Photonics is the hidden backbone of AI. As GPU clusters scale to hundreds of thousands, the bottleneck is no longer compute—it's data transfer between chips. Optical interconnects solve this, and demand is accelerating non-linearly. Goldman Sachs sees a total addressable market of over $154 billion by 2028. Key companies to watch: $Applied Optoelectronics(AAOI)$ : Vertically integrated with in-house lasers, benefiting from hyperscaler demand. $Credo Technology Group Holding Ltd(CRDO)$ : High-margin growth, with optical revenue booming. $Lumentum(LITE)$ : A blue-chip supplier of lasers and transceivers. $COHERE

Since April 23rd, the leaders in photonics: Sivers → +190% LPKF → +47% Soitec → +42% $Applied Optoelectronics(AAOI)$  → +28% $COHERENT(COHR)$  → +12% $Lumentum(LITE)$  → +2% The thesis played out: the underappreciated chokepoints delivered the biggest gains. Now the debate is on $Applied Optoelectronics(AAOI)$  — clear upside, but is it still early or already priced in? What's your market cap expectation for $Applied Optoelectronics(AAOI)$ ? The tape is watching...

$Roundhill Memory ETF(DRAM)$ That NVDA article on Friday, which came out with the jobs report, said NVDA was cutting its memory requirements in half. The algos misunderstood and dumped memory stocks on that news. As we've had the weekend to digest, the real story is clear: for flexibility, NVDA is offering rack requirements at half the memory but doubled the number of racks. There might be some volatility at the open for a day or two, but I'm confident we'll quickly revisit and surpass where we dropped from. Additionally, Micron should give us a solid lift after earnings on the 24th. Ignore the short-term noise, maintain your conviction, and stay the course.

$Roundhill Memory ETF(DRAM)$ Honestly, it still feels early to me. A lot of people are acting like the memory trade is already crowded, but demand keeps getting bigger every quarter. Think about what's happening: AI servers need insane amounts of HBM, AI PCs need more LPDDR, and robotics plus autonomous driving need even more memory. This ETF gives exposure to the entire space. It feels like people still don't fully understand how massive this cycle could become. Memory demand isn't disappearing anytime soon. If anything, it feels like we're just entering the acceleration phase.

If you zoom out on the semiconductor sector, the message becomes pretty clear. WSTS projections point to a massive expansion cycle: 2026 semiconductor market around $1.5T, up about 90% YoY, heavily driven by memory. 2027 around $1.9T, another strong +27% YoY growth. Memory specifically is where things get extreme: 2026 around $800B, up about 250% YoY. 2027 around $1.1T, still showing +32% YoY growth. That's not a slow cycle; it's acceleration with supply constraints still in play. In that kind of environment, pricing power tends to shift sharply in favor of $SanDisk Corp.(SNDK)$ , $Micron Technology(MU)$ , SK hynix, and Samsung. Even optoelectronics is picking up speed into 2027, with growth re-ac

$Roundhill Memory ETF(DRAM)$ There we go. I called 60 and a lot of people thought I was off base. It only took a few decent down days. I'll be buying some today, adding to my long-term position.

$Roundhill Memory ETF(DRAM)$  Memory prices seem to be only getting higher. I noticed when I went to buy a new SSD that the price had doubled. That's why I'm investing in this area, because prices aren't coming down anytime soon.

$Lumentum(LITE)$ Looks like those 860 new additions are already performing well. Still feel fortunate to be long this incredibly exciting stock - haven't sold a single share.

$LightPath(LPTH)$ The stock continues to act very constructively after the offering. Shares effectively put in a double bottom today near $14.70 without even testing the $14 offering price itself—that's generally not the type of price action you see in weak secondary setups. To me, $16.65 now becomes the key pivot as the post-offering high. A convincing volume breakout above that level would likely open the door for a push toward the previous highs, as supply from the raise appears increasingly absorbed. Meanwhile, the broader strategic story around specialty IR materials, thermal imaging, vertically integrated optics, and domestic germanium alternatives still feels materially underestimated by the market. Sam Rubin also continues to drop inc

$Roundhill Memory ETF(DRAM)$ Micron earnings on June 24. It's another catalyst.

The year 2026 is widely expected to be a big one for optical and photonics picks-and-shovels investments. $Applied Optoelectronics(AAOI)$  is the small-cap, high-beta transceiver name with lumpy revenue but massive AI upside if demand holds. $Broadcom(AVGO)$  sits at the top of the stack, owning custom AI chips and networking, with its scale and pricing power already largely priced in. $Lumentum(LITE)$  is the steady optical components leader tied to datacenters and telecom, slowly recovering margins as demand ramps up. $Credo Technology Group Holding Ltd(CRDO)$  offers pure high-growth connectivity e

$Roundhill Memory ETF(DRAM)$  Goldman Sachs raised its 12-month price target for Samsung to 480,000 won on June 1 — implying roughly 33% upside from Tuesday's close of 360,500 won — and reiterated a Buy rating. They cited a structural shift in memory from a commodity cycle to a predictable "AI infrastructure" market. The bank forecast that supply constraints will persist at least through 2028. This news is great for those of us holding Samsung stock.

Everyone thinks the photonics run is over. I disagree. AI cluster buildout is just getting started. Every GPU farm needs insane amounts of data moving at light speed. Optical demand isn't growing linearly; it's exploding. $Applied Optoelectronics(AAOI)$  Owns a critical part of the supply chain with in-house lasers. Their 2026 revenue guidance is $1.1B. Hyperscalers are lining up. $Credo Technology Group Holding Ltd(CRDO)$  Revenue has tripled, with 68% gross margins. The optical segment alone is projected to be over $600M next year. Analysts are raising their targets. $Lumentum(LITE)$  A blue-chip photonics player with decades of expertise. Probab

$Roundhill Memory ETF(DRAM)$  Overbought? Consider, Micron has a PEG of 0.08. SanDisk's is even lower at 0.04 (stock charts numbers). Room to run!

$Applied Optoelectronics(AAOI)$ 198 again? All the photonics were ripping AH Then money rotated to semis cause of Computex comments We not done $Lumentum(LITE)$  $COHERENT(COHR)$ 

$Kopin(KOPN)$ $Applied Optoelectronics(AAOI)$ $POET Technologies Inc(POET)$ $Lightwave Logic, Inc.(LWLG)$ $Lumentum(LITE)$ KOPN has 150,000 chips of AI optical capacity, potentially generating hundreds of millions in revenue per quarter. Microsoft was already researching MicroLed-based optical connectivity last year. That sounds promising. According to Microsoft, the advantages of MicroLed-based optical connectivity include: long range, low power consumption, low cost, and high reliability.

$Roundhill Memory ETF(DRAM)$ Samsung was up 3.3% in Korea last night, DRAM owns shares.

$Lumentum(LITE)$  isn't for the fainthearted, but it always seems to find a way to have huge up days that get the momentum headed back in the right direction.

$BigBear.ai Holdings(BBAI)$ is surging lately! The AI boom keeps driving it higher, pushing the stock up this rebound. Short-term gains look solid—adding on the dip and staying bullish on its long-term potential!

$Circle Internet Corp.(CRCL)$ has successfully built a solid base with significantly higher trading volume, and it's firmly standing at a turning point! Made some decent gains on this move—feeling like it's only a matter of time before it climbs back to $300. Plus, the recent Contra688 event is just awesome—free top-up vouchers really sweetened the holding cost!