Forgent: A $3.0 Billion Backlog, a 37% Sponsor Stake, and the Conversion Test

Our view

Forgent's fiscal fourth quarter settles the demand question and opens the conversion one. Bookings of $1.5 billion, a 3.3x book-to-bill and a $3.0 billion backlog arrived with fiscal 2027 guidance of $2.4-2.6 billion of revenue and $575-625 million of adjusted EBITDA, after the shares had fallen 28% in 20 sessions and 56% from their June closing high. [1][19]

Our business view is constructive on demand and unresolved on economics per share. Three things decide whether the backlog is worth what the market now pays: whether it converts on the back-half-weighted schedule and at the 24% margin the guide assumes, who funds the receivables and inventory the ramp requires, and how much of the equity story is the sponsor's exit. Vehicles controlled by Neos Partners still hold 37.2% after four offerings that moved 191 million shares, and the 10-K says additional follow-ons are expected. [2][7]

The 11:00 ET call was held before this note was finished; we use the published transcript below. We hold no five-session directional view, and the case may stay unresolved through the fiscal Q1 report.

Starting market state

The last completed session is Monday, September 14. FPS closed at $28.64 (20:00:02 UTC), down 10.0% from Friday's $31.82 on 12.19 million shares, 2.3 times the prior 20-session average of 5.29 million (about $171 million a day). Five sessions: -8.6%; 20 sessions: -27.8% from $39.66 on August 14. In the 126-session window since March 16 the highest close was $64.59 on June 4 (intraday high $66.00, June 22) and the lowest close $27.97 on March 30; Monday's close was 55.7% below the June closing high. [19]

Monday's decline coincided with an 18:04 UTC wire report that data-center infrastructure shares were lower after AI executives signaled the industry's growth may have to be slowed, the pacing discussion in our September 13 weekly; no Forgent filing or release was published that day. [16][23]

Tuesday's session is not complete. The premarket indication was $30.98 at 11:19:29 UTC (07:19 ET), 8.2% above Monday's close, inside an 08:00-11:19 UTC range of $28.90-32.37 (no premarket volume reported). At 16:06:10 UTC (12:06 ET) the regular session was at $31.61, up 10.4%, range $29.50-33.09, with 18.5 million shares traded. These are in-progress prints, not a close or a completed reaction. [20][21]

What changed

The release was posted at 06:30 ET and furnished on Form 8-K at 06:40:38 ET; the 10-K followed at 06:43:36 ET, then the deck and the 11:00 ET call. [1][2][3][5][6]

Fiscal Q4 (to June 30): revenue $461.7 million, up 94%; bookings $1,503 million, up 375% year over year and 73% from Q3's $867 million, a 3.3x book-to-bill after 2.3x; backlog $3,023 million, up 256% from $850 million and 53% from $1.98 billion at March 31; net income $66.1 million (14.3%); adjusted EBITDA $112.7 million (24.4%, from 22.4%); adjusted EPS $0.25; operating cash flow $74 million; capex $31 million. Revenue, adjusted EBITDA and adjusted net income exceeded the top of the May ranges ($1,350-1,390 million; $310-320 million). [1][3][8]

Fiscal 2026: revenue $1,420.1 million (+89%), adjusted EBITDA $322.9 million (22.7%), adjusted EPS $0.68 on 304.7 million adjusted shares. Operating cash flow $109.1 million against $115.9 million of capex; working capital absorbed $124.5 million, with receivables using $172.7 million and inventory $142.9 million, offset by $153.0 million of deferred revenue (customer deposits) and $68.5 million of payables. Year-end cash $97.5 million; $598.5 million on the term loan; $250 million revolver undrawn; tax receivable agreement liability $338.9 million. [1][2]

Mix: Powertrain Solutions (modular e-houses and skids) was $357.4 million for the year, up 259%, $147 million in Q4 (about 32% of revenue) and about 40% of the June 30 backlog. Data centers were about 59% of revenue, grid 21%, industrial 10%. One unnamed customer was about 11% of revenue and 15% of receivables; none exceeded 10% a year earlier. The company reports a first direct order from a frontier AI lab (transformers and switchgear for a campus above 1 GW) and a signed MSA with a hyperscaler. [2][3][4]

Guidance: fiscal 2027 revenue $2,400-2,600 million (+76% at midpoint), adjusted EBITDA $575-625 million, adjusted EPS $1.26-1.40; Q1 revenue $445-465 million and adjusted EBITDA $90-100 million including about $10 million of one-time hiring and start-up costs, both rising each quarter after Q1; capex about $87 million including a $35 million Tijuana expansion that lifts revenue capacity to about $5.8 billion by fiscal Q4 2027; operating cash flow above $300 million, weighted to the second half. Backlog covers more than 90% of the guide and is entirely firm purchase orders, management says. [1][3][4]

Expectations

Three layers need separating.

Company guidance is above. Wire-reported consensus (Benzinga, 10:30-10:31 UTC): fiscal 2027 revenue $2.078 billion against the $2.4-2.6 billion guide, adjusted EPS $1.14 against $1.26-1.40, Q4 revenue $429.9 million against $461.7 million. The wire's first Q4 EPS line ($0.31) was corrected at 13:11 UTC to the company's $0.25 against a $0.24 estimate, a one-cent beat. StockAnalysis, citing ten S&P Global analysts on a page rendered at 07:42 ET, showed $2.09 billion and $1.17 for fiscal 2027 and a $59.90 average target, possibly predating the release. The guide midpoint is 37% above the $1,828 million fiscal 2027 revenue in the company's own December IPO model. [3][12][13][14][15][17][18]

Our arithmetic on 304.43 million shares (274.53 million Class A plus 29.90 million Class B at September 8): at Monday's $28.64 the equity is $8.72 billion and enterprise value about $9.22 billion after $501 million of net debt ($9.56 billion with the TRA): 15.4x the guided EBITDA midpoint (14.8-16.0x across the range) and 21.5x guided EPS (20.5-22.7x). At the $30.98 premarket bar, 16.6x and 23.3x; at the 12:06 ET print of $31.61, 16.9x and 23.8x. [2][19][20][21]

The guide's implied margin is 24.0% at every point of the range, level with Q4's 24.4%, so guided earnings growth is volume. The shape matters: Q1's $455 million midpoint is below Q4's $462 million, so the other three quarters must average about $682 million, and the CFO put Q1 margin near 21% with about 200 basis points of recovery in Q2. Sensitivity, not forecast: $2.2 billion at 21% gives about $462 million of EBITDA and $0.98 of EPS, making Monday's close 20x and 29x; $2.4 billion at 22% gives $528 million and $1.15, or 17.5x and 25x. [1][3][4]

We have no market-implied expectation. A 20-session decline into a beat-and-raise fits a duration discount on data-center capital spending, sponsor supply, or both; price alone does not separate them.

Observed response

The observed response ends at Monday's close, before the release; Tuesday's prints are in progress, and wire items at 11:48 and 13:53 UTC describing the shares as higher restate them. [15][21]

The announcement and learned dates are September 15 before the open; the economic events are older. The bookings were taken April-June and the first AI-lab order in Q4. The Neos vehicles' sale of 43.65 million Class A shares around the $49 July offering (29.09 million secondary plus 14.56 million exchanged from Opco interests and sold as the company's primary shares) was reported on Form 4 on July 8. The four September 3 Form 4s are September 1 RSU and option awards to four officers (options struck at $29.27), not sales. Monday's 12.2 million-share session on a sector headline is the most recent completed evidence of how this holder base behaves under an AI-pacing scare, and it sold. [10][11][16]

On the call, management said backlog is entirely firm purchase orders, pricing was unchanged quarter to quarter, it sees no meaningful pushouts, and headcount roughly doubled to about 3,500 in a year with hundreds more hires needed each quarter. No questioner asked about the 11% customer or sponsor supply. [4]

Comparable history and its limits

ChartLibrary returned an empty packet for FPS on September 14: no happening identity, because the symbol has traded only since the February 5 listing and is not in the coverage register. We have no state or shape analog cohort and make no continuation claim. [22]

The closest covered comparable is Powell Industries, where our September 13 note framed the same test: $934 million of orders and a $2.4 billion backlog against a weaker stock, with the answer resting on conversion timing and margin quality rather than the order count. Forgent is a different instrument: a seven-month-old listing with a sponsor at 37.2%, a $600 million term loan, an Up-C with a tax receivable agreement, one customer above 10%, a workforce that doubled in a year, and a 3.3x book-to-bill. Powell has decades of contract-accounting history; Forgent's 10-K says it has limited historical experience of the realization it achieves on backlog. [2][4][24]

The mechanism under test is the weekly's: AI pacing, usage of deployed systems and the timing of physical projects can move separately, and an unchanged order book can later contradict a headline-driven exit. Forgent's order book grew 53% in a quarter; that says nothing about the next five sessions. [23]

The countercase

The constructive reading: a 3.3x book-to-bill with $5.8 billion of capacity by fiscal Q4 2027 means the guide needs no heroic conversion; $263.9 million of customer deposits show buyers pre-funding part of the build; Q4 margins expanded while new campuses were still ramping; backlog covers more than 90% of the year with capacity held back for book-and-ship orders; and a stock at 15x guided EBITDA after a 56% drawdown has already absorbed a large reset. The four offerings priced at $27, $29.50, $47 and $49, all above Monday's close. [1][3][4][7][9]

The cautious reading: the 10-K names the risks. Powertrain Solutions concentrates revenue in fewer, larger customers who may negotiate more favorable pricing and payment terms, which is how a 24% margin guide and a receivables build could both go wrong; backlog may not carry historical margins; tariffs and transformer efficiency standards are named cost risks; demand depends in part on continued interest in developing artificial intelligence. Every dollar of net primary proceeds from the four offerings, about $2.2 billion, redeemed Neos-controlled Opco interests rather than funding the balance sheet. The 113 million sponsor shares, about 21 days of average volume, are past the 180-day IPO and 60-day July lock-ups, carry registration rights, and the 10-K says additional follow-ons are expected. The no-pushouts answer is management's, one day after the sector sold off on a pacing signal. [2][4][7][10]

What would change our view

Weaken the constructive business view if the fiscal Q1 report shows revenue below $445 million or adjusted EBITDA below $90 million, one-time costs well above $10 million, gross margin back below the 35% fiscal 2026 level while revenue grows, receivables growing faster than revenue without matching deposits, cancellations becoming material, or a new secondary filed at or below Monday's close. Strengthen it if Q1 lands in range with Q2 margin up the guided 200 basis points, deposits keep funding a meaningful share of working capital, and the top customer's share falls as the base broadens.

Price alone settles none of this: a completed close above Tuesday's intraday prints or below Monday's close would be a market observation, not evidence about conversion.

What we will revisit

First: today's completed session, September 15, the first observation after the release; we will record the close, volume and the open-to-close move separately from the overnight gap and the intraday prints above.

Second: five sessions on September 21 and ten on September 28, counting September 15 as session one. The September 15-16 FOMC meeting is a confounder for the whole window. [25]

Third: any Form 4, Form 144 or prospectus supplement from the Neos vehicles, which would date the sponsor-supply question, and the fiscal Q1 report, not yet scheduled, the first operating test of the Q1 guide and the quarterly ramp. Items the call left open: the 11% customer's end market, Powertrain payment terms and deposit practice, and sponsor intent. No model position is created by this coverage. The issuer's events-and-presentations route was not machine-accessible to the scanner at the check time; the results deck itself was retrieved directly from the investor site and is cited, so the gap is limited to any other event materials.