Blackbaud's 95% Sprint Is Betting on a Model That Doesn't Ship Until 2027
Lantern, Platform for Good and a shrinking share count re-rated a cheap cash machine — but guidance still contains no AI revenue, and the renewal calendar runs the other way
BLKB · Earnings Call · 2026-10-28
A rerating that arrived before the product
Blackbaud added 94.7% in the ninety trading days into its 28 October print — a single unbroken up-17w:+95% segment off a stock that had lost half its value in the twenty-three weeks before it. The tape peaked at $56.48 on 22 October and sits 7% below that, on a $2.38bn market cap. Whatever happened at BBCON in late September, the market decided it was a different company. The number that moved the stock, though, is not in the transcript. It is the multiple. Coming into Q3, Blackbaud traded at 1.2x price-to-revenue and 8.9x price-to-net-income — the residue of a nine-year de-rating from six turns of sales. Q3 then delivered revenue of $293m (+4.1%), adjusted EBITDA of $111m at a 38% margin, non-GAAP EPS of $1.36 (+24%) and, per the CFO, “a very strong free cash flow quarter of $126 million” — Chad Anderson, Executive Vice President and Chief Financial Officer · 2026-10-28. Leverage fell from 2.7x to just under 2.3x, and 7% of the share count was retired in nine months — roughly a tenth of a quarter's revenue. So the setup was real: a levered, slow-growth, cash-gushing vertical software company with 22.7% free cash flow margins, a $271m net debt position being paid down, and a pledge to spend 50%+ of cumulative free cash flow through 2030 on buybacks. What the quarter added was a story to hang it on.What is actually new: Lantern
The keyword stack turned over almost completely. Prior quarters were full of mechanics — general availability, Early Adopter, annual subscription fee, seats based pricing. This quarter's company-unique leaders are architecture: Lantern, Platform for Good, Signal Graph, intelligence layer, collaboration with Databricks, data health. That is not boilerplate recycling; those are new nouns. The pitch is vertical specificity against general-purpose models: “General purpose models learn from the Internet Lantern learns from philanthropy.” — Michael Gianoni, Chief Executive Officer, President and Vice Chairman · 2026-10-28 Built with Databricks on open-weight models and post-trained on synthetic scenarios, it is due to power Raiser's Edge NXT features in 2027. A rebuilt Raiser's Edge was pulled forward by more than a year, and the CEO claims “the pace of innovation has picked up a lot. We have announced more products this year than we ever have.” — Michael Gianoni, Chief Executive Officer, President and Vice Chairman · 2026-10-28The uncomfortable part: this wave is not new, and the market has been selling it
Blackbaud has been telling the agents story for at least five quarters. In July the CEO was already describing the roadmap: “So development agent a while ago into general availability, 4 more in the press release just recently... predominantly at our BBCON conference.” — Michael Gianoni, Chief Executive Officer, President and Vice Chairman · 2026-07-29 The October call delivered that promise — it did not originate it. Worse, the wrappers Blackbaud just adopted are the ones the tape has been punishing. Across the market's 360-day decliners, system of record is down 30.9% across 25 tickers and agentic operating system is down 32.3%. Blackbaud's entire thesis is the migration from systems of record to a "unified operating system for social impact" — precisely the phrasing that has bled out in public SaaS. Meanwhile agentic anything is now consensus among this week's reporters: ServiceNow's control tower, Oracle's personal AI apps, Adyen and eBay's agentic commerce. The differentiation claim rests on the domain model, not the agent. There is a clean inversion on data center too. The market's AI-datacenter complex is deflating over 30 days — the aggregate keyword is off 5.4% with fifty decliners. Blackbaud's data-center keyword is a cost story pointed the other way: closing the last two company-operated colos by year-end, with Chad Anderson flagging that “it kind of blends in over time. Starting in 2027.” — Chad Anderson, Executive Vice President and Chief Financial Officer · 2026-10-28 Same words, opposite economics.The arithmetic the rally is not underwriting
That sentence has anchored every call since February, when the CEO said the long-term aspirational model does not include the AI products because “It's too early.” — Michael Gianoni, CEO, President and Vice Chairman · 2026-02-10 It is still too early. Lantern reaches customers in 2027. The Development Agent has been in market roughly seven months and is being bundled to pull through core Raiser's Edge seats — a good motion, but not a revenue line yet. What is in guidance is an unusually clean piece of bad-then-good math on renewals. This year's contractual renewal cohort is roughly 40% larger than last year's, which suppresses gross dollar retention into the first half of 2027; the 2027 cohort is nearly 25% smaller, so churn should fall materially and retention should return to the 91–92% historical norm by the end of 2027.Our 26 full year guidance does not include any meaningful revenue contribution from our new AI products.
Note the timing: the tailwind is a second-half 2027 event, and the retention recovery is a 2028 revenue event. So the Q4 step-up analysts pushed on — from ~4% organic toward something closer to 5.5% — is being bridged with holiday giving and a new platform fee catching up to industry norm, not with AI. And the CEO was explicit that Q3 contained none of the good stuff: no viral giving events, no holiday seasonality. Fine, but that means the fourth quarter is carrying the year, and the 2027 hand-off is a smaller cohort, not a new product.As we move beyond this larger renewal cohort, we are planning for a material reduction in churn to revenue as our recurring revenue renewal cohort in 2027 will be nearly 25% smaller than 2026.