the Massive income-statement pull because the SEC company-facts set carries fourth quarters only inside the annual figure. Q4 2025 net income of $123.050M includes a $74.976M income-tax benefit (Massive), so it is not an operating result — trailing earnings built on it overstate run-rate profitability. Massive's series ends at Q1 2026 and does not contain Q2 2026 at all; the Q2 row is SEC XBRL only. Massive's ratios endpoint returned 403 (premium tier), so no provider ratio is used anywhere above.

Full-year 2025 (10-K filed 2026-02-25): revenue $713.953M, operating income $97.601M, diluted EPS $0.95, operating cash flow $236.168M, capex $70.535M — free cash flow $165.633M.

Valuation, both bases. On 143.215M shares outstanding: at the 2026-08-05 settled close of $20.67, market cap $2.96B — 4.0x trailing-twelve-month revenue of $742.0M and 17.7x trailing net income of $166.9M. On the 2026-08-06 intraday snapshot of $24.92, market cap $3.57B — 4.8x and 21.4x. Trailing net income includes the Q4 2025 tax benefit; read the sales multiple as the cleaner one.

Not on file. The Q2 earnings release exhibit (8-K EX-99.1, filed 2026-08-05) is not held here, so management's Q2 commentary, updated adjusted-EBITDA guidance, and any Contribution ex-TAC figure for the quarter are unavailable. The channel split between CTV, mobile and web for Q2 2026 is likewise not held — the 43/40/remainder figures cited above are a vendor description, not a filed disaggregation.

Cross-references

• Ad-tech watchlist, opened 2026-08-06 (MGNI, TTD, APP, PUBM, CRTO, DV, ZETA, ROKU, MNTN, TBLA). Magnite is the reference name for the independent-SSP leg. The 2026-08-06 intraday snapshot shows the cohort split cleanly: Magnite +20.6%, ROKU and PubMatic and Zeta holding uptrends, against Trade Desk (−80.6% from its 52-week high), Applovin (−19.4% on the session), Criteo and Taboola in downtrends.
• Market-wide monster screen, 2026-07-30 sweep — Magnite was added to the discovery list in that batch alongside 53 other names, a week before the print that re-rated it. The screener flagged the base; the print supplied the reason.
• IAS / Novacap take-private — Integral Ad Science completed its $1.9B all-cash acquisition by Novacap on 2025-12-23 at $10.30/share and delisted. It is deliberately excluded from the ad-tech watchlist and is used here only as the sector's most recent take-private comp: the verification leg of ad tech cleared at that price, which is the closest thing available to a private-market mark on independent ad-tech assets.
• Google ad-tech antitrust — the remedies phase is a live sector-wide catalyst across the whole independent SSP/DSP cohort, not a single-name event; PubMatic and Trade Desk carry the same directional exposure with different signs.

Sources

Primary

• MGNI SEC EDGAR filing index (CIK 0001595974)
• 10-Q for the quarter ended 2026-06-30, filed 2026-08-05 (accession 0001628280-26-053343)
• 8-K Item 2.02 results release, filed 2026-08-05 (accession 0001628280-26-053334)
• 10-K for FY2025, filed 2026-02-25 (accession 0001595974-26-000007) — Item 1 Business and Item 1A Risk Factors quoted above
• Norton Rose Fulbright — closing arguments in US v. Google (ad tech)
• AdExchanger — DOJ and Google sharpen their remedy proposals
• National Law Review — judge leans away from breakup during closing arguments
• The Current — judge signals on structural remedies
• Digiday — Google's ad-tech empire faces its moment of truth
• PR Newswire — IAS announces completion of acquisition by Novacap (2025-12-23)
• Seeking Alpha — MGNI coverage — the five items characterized above (Q1 2026 earnings-call transcript; "CTV Momentum Should Continue To Translate Into Higher Value" and "Seems Like The Worst Has Been Priced In (Rating Upgrade)" by Gytis Zizys; "Great Start To 2026 Should Drive The Stock's Recovery" and "Magnite Stock Looks Poised For A Strong Recovery In 2026" by David Zanoni). The pull returned only this symbol-level URL for all five; no per-article URLs are held, and the item dates are inferred rather than published.

Repo receipts

• company-facts, key-financials.json, submissions.json — SEC XBRL company facts, fetched 2026-08-06
• income-statements, balance-sheets.json, cash-flow-statements.json, short-interest.json, sec-filings.json, company.json
• 10k-business, 10k-risk_factors.json — Item 1 and Item 1A text, 10-K filed 2026-02-25
• latest — fetched 2026-08-06T15:35Z, 5 items, ok
• 2026-08-06-social-search-mgni-magnite — social sweep, 2026-08-06
• monster-discoveries — 2026-08-05 settled row (price $20.67, RSI 62.1, 1d −0.91, 7d +4.76, 30d −0.82, 3m +54.37, vs 20-day +5.3, vs 200-day +36.1, 50-day 18.05, 200-day 15.18, −22.44% from 52-week high, strong-up)
• adtech — 2026-08-06 intraday snapshot (MGNI $24.92, RSI 77.8, 1d +20.56, relative volume 2.2, value-area high 20.05, nearest support shelf 18.41, 20-day 19.82, 50-day 18.28, 200-day 15.22) plus the nine-name cohort
• adtech, CHANGELOG — watchlist opened 2026-08-06; monster-discoveries add dated 2026-07-30
• Google remedy status checked 2026-08-06; no final remedies order found in coverage as of that check