I am sitting on Innodata (INOD) right now. The company delivered the kind of Q1 that makes analysts rewrite their models. The problem is the stock has already moved as if every remaining quarter will beat too.
What happened
Innodata reported record Q1 2026 revenue of $90.1 million on May 7 - up 54% year-over-year and beating consensus by roughly 18%. Adjusted gross margin hit 47%. Adjusted EBITDA came in at $25 million, also above estimates. Free cash flow for the quarter was $34.8 million. Management raised full-year 2026 revenue guidance to approximately 40% or more, up from the 35% floor they set just ten weeks ago.
The stock responded the way breakout names do. It had been sitting around $42 before the print. Within weeks it surged past $97, briefly touching $100 before settling back. A near-doubling in under a month. The P/E ratio jumped to about 86 times earnings, well above its 12-month average of roughly 49.
The growth is real - that's not the argument
Innodata cleans and labels data for AI model builders. It's a "picks and shovels" service in an AI infrastructure boom where the model companies don't want to do this work themselves. The company has won LLM programs from multiple big tech clients, including a $44 million annualized run-rate expansion announced last year. Q1 non-primary big tech revenue grew 453% year-over-year, which shows the customer diversification thesis is starting to work.
This is not a company pivoting to AI on narrative alone. The AI demand is showing up in revenue, margins, and cash flow. All at once. That combination is uncommon enough to respect.
The argument is about what $3.1 billion is buying you now
Here's what changed between the $42 zone and $97. The business didn't. The multiple did. At $42, the stock was trading at roughly half its current earnings multiple and offered margin for error if one of the next two quarters stumbled. At $97, the valuation has already absorbed four-plus quarters of flawless 40%+ growth, continued customer diversification, and expanding margins.
That matters because this is still a company where the largest single customer represented approximately 61% of revenue as recently as Q1 2025. Management says that share is declining - and the 453% spike in secondary big tech revenue supports it - but concentration risk doesn't disappear because growth is fast. If one of those big tech clients shifts internal priorities, renegotiates pricing, or pulls back on a model training cycle, a meaningful chunk of that 40% guidance comes off the table.
The Q1 print was exceptional, but a single quarter doesn't de-risk a structural dependency. It just gives the stock momentum.
The valuation problem in one metric
An 86x P/E on a $3.1 billion market cap requires the growth story not just to continue, but to accelerate without interruption. At that multiple, the market is pricing in near-perfect execution through 2027 and beyond. Any quarter that shows deceleration - even a modest one - will compress the multiple and drop the stock hard. That's the mechanical reality of trading at 86 times earnings when you were at 49 six months ago.
For context, fast-growing AI-adjacent software names typically command 40x to 60x forward earnings when the growth rate and risk profile justify paying up. INOD is asking investors to accept roughly double that premium.
What would make me buy
If the stock pulls back to the $55 to $60 range, the math starts working again. That would compress the multiple back toward a level where 40% revenue growth justifies the entry price, even if customer concentration takes another quarter or two to meaningfully decline. At that level, the risk/reward resets in the investor's favor.
What would make me sell a current position
A quarter showing revenue growth decelerating below 30%, or any evidence that big tech spending on AI data services is plateauing. At the current multiple, INOD cannot afford a slow quarter without a painful re-rating.
Bottom line
Innodata is a legitimate AI infrastructure play with accelerating revenue, expanding margins, and real customer diversification. The problem isn't the business. It's the stock. The rally has moved the price ahead of the proof. At nearly $100, there's not enough margin for error to justify a new Buy. Hold existing positions, wait for a pullback, and watch Q2 guidance carefully when it arrives in August.
Rating: Hold