This is an interesting $AAOI update from @JonahLupton, but one thing in particular caught my attention: “Demand is not the problem. Capacity is.” That’s a very important distinction. If hyperscalers are already willing to sign LTAs and AAOI is saying it could win additional business if it simply had more capacity, then the $600M ATM starts to look less like “raising money because the business needs it” and more like raising capital to remove the bottleneck. And I think there’s another piece here that the market may be underestimating. AI infrastructure is becoming increasingly optical. Higher GPU density, faster networking and the transition to 800G/1.6T all mean more optical content moving through each generation of AI infrastructure. So AAOI doesn’t necessarily need to win some massive share of the AI market. It needs to keep increasing capacity while the amount of optical infrastructure being deployed keeps growing. That’s the part of the thesis I find most interesting. The numbers are obviously where this gets crazy: $1.1B revenue in 2026 Then $4B in 2027* *Potentially $4.5–5B if capacity ramps faster And potentially $8–10B in 2028 if the growth trajectory continues. This HUGE! But of course execution is the key risk, but if AAOI can actually build the capacity to meet this demand, I think the upside is enormous. And honestly, after connecting the dots, I completely agree with @JonahLupton’s $500–600 target over the next 18–24 months is very realistic. The market may be underestimating what happens when the bottleneck is not demand, but capacity. This isn’t just a $AAOI story either. $LITE and $COHR are saying essentially the same thing - everything they can produce is already sold out. And I think $AAOI will be the one to ramp capacity the fastest.