CXApp Inc. (CXAI) Q3 2026 earnings review
CXAI revenue nearly doubles, but its older business stays flat
CXAI's Australian purchase now supplies most of its sales, and the original business is standing still. Revenue reached $3.0 million, nearly double the previous quarter. Two-thirds of that came from EngineRoom, bought in June, by our math. Chairman and chief executive Khurram Sheikh narrowed the cash operating loss by 37% and called the quarter "an important operating inflection for CXAI".
| Sales excluding the EngineRoom purchase | About $1.0M roughly 2% under the previous quarter, by our math |
|---|---|
| Revenue | $3.005M +77% from the previous quarter |
| Cash operating loss, as CXAI measures it | $1.7M $2.7M three months earlier |
| Breakeven target | Unchanged positive cash flow in the third quarter of 2027 |
⚖️ Verdict: ⚪ Neutral
The case for CXAI sits where it did three months ago. Operating progress is real: the loss narrowed by over a third while sales rose, and management now names a single quarter for breakeven instead of a half-year. One caution: the recurring share of revenue, at 94%, trails the 99% CXAI reported a year ago.
The question now is whether CXAI can expand without buying something. Cross-selling into EngineRoom's mid-market base would show up as fourth-quarter growth; failing that, revenue settles near $3 million. That quarter, the first with the acquisition on both sides, gives the first clean reading.
🐂 Bull Case
Losses Narrow as Revenue Climbs
CXAI judges its own progress on cash EBITDA, which chairman and chief executive Khurram Sheikh calls "a measure we use to assess the underlying operating performance of our business". It leaves out financing, accounting and non-cash items. On that measure the quarterly loss came to about $1.7 million.
Three months earlier the same loss ran to $2.7 million. The improvement works out at 37% in a single quarter, and it arrived while revenue climbed 77%. Sheikh credited AI-enabled automation, shared technology infrastructure and cost work with Google's cloud team, where he said further savings have turned up.
Part of that gain is arithmetic rather than efficiency. EngineRoom carried about $1.6 million of annual adjusted profit when CXAI bought it, so holding it for three months instead of one adds roughly $0.3 million, by our math. Neither the release nor the call names a single expense line that fell.
What to watch: the fourth-quarter cash loss, reported early next year, when EngineRoom sits in both periods and the acquisition no longer flatters the comparison. Another step down of this size would show the efficiency programme, rather than the purchase, is doing the work.
Beat Reaches Beta Before a January Launch
Beat is CXAI's newest agentic AI product: software that acts on a worker's behalf. It reads the jobs already queued for a software developer or project manager, among them code reviews, tickets and follow-ups, then prepares the next step and carries it out "following human review and approval".
Outside testing began this week, after customer demos in October, and CXAI is "targeting commercial go-live in January 2027, subject to successful testing and launch readiness". Sheikh said Beat is "designed to work within tools teams already use", naming Jira, GitHub, Slack and Google Workspace.
What to watch: whether that January date holds, and whether CXAI names a paying Beat customer when it reports the first quarter of 2027. A launch that slips would push the product's first revenue past the quarter in which Sheikh has promised positive cash flow.
Positives this quarter didn't test
Several of the reasons to back CXAI got no fresh numbers this time. Each was quantified at some point over the past year, and each needs a figure in the fourth quarter or the annual report before anyone can call it progress rather than intent.
- Cross-selling: no customer count and no revenue figure for CXAI software sold into EngineRoom's 50-plus mid-market accounts, the core of the growth argument.
- Customer retention: net revenue retention stood at 99.3% in June and went unmentioned this time.
- TouchSource: the partnership aimed at 11,000 digital directory screens has not come up since last year.
- Pipeline: the enterprise funnel of more than 20 opportunities described a year ago got no update.
What to watch: a number attached to cross-selling, which is the one lever that would separate CXAI's growth from the acquisition. Even a handful of named mid-market accounts buying enterprise modules would show the two businesses fit together.
🐻 Bear Case
Almost All the Growth Came From EngineRoom
CXAI bought EngineRoom, an Australian firm selling growth intelligence to mid-market companies, in the first week of June. It carried more than $8 million of annual revenue, or roughly $2 million a quarter. July to September is the first stretch that includes all three months of it.
Take that $2 million out and about $1.0 million remains, by our math. The original CXApp business sold much the same three months ago, and a touch more in the equivalent quarter of 2025. Sheikh gave no standalone figure for it.
- EngineRoom's share of third-quarter revenue, by our math: about two-thirds
- New recurring revenue: $1.26 million, of which about $1.25 million is EngineRoom's two extra months
- Recurring revenue growth of 81% in three months, against roughly nothing once those months come out
So the quarter reads differently from the headline. CXAI describes an operating inflection, but the step-up in sales is a calendar effect. The flywheel Sheikh describes, in which enterprise and mid-market products sell each other, has yet to show up in any disclosed number.
What to watch: fourth-quarter revenue, the first period with EngineRoom on both sides of the comparison. Growth there would come from customers rather than from the deal, while a flat quarter leaves the 2027 revenue target resting on cross-selling alone.
Recurring Share Trails Last Year's Level
The release leads with recurring revenue at 94% of the total, and the mix did improve by 1.8 percentage points over three months. Set against CXAI's own history, though, 93.7% is not a high reading. Recurring revenue is simply the part customers pay for on a subscription.
Subscription sales made up 99% of the total in the third quarter of 2025, and 98% both across that year and in the first quarter of 2026. EngineRoom pulled the share down to 91.9% in June. The operating model management set out in August calls for more than 95%.
What to watch: the recurring share in the fourth quarter, once the acquisition stops changing the blend. A reading back above 95% would mean the mid-market revenue is as durable as CXAI says, while a flat share keeps the question open.
Cash Covers About Five Quarters of Losses
A shareholder asked on the call whether CXAI needs more money. Sheikh answered that it holds "north of $9 million in the bank" and has "enough cash runway for up to 6 quarters", and that it can reach positive cash flow "without any additional financing or shareholder dilution".
Cash stood at $11.7 million at the end of June, so roughly $2.7 million has left the business since. At the third quarter's loss rate, $9 million funds about five quarters, by our math. The date Sheikh has set for breakeven is four quarters away, which leaves little room.
The release puts the dilution question differently. It states that "Existing financing obligations remain outstanding, and potential dilution continues", and the standstill with Avondale Capital lasts only 90 days and depends on conditions being met. Alongside that, the board authorised buying back up to $1 million of shares, about a ninth of the cash.
What to watch: the cash balance in the annual report, due early next year, and any shares issued to the financing partner once the standstill lapses. Cash under $8 million by December would cut the funding to less than four quarters at the current burn.
Risks this quarter didn't answer
The third quarter arrived with fewer numbers than the second. The release carries no income statement, no balance sheet and no cash flow statement, and several figures CXAI itself published in August have gone missing. Each belongs in the quarterly filing, so investors can find them there.
- Annual recurring revenue: $11.5 million in June, absent this time; CXAI reported an annualized run rate instead.
- Net revenue retention: 99.3% in June, not repeated.
- Gross margin: 83% in the first quarter and 87% for 2025, with nothing published since EngineRoom arrived.
- Customer wins: "two Fortune 500 enterprise renewals" and a multi-year financial-services contract match the three wins described in August.
- EngineRoom earn-out: a two-year obligation linked to revenue, with no sizing this quarter.
What to watch: the quarterly filing, which should carry the statements and the earn-out liability. A recurring revenue figure well above $11.5 million would settle whether the acquired base is expanding or merely consolidated.
👓 Other Themes
The Gap Between Using AI and Running On It
Sheikh cited research he says shows "76% of small businesses already using AI, yet only 14% have fully integrated into their core operations". That gap is the market CXAI sells into, because it offers "an agentic operating layer" meant to turn AI into finished work. At six industry events in October, buyers asked for production systems rather than pilots, he said.
💲 Other KPIs
The quarterly loss per share fell to $0.02 from $0.10 three months earlier, and stood at $0.13 a year ago. CXAI presents these figures on the share count before its reverse split. Neither the release nor the call explains why the per-share loss shrank far faster than the underlying cash loss.
Four times the quarter's revenue, the yardstick CXAI uses for scale. It passed $12 million for the first time, against about $6.8 million on the same basis in June. Because the measure multiplies a single quarter by four, it carries the full weight of the acquisition and moves with one period.
🔮 Guidance
Unchanged. CXAI still aims to turn cash-flow positive in the third quarter of 2027, and now names the quarter rather than the half-year it used in August. By our math the gap to close is $1.7 million over four quarters. The two levers Sheikh named cover about 92% of it, and only if every extra dollar of sales carries the 87% gross margin last reported.
New. Sheikh set a target of "approximately $4 million in quarterly revenue by Q3 2027", about a third above this quarter. That works out at roughly 7% a quarter, by our math, against the 77% CXAI posted in the third quarter on the back of the acquisition. He named cross-selling and EngineRoom's mid-market reach as the sources.
Unchanged. The company repeats the scale it set out when it bought EngineRoom in June, and the third quarter delivered precisely that. Because the yardstick multiplies one quarter by four, it rises or falls with a single period. Meeting the 2027 revenue target would lift it to about $16 million.
New. Sheikh is chasing "more than $700,000 in additional quality cost efficiencies through AI automation", shared infrastructure and tighter delivery, with pricing work alongside Google named as one source. He attached no date beyond the 2027 cash-flow goal. The sum covers about 41% of the distance to positive cash EBITDA, by our math.
❓ Key Questions
Where does annual recurring revenue stand now?
CXAI put it at $11.5 million in June and skipped it this time, offering an annualized run rate instead. That run rate simply multiplies one quarter by four and says nothing about contracts signed.
What is the gross margin with EngineRoom included?
The last figures were 83% for the first quarter and 87% for 2025, both before the acquisition. EngineRoom is a services-heavy mid-market business, so the blended margin matters for the breakeven arithmetic.
How fast is the business growing without EngineRoom?
By our math the rest of CXAI sold about $1.0 million, roughly flat in three months. Management has never published a standalone figure for the original business, which makes the growth claim hard to test.
Are these customer wins new, or August's wins again?
The release lists two Fortune 500 renewals and a multi-year financial-services agreement, matching the three wins described three months ago. Investors cannot tell what the third quarter actually added.
How many shares could Avondale Capital receive?
The standstill runs 90 days and depends on conditions, and the release says dilution continues. Neither document sizes the shares that existing financing obligations could create.