On May 8th, 2026, Innodata Inc. (NASDAQ: INOD) did something that most small-cap stocks only dream about. It nearly doubled in a single session. Not on a rumor. Not on a buyout bid. On earnings.

The stock closed up roughly 92% after the company reported its Q1 2026 results the evening before. If you had been watching this one over the past month, you were watching a stock that had already been building momentum. The Q1 report just lit the fuse.

What Happened

Innodata posted Q1 2026 revenue of $90.1 million, up 54% year-over-year. That number beat analyst consensus of $76.5 million by approximately $13.6 million, or about 18%. Not a small miss in the other direction. An 18% beat on the top line for a company this size is a significant gap between what the street expected and what actually showed up.

But the revenue beat was only part of the story.

Adjusted EBITDA came in at $25.0 million, or 28% of revenue. That number beat analyst consensus of $10.4 million by 139%. Diluted EPS hit $0.42 against a consensus estimate of just $0.08. The company ended the quarter with $117.4 million in cash and short-term investments, up $35.1 million sequentially from $82.2 million at year-end 2025. No debt drawn on a credit facility that was actually expanded during the quarter from $30 million to $50 million on a fresh three-year term with Wells Fargo.

That is a clean balance sheet. Rare in this space.

What Innodata Actually Does

Here is where a lot of investors get tripped up. Innodata is not building the large language models. It is not selling chips to run them. What it does is arguably more foundational to the whole operation: it cleans, structures, labels, and engineers the raw data that goes into training these models in the first place.

Think of it this way. Before any frontier AI model gets to learn anything useful, someone has to transform messy, unstructured text, images, and multimedia into clean, machine-readable formats. That is Innodata’s core business. The company has been doing some version of this for over 36 years, originally in document processing and publishing. The AI wave essentially turned its legacy skills into one of the most in-demand capabilities in enterprise technology.

Slight tangent, but it matters: this is exactly why data engineering companies like INOD attract less attention than the GPU plays or the hyperscalers. They are not running the AI. They are the reason the AI runs well. The distinction is starting to show up in revenue growth rates.

The company operates across pre-training data, post-training alignment, model evaluation, trust and safety, and is now moving into agentic AI systems. It is also expanding into government work, having secured a prime contract position under the Missile Defense Agency’s SHIELD program, and generating traction with Palantir on computer vision tasks in the federal space.

The Numbers That Matter

• Q1 2026 Revenue: $90.1 million (+54% YoY, +24% sequentially from Q4 2025’s $72.4 million)
• Revenue vs. Consensus: Beat by $13.6 million, or 18%
• Adjusted Gross Margin: 47% – six percentage points above Q4 and seven points above the company’s own 40% public target
• Adjusted EBITDA: $25.0 million (28% of revenue), nearly doubling from $12.7 million in Q1 2025
• Net Income: $14.9 million vs. $7.8 million in Q1 2025
• Diluted EPS: $0.42 vs. $0.22 in Q1 2025
• Operating Cash Flow: $37.3 million
• Cash Balance: $117.4 million at March 31, 2026
• Full-Year 2025 Revenue (for context): $251.7 million (+48% YoY)
• 2026 Revenue Guidance (raised): ~40% or more YoY growth, up from 35% or more guided just ten weeks prior

The margin expansion story is actually the part that should be getting more attention. An adjusted gross margin of 47% is well above what most labor-intensive data services companies can deliver. The fact that it is expanding while revenue is simultaneously accelerating suggests the company is capturing operating leverage as its largest programs scale.

The New Contract Win and Customer Diversification

One of the more compelling disclosures in the Q1 report was a new set of engagements with a major Big Tech customer expected to generate approximately $51 million of 2026 revenue. Twelve months ago, that customer contributed zero revenue to Innodata. It is now expected to become the company’s second-largest account by year-end.

What is interesting here is how management described the diversification dynamic. Revenue from the company’s other Big Tech customers in aggregate grew 453% year-over-year in Q1. The largest customer is still growing in absolute dollar terms, but at a slower rate than the rest of the base. That is customer diversification working in the best possible direction for a company trying to reduce concentration risk.

That said, the 10-Q filing is worth reading carefully. As of Q1 2026, one customer still represented 56% of revenue and another 17%. That is a combined 73% from two clients. The diversification is real and moving in the right direction, but it is not finished yet.

Bull, Base, and Bear

Bull Case: AI model training demand continues to expand into multimodal, agentic, and physical AI use cases. Innodata’s $51 million new Big Tech engagement scales further. Federal contracts begin generating meaningful revenue. Adjusted gross margins hold at or above 47%. The company exits 2026 with annualized revenue approaching $400 million and further cash generation, with the balance sheet remaining debt-free. Analysts currently have a consensus price target around $93 to $95 on this name, with at least one firm at $111.

Base Case: Full-year 2026 revenue growth lands at the guided 40% floor. Margins stabilize in the 44–47% adjusted gross margin range. Customer diversification continues but concentration remains elevated through most of the year. The stock consolidates after its parabolic move and trades in a range as investors wait for Q2 confirmation of the trend.

Bear Case: The two largest customers pull back or delay programs. Margins compress as the company invests more heavily in headcount and platforms to support growth that does not materialize on schedule. The labor-intensive nature of data annotation limits how fast the business can scale without proportional cost increases. The Q1 results may reflect front-loaded contract activity that does not repeat uniformly across the year.

Cheap Investor Scorecard

• Revenue growth above 40% YoY for full-year 2026: On track per raised guidance
• Adjusted gross margin sustaining above 44%: Currently 47% – above target
• Adjusted EBITDA margin expanding toward 30%: At 28% in Q1, expanding
• Cash balance growing without debt: $117.4M cash, zero drawn on credit line
• New Big Tech customer delivering $51M in 2026 revenue: Announced, in progress
• Top customer falling below 50% of total revenue: Not yet – still at 56% in Q1
• Federal revenue becoming material: Early-stage, SHIELD contract awarded
• Diluted EPS continuing to grow: $0.42 in Q1 vs. $0.22 a year ago
• Agentic AI platform gaining paying customers: First $1M engagement closed post-launch
• Analyst price targets converging above current price: Consensus near $93–$95; high target $111

Action Plan

If you missed the 92% single-day move, the honest answer is that chasing it the day after is a different risk profile than owning it into the Q1 report. The stock has already priced in a significant amount of the Q1 upside. What matters now is whether Q2 confirms the trend or stalls it.

For those with conviction on the AI data engineering theme and tolerance for concentration risk, a scaled entry approach makes sense here. Consider a partial position on any meaningful pullback toward prior resistance levels, with a second tranche reserved for Q2 confirmation. Full-position sizing into a stock that just doubled in a day is a discipline question as much as a valuation one.

If you already own it: the raised guidance, expanding margins, and new contract wins give you a reasonable fundamental reason to hold. The risk is not the business right now. The risk is the multiple you are paying for it after a near-doubling, and whether the next quarter can keep pace with what Q1 just delivered.

What I keep coming back to is the cash flow. $37.3 million in operating cash flow in a single quarter, no debt, and a company that just delivered a quarter exceeding its own annual revenue from three years ago. The fundamentals have changed. Whether the stock has already reflected all of that change is the question nobody can fully answer yet.

Worth a closer look.

– The Cheap Investor