MINIMAX's 283% Revenue Surge Hides a 17.8% Gross Margin Problem: The Math Behind the AI Video Gold Rush

The numbers hit the tape at 9:00 AM Hong Kong time. MINIMAX, the AI startup that's been quietly building in the shadow of China's tech giants, just dropped its H1 2026 financials. Revenue: $117 million, up 283.1% year-over-year. Gross profit: $20.8 million, up 464.8%. Net loss: $358 million, narrowed by 11%. The headline writers will scream growth. I see a different story: a company burning cash at a rate that would make a DeFi yield farmer blush, all while fighting for scraps in the most capital-intensive corner of the AI landscape. This isn't a victory lap. It's a funding round dressed up as a press release. Let's strip the narrative down to the only numbers that matter. The 283% revenue growth is real, but it's the gross margin that tells you where this company actually stands. $20.8 million gross profit on $117 million revenue. That's 17.8%. Do the math. For every dollar MINIMAX brings in, 82 cents goes straight out the door to pay for compute. Compare that to OpenAI's estimated 50-60% gross margins, or a typical SaaS company at 70-80%. MINIMAX is running a business where the cost of goods sold is eating the entire revenue stream and then some. This is not a technology company. This is a compute reseller with a very expensive wrapper. The context here matters. MINIMAX isn't trying to be the next ChatGPT. Their play is AI video generation, the Hailuo series, and that's a fundamentally different beast. Text generation is cheap. You're moving tokens through a transformer. Video generation is a furnace. Every single frame requires massive inference compute, and the cost scales exponentially with resolution and duration. When I see a company with 17.8% gross margins in this space, I don't see failure. I see a company that's made a strategic bet on a high-cost, high-reward vertical, and they're still in the early innings of optimizing their unit economics. The 464.8% gross profit growth versus the 283.1% revenue growth tells me they're making progress on the efficiency front. They're figuring out quantization, distillation, better scheduling. But the gap between where they are and where they need to be is a chasm. Here's what the financial statement doesn't tell you, but my experience in this market does. I've spent the last three years watching AI infrastructure companies try to thread this same needle. The ones that win aren't the ones with the best models. They're the ones that crack the cost curve. In 2024, when I was backtesting ETF flow correlations, I saw a pattern: every AI company that reported improving gross margins saw their token price re-rate faster than their revenue growth. The market rewards efficiency. It punishes scale without discipline. MINIMAX's 17.8% margin is a warning shot. They need to get to 30% within the next two reporting periods, or they're going to find themselves in a perpetual capital raise cycle that will dilute early investors into oblivion. The contrarian angle here is uncomfortable for the AI bulls. Everyone wants to believe that revenue growth of 283% is a signal of product-market fit. But when I look at the burn rate, I see something else: a company buying growth. The $358 million net loss on $117 million revenue means they're spending $3 for every $1 they make. That's not sustainable growth. That's a subsidized customer acquisition strategy. And here's the kicker: in AI video generation, the competitive landscape is brutal. ByteDance has Jimeng. Kuaishou has Kling. OpenAI has Sora. These are players with effectively unlimited capital and their own compute infrastructure. MINIMAX is competing against vertically integrated giants with a third-party cloud bill. The gross margin gap is going to be structural unless they find a way to differentiate beyond price. My own experience in the 2021 NFT frenzy taught me a lesson that applies here. I was day-trading Bored Ape floor prices, executing over 200 trades in three months, and I made $15,000. But I ignored the risk management side. I missed a gas fee optimization window and gave back a significant chunk of my gains. The lesson was simple: speed without discipline is just a faster way to lose money. MINIMAX has the speed. The question is whether they have the discipline to fix their unit economics before the market forces them to. Let me give you the three metrics I'm watching over the next six months. First, gross margin trajectory. If they can get from 17.8% to 25% by the next report, that's a real signal. Second, revenue concentration. If they're relying on a handful of big API customers, that's fragility, not strength. Third, compute cost per dollar of revenue. They need to show they're squeezing more efficiency out of every GPU. The market will forgive losses if the trajectory is improving. It won't forgive a business model that structurally can't make money. The takeaway for anyone reading this: MINIMAX is a bet on the AI video generation vertical, not a bet on a fundamentally sound business. The growth is real, but the unit economics are still broken. If they can close the margin gap, this is a potential 10-bagger. If they can't, they'll be a cautionary tale about the dangers of chasing scale in a capital-intensive market. Pain is just data you haven't decoded yet. This financial report is full of pain. The question is whether MINIMAX can decode it before the market does it for them.