The numbers hit the terminal like a flash crash in reverse. MINIMAX, the Chinese AI video generation powerhouse, just reported revenue of $117 million for the first half of 2026. Up 283.1% year-over-year. Gross profit surged 464.8%. The headline writes itself: another AI unicorn galloping toward the promised land.
But I've spent the last decade dissecting financial statements that are engineered to deceive. And this one has a smell. A 17.8% gross margin. A net loss of $358 million—three times the revenue they generated. This isn't a growth story. It's a capital incineration event dressed in a bull-market costume.
Let me be clear about what I'm seeing. The market will celebrate the top-line growth. The sophisticated money will read the footnotes. And the footnotes are screaming that this business model, as currently constructed, is a high-performance vehicle with a fuel leak the size of a fist.
The Context: A Race to the Bottom Paved with GPUs
MINIMAX sits in the most capital-intensive corner of the AI landscape: video generation. This isn't text prediction. This is the heavyweight division. Generating a single minute of high-definition video requires thousands of GPU inference calls. The compute bill for a company like this isn't an operating expense—it's a existential line item.
The company's differentiation was supposed to be its multi-modal prowess, particularly its Hailuo video generation series. In a market where OpenAI's Sora and Google's Veo are the theoretical benchmarks, MINIMAX has been the scrappy challenger shipping actual products. Their growth suggests product-market fit. Their margins suggest they're paying for every dollar of that fit with two dollars of compute.
This is the classic "scale at all costs" playbook. And in a bull market, it works. Capital is cheap. Investors are chasing the next frontier. But the underlying physics of the business haven't changed. Video generation is a commodity service with a premium veneer. The moat isn't the model—it's the cost structure. And MINIMAX's cost structure is currently a sieve.
The Core: Forensic Accounting for the Decentralized Age
Let me walk you through the math that matters. Revenue of $117 million. Gross profit of $20.8 million. That's the 17.8% margin. For context, a mature SaaS company runs at 70-80% gross margins. Even a capital-heavy cloud provider like AWS operates above 60%. MINIMAX is running at a level that suggests they're paying retail prices for their most critical input: compute.
This is the hidden grid where value leaks out. The 464.8% growth in gross profit sounds impressive until you realize it's growing off a tiny base. The absolute number—$20.8 million—is what you'd expect from a mid-sized B2B software company, not a frontier AI lab with a $358 million burn rate.
The loss structure is even more revealing. A $358 million net loss on $117 million in revenue means they're spending over $4 for every $1 they bring in. This isn't a company approaching profitability. It's a company that's still in the "prove the market exists" phase, except the market has already been proven. The question is whether they can build a business around it before the capital spigot closes.
Based on my audit experience with high-burn tech companies, I can tell you what's happening here. The revenue growth is real, but it's purchased. They're likely underpricing their API access to win enterprise contracts. They're probably offering aggressive free tiers to build consumer habit. And they're almost certainly paying premium prices for GPU capacity because they can't access the most advanced chips directly.
This last point is critical. MINIMAX is a Chinese company. The US export controls on advanced semiconductors mean they're either relying on domestic chips like Huawei's Ascend line or paying through the nose for indirect access to NVIDIA hardware. Either path is a competitive disadvantage. The friction is where the opportunity hides—but in this case, the friction is their cost structure, and it's not an opportunity for them.
The Contrarian Angle: The Real Story Is the Capital Structure, Not the Technology
Everyone will focus on the technology. The model quality. The video generation capabilities. The race against Sora. That's the narrative the company wants you to follow. It's a distraction.
The real story is that MINIMAX is a capital arbitrage play. They're converting investor dollars into market share at a terrible exchange rate. The question isn't whether their technology is good—it clearly is. The question is whether they can achieve unit economic breakeven before their cash reserves hit zero.
At a $7 billion annualized burn rate, they need either a massive new funding round or a dramatic improvement in margins within the next 12-18 months. The window is closing. And here's the uncomfortable truth: their competitors with deeper pockets—the ByteDances, the Kuaishous, the Tencent-backed labs—can sustain this exact same strategy for longer. MINIMAX is fighting a war of attrition with less ammunition.
The bull case is that they become the "infrastructure layer" for AI video. The bear case is that they become a cautionary tale about the difference between revenue and value. I've seen this movie before. In the DeFi summer of 2020, I watched protocols with astronomical total value locked and zero sustainable revenue get valued like they'd solved banking. The ones that survived weren't the ones with the best technology. They were the ones with the most patient capital.
MINIMAX's technology is real. Their growth is real. But their margin structure is a warning sign that the market is mispricing the sustainability of their business model. Speed is the only moat when the gate opens—but this gate is opening into a field where the cost of entry keeps rising.
The Takeaway: Watch the Margin, Not the Headline
The next 12 months will determine whether MINIMAX is a legitimate AI powerhouse or a cautionary tale. The signals to watch aren't the revenue numbers—those will look great. Watch the gross margin. If it starts climbing toward 30-40%, they're solving the compute problem. If it stays in the teens, they're just buying growth with investor capital.
Watch their next funding round. The valuation will tell you whether sophisticated investors believe the unit economics can improve. And watch for strategic partnerships with cloud providers or chip manufacturers—those would signal they're locking in cost advantages.
I'm not saying MINIMAX is a fraud. I'm saying the market is looking at the wrong numbers. The revenue growth is the bait. The margin is the hook. And right now, that hook is buried deep in the side of a company that's bleeding cash at a rate that would make most traditional businesses insolvent within a quarter.
The question isn't whether AI video generation is the future. It is. The question is whether MINIMAX can survive long enough to be part of that future. And based on these numbers, that's a coin flip at best. Mapping the invisible grid where value leaks out—that's my job. And the grid here is leaking faster than the revenue can fill it.
