SK HYNIX: - HAS SIGNED LTAS WITH 14 MAJOR CUSTOMERS, INCLUDING LARGE GLOBAL CSPs. - THE STANDARD CONTRACT TERM IS FIVE YEARS, WITH DISCUSSIONS UNDERWAY WITH SOME CUSTOMERS TO EXTEND EXISTING CONTRACT TERMS. - PRICING GENERALLY REFLECTS MARKET CONDITIONS WITHIN A RANGE AGREED UPON WITH EACH CUSTOMER. - THE AGREEMENTS INCLUDE DEPOSITS AND OTHER MECHANISMS TO ENSURE CONTRACT FULFILLMENT. - VOLUMES CURRENTLY SECURED UNDER LTAS ACCOUNT FOR MORE THAN 50% OF TOTAL CAPACITY IN EACH OF 2027 AND 2028.

My favorite topic: LTA Samsung, SK Hynix, and Micron are signing 3- to 5-year supply deals. Micron has gone further with take-or-pay “strategic customer agreements.” Nanya already has about half of capacity under contracts of various lengths. Samsung is pushing new deals to at least three years, some to five. It has multi-year pacts with the top five data-center operators and wants about two-thirds of memory output under long contracts, some with price floors and prepayments. SK Hynix has finished LTA talks with about ten core customers. The new deals tie volume to product roadmaps so it can plan DRAM and HBM lines earlier. Micron has 16 SCAs running from 2026 through end-2030. Auto deals are often three years. Take-or-pay means the buyer owes a minimum even if it takes less. Those pacts already cover about 20% of DRAM shipments and about one-third of NAND. It wants at least half of revenue inside SCAs. Nanya’s mix runs from one quarter to several years. SanDisk and Kioxia also set long DRAM supply after taking stakes. The twist: even if makers add capacity in 2027 and 2028, much of the new output may already be spoken for. Spot supply could stay tight longer than an old cycle. $MU $SKHY $SNDK If two-thirds of bits sit under multi-year deals, does the next downturn even show up on the spot tape?

Let's talk about LTA's / pricing for Photonics $LITE $AAOI $GLW $COHR etc. Historically a lot of this business was pretty tactical. Customers bought quarter to quarter, suppliers fought for share, and price-downs were part of the game. This is not the case anymore due to the supply being this scarce. $LITE : Mgmt said they basically never used to have LTAs like this. Now all of their EML capacity is spoken for through 2027, and customers that commit get supply priority. They also said LTAs have helped pricing because they no longer have the same quarterly negotiations for price cuts. “If anything, prices are holding or they’re increasing.” And when customers come back asking for MORE capacity than they originally contracted that opens the door for another pricing discussion. $COHR : Their LTAs can include > multi-year volume commitments > pricing commitments > minimum demand guarantees > take-or-pay structures And many of these agreements extend through the end of the decade. $GLW: Says essentially every major optical capacity expansion is now backed by these kinds of LTA's, and eventually they expect this structure to represent the “lion’s share” of the optical business. So this landscape is increasingly becoming: Customer commits volume → supplier builds capacity → supply gets reserved → pricing gets protected.

$AAOI earnings: - $1B-$1.5B LTA disclosed with a hyperscaler customer - 2nd LTA coming soon, worth 5x+ more (implying $5B-$7.5B), in its final stages. This LTA explicitly includes lasers, very material for AOI that they're moving to supplying merchant lasers. - "The demand is exceptional, unprecedented. I do not know what kind of superlatives to use to describe the amount of demand that we are seeing out there in the industry" Yeah this commentary doesn't sound like any optical demand would slowed down if it's "exceptional + unprecedented" - Rejected hyperscaler slowdowns from political changes with little effect. - AOI Can only meet 80% of demand, in case there's a kind of reduction, the demand would be bigger than AOI future capacity - Moving toward other financing methods than ATMs (thank god), US bank loans similar to what $NBIS is doing, I'd argue is positive, given FCF coming in later. - Delay in meeting capacity projections into H1 2027. $471m/month targets pushed back into H2 2027, due to electrical procurement and construction logistics, which existed before Texas governor changed. This part is negative part of the earnings call. To qualify this, I don't quite think hyperscalers would sign a $5-$7.5B LTA if they didn't think AOI couldn't deliver. - Near term Q4 guidance cut was DSP availability. AOI revised near term projections down slightly to $385M and $420M in revenue. Said DSP supply would ease early 2027 and that they're qualifying more suppliers. - Gave hiked H2 2027 capacity projections from their expansion, near term revision downward. - 4 to 5 ongoing UHP laser eval/quals for CPO UHP laser segment alone would far outweigh their $3.8B MC, especially if you're executing on multi billion dollar contracts with 60%+ gross margin laseres. TLDR: - Long term visibility improved a lot, with est. $6-$9B cumulative LTA contracts coming soon. - Shift toward supplying CPO lasers merchant is fundamentally material vs. just transceivers, and what I wanted to hear - Capacity guidance hiked into H2 2027, near term delays from electrical/DSP related procurement + installation. I still think we're reached peak fear levels, but wouldn't dismiss disappointing short term procurement delays + Q4 DSP bottlenecks. AOI likely signing an additional +$5-$7.5B in likely hyperscaler LTAs and entering merchant UHP CPO market dominanted by $LITE + $COHR. Factored in with the valuation drop has made medium-long term opportunities very compelling.

SK HYNIX SAYS IT HAS SIGNED LTAS WITH 14 KEY CUSTOMERS, INCLUDING MAJOR GLOBAL CSPS, ON A STANDARD 5-YEAR TERM, HAS SECURED LTA VOLUMES OVER 50% OF TOTAL CAPACITY FOR 2027 AND 2028, AND EXPECTS 2027 CAPEX TO INCREASE SIGNIFICANTLY COMPARED TO 2026