# Innodata: free cash flow looks like 10 times, but customers funded most of it INOD, Nasdaq. Published October 20, 2026. Price is an intraday trade at 7:46 pm Prague time (1:46 pm New York) on October 20, 2026, during the regular session; the October 19 close was $54.00. ## Verdict: HOLD at $54.92 At $54.92 the shares trade 7.5% above my Base value of $51.07, a downside of 7.0%. The growth is real: revenue rose 58% in the second quarter and the operating margin is 18%. But most of the $159 million of first half free cash flow was customer prepayments and unpaid project costs, two customers were 71% of second quarter revenue, and a $300 million share sale program is open. Confidence is C and the required margin of safety is 40%, which puts the maximum buy price at $30.64. Hold for holders, watchlist for new money. Third quarter results on November 5 are not in this analysis. - Bear value: $24.47 - Base value: $51.07 - Bull value: $71.92 - Downside to base value: -7.0% - Maximum buy price: $30.64 - Required margin of safety: 40% - Normalized owner cash yield: 2.0% - Price to trailing sales: 5.95 times - Top two customers, Q2 revenue: 71% - Modeled annual return to end of 2028: 5.7% - Business quality: 69/100 - Valuation score: 40/100 - Confidence: 15/25, grade C A screener will tell you Innodata trades at about 10 times free cash flow. In the first half of 2026 the company reported $159.1 million of free cash flow on $182.2 million of revenue, which would be an 87% cash margin. It is not that kind of business. The cash came from $65.2 million of customer advances and a $71.9 million rise in payables and accruals, including $61.8 million of project costs incurred but not yet paid at June 30. Take those out and the trailing twelve-month owner cash is about $38 million (my estimate), a 2.0% yield on the $1.9 billion market value. At $54.92, an intraday trade at 7:46 pm Prague time (1:46 pm New York) on October 20, 2026, the shares trade 7.5% above my triangulated Base value of **$51.07**, a downside of 7.0%. The Bear value is $24.47, 55.5% below the price, and the Bull value of $71.92 is 31.0% above it. Confidence is grade C (15/25), the required margin of safety (MOS) is 40%, and the maximum buy price is **$30.64**. The price is close to my Base and well above the buy ceiling, so the verdict is Hold. Valuation route: a listed provider of data engineering for AI developers, a services business paid per project, not a subscription software company. Primary method: 25 times normalized owner cash in 2028, discounted to today. Secondary method: 18 times 2028 operating profit. Weights are 60% and 40%. A ten-year cash flow model ($55.03) is a cross-check and carries no weight. Classification: quality growth, with no compounder premium. ## What Innodata is Innodata builds the data that AI developers train and test their models on: large training datasets, human evaluation of model output, and, more recently, programs for agentic and robotic systems. Its customers are a small number of large technology companies, and it competes with Appen, Scale AI, TELUS Digital, Surge AI and the customers' own teams. Smaller Synodex and Agility businesses sit alongside the main AI data services unit. Revenue went from $86.8 million in 2023 to $170.5 million in 2024 and $251.7 million in 2025, a 70.3% compound rate over two years. In the second quarter of 2026 it reached $92.1 million, up 58%, with an adjusted gross margin of 49% against 43% a year earlier. That is a real improvement, but it is project revenue. At-will programs can be cut, and nothing in the filings turns it into subscription income. The customer mix is the sharper story. The largest customer fell from 56% of revenue in the first quarter to 37% in the second, and a second customer rose from 17% to 34%. In dollars, from the rounded percentages, the first dropped from about $50.5 million to $34.1 million a quarter while the second doubled from about $15.3 million to $31.3 million (my arithmetic). Concentration moved, it did not disappear: the top two accounts were 71% of second quarter revenue. ## The latest results | Latest reported period | Result | | --- | --- | | Second quarter 2026 revenue | $92.1 million, up 58% from $58.4 million | | Second quarter adjusted EBITDA | $25.4 million, 27.5% margin | | Second quarter net income | $14.4 million, $0.41 a diluted share | | First half revenue | $182.2 million, up 56.1% from $116.7 million | | First half operating income (my calculation) | $32.8 million, 18.0% margin, against 14.7% | | First half cash from operations | $164.4 million, against $15.0 million | | Cash and Treasury notes, June 30 | $250.4 million, against $82.2 million at year end | | 2026 revenue guidance | At least 40% growth, raised from about 35% | Adjusted EBITDA adds back stock-based compensation (SBC, shares given to staff instead of cash) and depreciation, so it is not the profit I value. The operating margin of 18.0% after SBC is. The company will report third quarter results after the close on November 5 (the call is at 5 pm New York time, 11 pm in Prague), and none of it is in this analysis. Leadership changed too: on September 30 Rahul Singhal became CEO and founder Jack Abuhoff moved to executive chairman. ## Which cash to value Cash from operations of $164.4 million in six months looks like a windfall, so I checked where it came from. Net income and non-cash items explain $46.5 million. The other $117.9 million came from working capital: customers paying in advance and the company holding back what it owes suppliers. | $ million, first half 2026 | Amount | | --- | --- | | Net income, SBC, depreciation and other non-cash items | 46.5 | | Customer advances | 65.2 | | Payables, accruals and other liabilities | 71.9 | | Receivables, prepaids and other working capital | -19.2 | | Cash from operations | 164.4 | The company says that net of these prepayments its cash was about $134 million. The 10-Q gives $67.0 million of unfulfilled customer advances and $61.8 million of project costs incurred but unpaid, which is more than the release's $116 million of difference, and I could not reconcile the two. I use the filing's figures. The line between a customer's money and Innodata's money matters, because Innodata passes some of it through to vendors. | Owner cash, $ million | Twelve months to June 2026 | | --- | --- | | Reported free cash flow (cash from operations less capital spending) | 183.8 | | Less SBC | -18.6 | | Free cash flow after SBC | 165.2 | | Less customer advances received | -65.2 | | Less project payables outstanding (my estimate) | -61.8 | | Normalized owner cash (my estimate) | 38.3 | That is 23.2% of the headline figure. It is deliberately strict: the opening split of project payables is not available, so I treat all $61.8 million as new, and the exact answer is somewhere between this and the full number. A second approach, which strips working capital out of the first half altogether, taxes the first half at 24% instead of 16.0% and keeps pension cost, gives $24.6 million for six months. Either way the price is not 10 times cash. On the normalized figure it is 49 times. | Period, $ million | Free cash flow | After SBC | | --- | --- | --- | | 2024 | 27.1 | 23.1 | | 2025 | 35.6 | 24.5 | | First half 2026 | 159.1 | 146.0 | | Twelve months to June 2026 | 183.8 | 165.2 | SBC itself has more than doubled, from $4.0 million in 2024 to $11.1 million in 2025 and $13.1 million in the first half of 2026, which is 7.2% of revenue. I count it as a cost throughout. Reported net income of $46.5 million over twelve months is real, and so is the 17.5% trailing operating margin, so the business earns money. What it does not yet show is how much of that reaches shareholders as cash once the projects are settled. ## Shares, cash and the share sale program | $ million | Amount | | --- | --- | | Cash | 240.3 | | Short-term Treasury notes | 10.1 | | Less unfulfilled customer advances | -67.0 | | Less project costs incurred but unpaid | -61.8 | | Less accrued pension | -9.5 | | Less financed software licenses | -1.7 | | Net financial assets | 110.4 | The net financial assets of $110.4 million are worth $3.11 a share and are added to enterprise value once. I treat the customer and project balances as a reserve, not as spendable surplus. It is an economic adjustment, not restricted cash, and forecast cash flows assume ordinary future receipts and payments, so the reserve is not charged twice. Leases stay in operating costs, and the undrawn credit line is not counted as cash. If I used the release's $134 million instead, Base would be $51.42 to $51.70. If I counted all gross cash with no reserve, Base would be $54.70. There were 34.383 million common shares on July 31, 2026. About 2.0 million options were exercised in the first half for $10.9 million of proceeds, which is most of the increase from 32.3 million at year end and is already in the June cash. My valuation count is 35.5 million shares: the common shares plus an allowance for earned options, with future awards charged as a cash-like cost in forecast profit and not as extra shares. A full statutory count near 36.5 million would give $49.67. The bigger share question is the new at-the-market (ATM) program, filed on August 6: the company may sell up to $300 million of stock through five sales agents, paying up to 2% commission. It has no obligation to sell, and the 8-K gives no specific use for the money. If it sold the full amount at $54.92 it would issue 5.46 million shares, about 16% more, and raise $294 million net. If the cash keeps its value, Base becomes $51.44. If half of it is wasted, $47.85. A buyer who sees only the dilution and ignores the cash is making a mistake, but so is one who ignores that the cash needs a return. I do not know what was sold after August. ## The primary valuation: 25 times owner cash in 2028 My Base starts from 2026 revenue of $360 million, which is the consensus level and above the guidance floor of about $352 million, and grows it 28% in 2027 and 25% in 2028, to $576.0 million. At a 21% operating margin after SBC, operating profit is $121.0 million. After 24% tax, $14.4 million of depreciation, capital spending at 3.5% of sales ($20.2 million) and $13.8 million of working capital to fund growth, owner cash in 2028 is $72.3 million. A buyer at the end of 2028 paying 25 times that cash would pay $1,808.6 million, a 4% cash yield. That multiple is my assumption for a business still growing fast, not a peer median, and it would be too rich for a mature outsourcing company. I discount the exit value and the interim cash back to October 20 at 10% (a 2.197 year transition), add the $110.4 million of net financial assets, and divide by 35.5 million shares. The primary Base is **$47.70**. | Exit multiple on 2028 owner cash | 15 times | 20 times | 25 times | 30 times | 35 times | | --- | --- | --- | --- | --- | --- | | Primary value per share | $31.17 | $39.44 | $47.70 | $55.97 | $64.23 | ## The secondary valuation: 18 times operating profit The second method ignores cash conversion and capitalizes profit: 2028 operating profit of $121.0 million times 18 is $2,177.3 million at the end of 2028. With the same interim cash, discount rate and bridge, the value is **$56.12**. It is higher because profit is not reduced by the capital and working capital that growth consumes. | Base case | Revenue, $m | Operating margin | Owner cash, $m | | --- | --- | --- | --- | | 2027 | 460.8 | 20% | 53.3 | | 2028 | 576.0 | 21% | 72.3 | | 2030 | 808.7 | 23% | 119.2 | | 2033 | 1,135.8 | 23% | 174.8 | | 2036 | 1,365.3 | 23% | 217.2 | As a duration check I also ran a ten-year cash flow model with growth fading to 5% and terminal growth of 2.5% at an 18% return on new capital. It gives $55.03, with 57.7% of the value in the terminal year, and it swings with the discount rate: | Cash flow model value per share | Growth 2.0% | Growth 2.5% | Growth 3.0% | | --- | --- | --- | --- | | Discount rate 8% | $74.04 | $76.91 | $80.34 | | Discount rate 10% | $53.92 | $55.03 | $56.28 | | Discount rate 12% | $42.05 | $42.52 | $43.04 | All three views start from the same customer growth and margin story, so their agreement is not proof that the story is right. They differ in how much of the profit survives as cash. ## A price check against history Historical cash gives little to hold on to. Free cash flow was close to zero in 2023, $27.1 million in 2024 and $35.6 million in 2025, and then the first half of 2026 jumped. I could not build a clean multi-year history of enterprise value to owner cash with matching definitions, so history gets no weight in my value. On today's numbers the shares are 5.95 times trailing sales and 32 times trailing operating profit after the net financial assets. The growth-adjusted check is not decisive either. Four analysts expect adjusted EPS of $1.65 in 2026 and $2.09 in 2027, which is 33.3 times and 26.3 times at $54.92, with a PEG of 0.99 on 26.7% growth. These are non-GAAP figures, and the 2027 numbers sit behind a paywall in more detail, so I read the PEG as a rough diagnostic. The snapshot I was given also shows 2026 free cash flow of $88.4 million, below the $159.1 million already reported for six months, which suggests the analysts expect part of the prepayment cash to go back out. ## Triangulating the values | Value per share today | Bear | Base | Bull | Weight | | --- | --- | --- | --- | --- | | 25 times 2028 owner cash | $22.04 | $47.70 | $67.31 | 60% | | 18 times 2028 operating profit | $28.10 | $56.12 | $78.83 | 40% | | Triangulated value | $24.47 | $51.07 | $71.92 | 100% | | Ten-year cash flow model (not weighted) | $16.19 | $55.03 | $99.39 | | The Base methods differ by 16.2% (difference over midpoint), under the 20% warning level, so there is no downgrade. I gave cash 60% because cash conversion is the binding constraint, and profit 40% because current profitability is genuine. The unweighted Base band is $47.70 to $56.12. The Bear methods differ by 24.2%, and the cash flow model is far below both at $16.19, so the weighted Bear of $24.47 is not a floor. ## What the price already assumes | Change needed to match $54.92 | Needed | My Base | | --- | --- | --- | | Operating margin 2027 / 2028, same revenue and multiples | 21.46% / 22.46% | 20% / 21% | | Which is a rise in every year of | +1.46 points | n/a | | Or a multiple on 2028 owner cash of about | 28.9 times | 25 times | | Operating enterprise value implied by the price | $1,839 million | $1,583 to $1,882 million | This is not a demanding set of assumptions. The price needs margins about 1.5 points better than my Base, or a slightly richer cash multiple, and either is reachable if the second customer scales and the mix keeps improving. That is why the verdict is Hold, not Reduce. It is also why Hold is the most I can say: a stock priced for a small positive surprise has no cushion if the surprise does not come. ## Bear, Base and Bull | | Bear | Base | Bull | | --- | --- | --- | --- | | 2026 revenue | $350m | $360m | $375m | | Growth 2027 / 2028 | 10% / 8% | 28% / 25% | 40% / 32% | | Operating margin 2027 / 2028 / mature | 14% / 14% / 14% | 20% / 21% / 23% | 23% / 25% / 28% | | Capital spending / extra working capital | 4.5% / 15% | 3.5% / 12% | 3.5% / 12% | | Terminal growth / return on new capital | 2.0% / 12% | 2.5% / 18% | 3.0% / 22% | | Net bridge adjustment | -$15m | 0 | 0 | | Cash multiple value | $22.04 | $47.70 | $67.31 | | Operating profit multiple value | $28.10 | $56.12 | $78.83 | | Value per share today | $24.47 | $51.07 | $71.92 | | Against $54.92 | -55.5% | -7.0% | +31.0% | All three cases share the 25 and 18 times multiples and the 10% discount rate, so the Bear does not cut the multiple on top of the weaker operations. It adds a separate $15 million permanent loss in the bridge. The Bull needs the second customer to keep scaling and margins to reach 28%. Neither carries a probability. Weights of 25%, 50% and 25% would give $49.63, for orientation only. ## Sensitivity and the case that flips the verdict | Base value per share | Margins 2 points lower | Base margins | Margins 2 points higher | | --- | --- | --- | --- | | Capital spending 2.5% of sales | $48.02 | $53.29 | $58.56 | | Capital spending 3.5% of sales | $45.80 | $51.07 | $56.34 | | Capital spending 4.5% of sales | $43.58 | $48.85 | $54.12 | Discount rates of 8%, 10% and 12% give $53.01, $51.07 and $49.24. A $25 million adverse change in the cash bridge costs $0.70 a share. Cutting all future owner cash by 20% or 40%, holding the profit multiple fixed, gives $45.51 or $39.96. | Case | Triangulated value | Entry at 40% MOS | Buy at $54.92? | | --- | --- | --- | --- | | Base, 10% discount rate | $51.07 | $30.64 | No | | 8% discount rate | $53.01 | $31.81 | No | | 12% discount rate | $49.24 | $29.55 | No | | Consensus revenue for 2026 and 2027 | $50.99 | $30.60 | No | | Top of the 2026 revenue range, consensus 2027 | $51.01 | $30.61 | No | | Full Bull path | $71.92 | $43.15 | No | | Bull path, 8%, 35 times cash, 24 times profit | $99.52 | $59.71 | Yes, conditional | Two single changes lift the value above the price: margins 2 points higher ($56.34) or a 30 times cash multiple ($56.03). Neither gets near a 40% margin of safety. Only a generous joint case does: Bull operations, an 8% discount rate and richer multiples all together give $99.52 and a $59.71 entry, and the call turns to a conditional Buy. I do not use that as the central case, because it needs execution, premium multiples and a low discount rate at the same time. ## How the buy price is set Confidence C calls for a margin of safety of 35% to 50%. I use 40% and add nothing on top: concentration, customer-funded cash, SBC and the share sale program are already in the scenarios and the bridge. The maximum buy price is $51.07 x (1 - 40%) = **$30.64**. Four numbers answer different questions: what the shares are worth ($51.07), the return from buying at $54.92 and holding to the end of 2028 (5.7% a year), the price that gives a 12% annual return ($48.36) and the price that leaves a 40% margin of safety ($30.64). The return path assumes no distributions, no share sale, normal working capital and $135.7 million of retained owner cash, and sells at $62.03 a share at the end of 2028. It leaves out interest on the cash, fees and personal taxes, and it is not today's value. ## The dividend There is none and I assume none. This is a capital appreciation case. A large cash balance and an undrawn credit line do not create a dividend or a buyback. ## Reasons to own - **Growth and operating leverage that show up in the accounts.** Revenue grew 56.1% in the first half, and the operating margin rose from 14.7% to 18.0%. - **A better mix.** The adjusted gross margin is 49%, up from 43% a year earlier, so growth has not come at the cost of margin. - **A second large account.** If the customer that went from 17% to 34% of revenue stays and grows, concentration risk becomes growth, not exposure. - **Reusable data and research.** Datasets and evaluation tools may serve more than one customer, which would add profit without matching labor. - **A positive balance sheet after the reserve.** Net financial assets of $110.4 million remain after counting customer and vendor money as not yours. ## Main risks - **Two customers.** They were 71% of second quarter revenue, and the largest one's revenue fell by about a third in a single quarter. Programs can be cancelled. - **A cash reversal.** $128.8 million of customer advances and unpaid project costs will go out as projects settle, so reported cash flow can turn negative. - **Dilution.** SBC is rising, and the $300 million share sale program could add about 16% to the share count if used in full at today's price. - **Project, not subscription, revenue.** Renewal depends on delivery, and customers can bring work in house or switch providers. - **Cost pressure.** Cloud, labor and licensing costs can pull margins below 20% while revenue still grows. - **Leadership change and capital use.** A new CEO and an untested plan for new cash add execution risk. ## Management execution | Year | Initial growth target | Outcome | | --- | --- | --- | | 2024 | 20% | 96% (revenue $170.5 million) | | 2025 | 40% or more | 47.6% (revenue $251.7 million) | | 2026 | About 35% or more | Raised to 40% or more after Q2 | Management has beaten its own growth targets each year, from a low base in 2024 and a larger one since. That does not cover cash, and the release's $134 million cash figure does not match the filing's separate reserve balances. I score management credibility 15/20: revenue delivery 4/5, profitability and capital allocation 4/5, disclosure and cash transparency 3/5, strategic execution 4/5. The cash definition and the use of any ATM proceeds are the two items to watch. ## Stock Analyza scorecard | Business quality | Points | Maximum | | --- | --- | --- | | Moat and pricing power | 13 | 20 | | Return versus cost of capital | 12 | 20 | | Balance sheet and resilience | 13 | 15 | | Earnings and cash quality | 7 | 15 | | Growth and reinvestment runway | 13 | 15 | | Management and capital allocation | 8 | 10 | | Dilution and governance | 3 | 5 | | Total | 69 | 100 | On the sector scorecard Innodata scores 7/12, quality growth: revenue growth is excellent, but recurring economics and customer breadth are not proven. The valuation score is 40/100 (modestly above central value, unattractive on the risk-adjusted entry): 16/40 for the discount to value, 4/20 for protection against the Bear, 8/15 for agreement between methods, 4/10 for history, 5/10 for what the price implies and 3/5 for data quality. ## Confidence score | Component | Points of 5 | | --- | --- | | Data quality and provenance | 4 | | Business predictability | 2 | | Valuation robustness | 3 | | Accounting transparency | 4 | | Scenario dispersion | 2 | | Total, grade C (13 to 17) | 15 of 25 | ## What I would watch from here ### Supports the thesis - 2026 growth at 40% or more with the operating margin heading to 20% or 21%. - Top two customers below 60% of revenue, and the largest customer stable or growing. - Customer advances and project payables settled without a sharp fall in cash, and ATM shares, proceeds and use reported together. ### Watch closely - Top two customers between 60% and 75%, or the largest customer's revenue falling again with replacement elsewhere. - Operating margin between 15% and 20%, or capital spending and cloud costs above plan. - Cash growth that depends on timing, or issuance and its use that are not clear. ### Breaks the thesis - Guidance cut below 40% growth, or top two customers above 75%, or the loss of a major client. - Normalized owner cash weakening after advances and payables settle, or a shortfall that makes cash quality impossible to verify. - Share issuance used on projects that earn less than the cost of capital, or a restatement or control weakness. ## The most important thesis breakers - **A major client.** Losing or sharply cutting one without profitable replacement. - **Cash conversion.** Weak owner cash after the advance-funded obligations are settled. - **Capital allocation.** ATM-funded spending that earns too little, or compensation dilution that keeps climbing. ## Stress test and independent check I rebuilt the model from its stated inputs with separate code and reproduced: the twelve-month and first half bridges, including the $117.9 million working capital swing and the $110.4 million of net financial assets; the annual rows for Bear, Base and Bull; the $47.70 and $56.12 values and the $51.07 blend; the $55.03 cash flow value; the 16.2% divergence; the $30.64 buy price; the $62.03 end-2028 value, the 5.7% return and the $48.36 entry; the 1.46 point reverse margin; the ATM cases ($51.44 and $47.85); the rate, margin and capital spending tables; and the $99.52 joint case. I found no construction error. Cutting every future cash flow by 20% and 40% gave $45.46 and $39.85 in my run against $45.51 and $39.96; the difference is the 2026 stub, which the original does not cut and is immaterial. From the company's documents I confirmed the second quarter and first half results, the cash and Treasury balances, the $67.0 million of advances and $61.8 million of unpaid project costs, operating cash flow, capital spending and SBC, the option exercises and proceeds, the July 31 share count, the ATM terms, the November 5 reporting date, the CEO change, the customer percentages, the revenue and growth targets for 2023 to 2025 and the 2026 guidance changes, and the four-analyst consensus. I could not re-read the $9.5 million pension claim, the financed licenses, the first half customer shares (46% and 26%), the $13.1 million expected capital spending, the $88.4 million consensus free cash flow for 2026, the controls conclusion and the current ratio, so those come from the supplied analysis. Three further points. Discounting at year end is conservative, and mid-year timing would give a Base of about $51.22, 0.3% higher. The price is an intraday trade of $54.92; the October 19 close was $54.00, 1.7% lower, and the verdict is the same at either price. And two screen figures do not reconcile: the broker page shows EPS of $1.43 with a P/E of 41.86, which would be 38.4 times, and a price to free cash flow of 29.25 against my 10.3 on reported cash flow. Main data limits: I do not know October cash or how much stock the company has sold under its program, the release and the filing define cash after prepayments differently, and customer contract terms are not public. These are why Confidence is C. ## Final verdict: hold at $54.92 Innodata is growing fast, earns a real 18% operating margin, and its second large customer is arriving as the first one shrinks. But the free cash flow that makes the screens look cheap is largely other people's money, the price already sits near my Base, and the buy ceiling of $30.64 is 44% below it. At $54.92 there is no margin of safety in either direction. **Verdict: HOLD at $54.92, watchlist for new money. Triangulated Base value $51.07, downside 7.0%, Confidence C (15/25), required margin of safety 40%, maximum buy price $30.64.** For holders this means keeping the position and watching the November 5 report, because the gap to Base is small. For new money it means waiting for a lower price or for owner cash to hold up after the projects settle. About half of the $250 million in cash and Treasury notes belongs, for now, to customers and suppliers, and the stock is priced as if all of it were Innodata's. ## More analyses on Patreon This analysis is free. Premium members get every in-depth analysis, the full archive and a vote on which stocks get analyzed next: https://www.patreon.com/stockanalyza ## Sources Primary sources include Innodata's [second quarter 2026 quarterly report on Form 10-Q](https://www.sec.gov/Archives/edgar/data/903651/000110465926092021/inod-20260630x10q.htm), the [second quarter 2026 results release](https://www.sec.gov/Archives/edgar/data/903651/000110465926092010/tm2621499d1_ex99-1.htm) and its [version on the investor site](https://investor.innodata.com/news/news-details/2026/Innodata-Reports-Record-Second-Quarter-2026-Results/default.aspx), the [2025 annual report on Form 10-K](https://www.sec.gov/Archives/edgar/data/903651/000110465926020655/inod-20251231x10k.htm), the [August 6 Form 8-K](https://www.sec.gov/Archives/edgar/data/903651/000110465926092133/tm2622364d1_8k.htm) and [prospectus supplement](https://www.sec.gov/Archives/edgar/data/903651/000110465926092131/tm2622086-4_424b5.htm) for the share sale program, the [third quarter reporting schedule](https://investor.innodata.com/news/news-details/2026/Innodata-to-Report-Third-Quarter-2026-Results/default.aspx), and the results releases for [2025](https://investor.innodata.com/news/news-details/2026/Innodata-Reports-Fourth-Quarter-and-Full-Year-2025-Results/default.aspx), [2024](https://investor.innodata.com/news/news-details/2025/Innodata-Reports-Fourth-Quarter-and-Full-Year-2024-Results/) and [2023](https://investor.innodata.com/news/news-details/2024/Innodata-Reports-Fourth-Quarter-and-Fiscal-Year-2023-Results-02-22-2024/). Consensus revenue and EPS come from a third-party [forecast page](https://stockanalysis.com/stocks/inod/forecast/), last updated October 7, and were used only as a cross-check and a robustness test, not as an input to the Base. The reference market price is $54.92, an intraday trade at 7:46 pm Prague time on October 20, 2026 during the regular session. The October 19 close was $54.00 on the [quote page](https://stockanalysis.com/stocks/inod/). Per share values use 35.5 million economic shares. The 2028 margin, the exit multiples, capital spending, the normalized owner cash, the discount rate and the terminal assumptions are analytical estimates, not company guidance. **Disclaimer.** This analysis is provided solely for research and educational purposes. It is not personalized investment, financial, legal or tax advice. Values are estimates, not forecasts or trading instructions. Innodata is exposed to customer concentration, project cancellations, a possible reversal of customer-funded cash, dilution from share issuance and compensation, and competition in AI data services. Investors can lose part or all of their invested capital. Method: Master Investment Analysis Framework v3.4. Approach: growth multiples on normalized 2028 owner cash and operating profit, cross-checked with a ten-year cash flow model, with customer-funded cash treated as a reserve. Confidence: C/15. Data checks: passed with warnings. Conditional research only. Open data gaps: the October cash balance, share sales under the ATM program and customer contract terms.