Sea SE 0.00%↑ really did pick about the worst possible time to announce its cloud ambitions…
The AI trade is getting questioned, investors are arguing about whether all this spending will ever earn a decent return, and here comes Sea telling us it wants to sell cloud services.
Reading Forrest Li’s October 20 letter, though, I can see why they want to try. Sea has been running its own cloud for more than a decade to support Shopee, Garena and Monee. Roughly 500 employees have spent the past two years rebuilding that platform for AI. They now want to offer it to outside companies through Sea Cloud.
At NVIDIA’s AI Day in September, Sea disclosed roughly 25,000 GPUs across a network of more than 40 data centers. Much of that capacity is there to support its own businesses, but it gives you some idea of how much experience and infrastructure they’re bringing to this.
The appeal, to me, is being able to earn more from money they were going to spend anyway. Shopee is going to need computing capacity as it grows. So are the financial services and gaming businesses. If outside customers help spread the cost of the servers, software and people across more revenue, I can see the economics working.
I do wonder how much capacity they have available when their own businesses are still growing, though.
Forrest didn’t give a spending budget in the letter. Sea’s cash purchases of property and equipment were already $495 million in the first half of 2026, up from $216 million a year earlier.
Those figures cover the whole company, including logistics. What I’d like to hear is how much faster spending needs to grow because they’ve decided to sell cloud to other people.
I’m watching that spending closely because I already like SE without a cloud business. At roughly 10–11x estimated 2027 free cash flow, with revenue forecast to grow more than 20% next year, I think the stock looks very cheap. I’d happily own it for Shopee, Monee and Garena alone. I’m not giving Sea Cloud any credit in my bull case.
