@investmattallen $DGXX Digi Power X — $541M market cap. 10-yr Cerebras colo deal: 40 MW in Alabama, $1.1B initial, up to $2.5B with renewals. 195 MW pipeline target. Anchor tenant Cerebras IPOs Thursday as $CBRS at ~$48B. Q1 results May 15.
$MXL — the next $SNDK?
The setup looks familiar. Sleepy semi name, sudden AI rerating, Q1 Infrastructure +136% YoY (now their #1 segment), stock ~$20 → ~$77 in 3 weeks, 5+ analyst upgrades inside 48 hours.
The products actually shipping into AI: Keystone (800G optical DSP, millions shipped), Rushmore (1.6T DSP, ramps late '26), Annapurna (224G copper retimer for GPU-to-GPU links), Washington (200G/lane TIA, just launched Apr 30), Panther V (storage accelerator for AI inference, at Dell Tech World this month).
$SNDK rerated on AI storage demand. Could $MXL be setting up to do the same on AI optical interconnect demand?
🚨 $LITE just printed a monster Q3. Huge read-through for $MXL.
Q3 numbers: Rev $808 M (+90% YoY, record), EPS $2.37 vs $2.26 est, 9th straight beat.
The line from $LITE CEO Michael Hurlston that matters for $MXL: "We appear to be ahead on 1.6T… our constraint is on transceivers… we could ship quite a bit more." → 1.6T transceiver demand is supply-constrained. $MXL is one of three DSP suppliers ($MRVL, $AVGO, $MXL) competing for the socket inside those transceivers — Rushmore is their entry. (DSP = digital signal processor, the chip that cleans up the optical signal so it arrives intact at multi-gigabit speeds.)
$LITE makes the laser. $MXL makes the DSP. They're both content suppliers into 800G / 1.6T modules.
And $MXL has been pushing hard into the 1.6T + AI inference story just in the last week:
Apr 30: Washington 200G/lane TIA available — the analog front-end chip that pairs with any 1.6T DSP (Marvell, Broadcom, or $MXL itself). Mass production 2H'26.
May 5: Panther V storage accelerator at Dell Tech World — purpose-built silicon for AI inference / RAG architectures, $5 B serviceable market per management.
Macro: top-5 hyperscaler '26 capex ~$705B (+67% YoY), 12–18 month visibility (unprecedented for optical), $NVDA put $2B into $LITE in March.
$LITE went from $63 → $1,000 in 12 months on this exact thesis. $MXL just printed +136% YoY infra growth and is one cycle behind.
Why it matters for AI: GPUs in modern AI clusters need to talk to each other at insane speed. The "wires" are fiber optics — data gets converted to light by a laser, sent across the data center, then decoded at the other end. 1.6T = 1.6 trillion bits per second per cable, double today's 800G standard. As AI models scale, the network becomes the bottleneck before compute does. Lasers ($LITE) + DSPs ($MXL / $MRVL / $AVGO) are the picks-and-shovels of the AI build-out.
AI-assisted | NFA | DYOR
@HyperTechInvest One more worth adding: $AIXA (Aixtron).
As AI data centers shift to faster optical links, demand for III-V lasers rises. Aixtron makes the tools many suppliers use to build those lasers, so it benefits early in the chain — upstream of $LITE, $COHR, and others.
@OpenMedFuture@AdrianP_doc Belgium says “all EU states,” even though everyone knows you will never get real solidarity from Hungary. So it feels like Belgium is adding this roadblock on purpose.
【NAND Spot Market in Chaos: Prices Adjusted Daily as Sellers Renege on Deals】
Frequent adjustments of quotations within a single day have been observed from the spot market under the expectation of exceedingly constrained supply and increasing prices. Some of the sellers, who are pursuing for an even higher profitability, are denying concluded prices confirmed previously at the cost of their reputation, and requesting new price negotiations, which fully reflects the increasingly intense level of panic buy orders within the spot market. Spot prices of 512Gb rose by 5.90% this week, arriving at US$9.607.
📷 Related news: https://t.co/d1tYF6STD7 🔗
📷 For more on NAND Price: https://t.co/2tMXrV83Is 🔗
@crux_capital_ Do you think the tight supply of InP lasers also means companies like $aixa, who make the tools for laser factories, could see more demand?