Monday, November 9, 2026

@convequity Oct 19 7 clicks

Convequity on X

Nvidia ($NVDA) faces a major issue: most AI compute is moving to inference, where its CUDA moat protects less. CUDA's programming flexibility is essential for training, which is complex and changes rapidly. Inference runs a trained model with fixed weights, and we think inference chips are easier to build than training chips. The mix is shifting. Post-training RL, reasoning tokens and final replies are all forms of inference. On our view, pre-training could fall below 20% of compute in the future. In February, Cursor said the compute used to post-train its Composer 1.5 model surpassed the compute used to pre-train the base model. At Convequity we think CUDA's depth now protects a shrinking share of the workload, and we expect that to continue. Deeper dive available at Convequity.

@aleabitoreddit Oct 19

Serenity on X

Feels like the public interpretation is all wrong… I’m pretty sure $MSFT backing out on NScale deal is bullish for Neoclouds like $NBIS or even $IREN with energized capacity with Microsoft? Pretend Microsoft just ate a bunch of Chipotle burittos: - They order toilet paper delivery scheduled Q1 2027 - NScale says: sorry 3-4+ year delay of the toilet paper (into 2030’s) because of UK grid constraints - Microsoft still urgently needs toilet paper so they will get it elsewhere The cancellation does send warnings about what happens if a Neocloud can’t execute on contracted demand. But that just makes the other toilet paper sellers more valuable since Microsoft will go elsewhere for supply.

@aleabitoreddit Oct 19

Serenity on X

I don't think it's unclear, but I do think segments outside of photonics are noise for what drives valuations (which I don't think management understood). Next earnings, $SIVE should use laser capacity revenue ceiling from foundry allocations + existing fabs as the core metric alongside customer qualification diversity. Not defending what the definition of a revenue pipeline is and using current term cellular AllSpace contract revenue as an example. Potential volume from $GFS probably dwarfs revenue from other segments. Talking about existing qualification cycles is enough since investors how they convert given 1.6T/NPO/CPO transitions are coming up.

@finphysnerd Oct 19 6 clicks

Myles on X

Somebody is absolutely puking $CTOS here. Down 14% on no news today and 8% in the last hour. The cybersecurity breach is clearly a concern, but realistically I don't see it affecting long term earnings significantly. I'm steadily buying the dip.

@aleabitoreddit Oct 19 1 click

Serenity on X

Last major $SIVE disclosure was 100M CW DFB laser capacity Q4 2027. And a new foundry partner (that makes 2 including Win Semi). Expected $JBL revenue coming in H1 2027 (with lasers for Jabil's 1.6T LRO scaling up H2 2027). There's channel checks along the way like CIOE where Sivers was engaged with Chinese pluggable makers + laser ASPs being raised. 70mW/100mW ranges were the most tight to Chinese pluggable makers, which was surprising. Biggest industry read through was $LITE selling out into 2029, making agreements for 2030, and having hyperscalers pay for capacity expansion for 2031. I don't think anyone has enough information to do any sort of accurate modeling for $SIVE like you can for $AAOI given they don't release product breakdown + capacity targets by quarter. The 200mW NPO lasers Sivers is sampling is likely different ASP than your 70mW lasers shipped to pluggable makers. And your custom CPO lasers are different pricing too. Like all sampling/qualification players from $AEHR to $AAOI back in 2025, usually the risk/reward comes in whether qualifications convert and what new customers they have, which is always a risk. In a bull case, if $SIVE comes online with 300m laser supply (2:1 implied ratio, already has "tremendous" supply online from likely US fab) during a structural shortage. Pluggable players like $JBL, 6 other vendors. $GFS as the reference design laser for their CPO/NPO platform. Your minor players vendors like $POET, $AEVA and other players end up scaling. And all the other custom products they hinted they were working on in the past for CPO years back. I think the probability that they cook is higher since there's so many ongoing qualifications during a bottleneck. We've already seen ownership changes from Swedish investors to US/International retail from $SIVE shareholder disclosures. My guess is that trends institutional from retail before US listing next year amid the market correction. Think people forgot institutions have already financed their expansion, which was a 700M SEK round 100%+ ago (multiple times oversubscribed). And that $SIVE uses primarily external fabs, so it's not the same as a lot of the major financing concerns affecting Neoclouds. Anyway, up to management execution, I'm not running Sivers so no clue what other companies are working with them now, whether it's Cambridge or some other US player like Semtec. TLDR: - Just need to wait for earnings for more disclosures around qualifications + partners, only so much public can guess from channel checks and supplier mapping. Not quite the same as $LITE CEO where you can see the guy on Fox news or Bloomberg interviews every week (which is a good thing)

@aleabitoreddit Oct 19 2 clicks

Serenity on X

Last time I've seen "It's all over" type posts was... 3 months ago during the situational awareness liquidation. And... 3 months ago before that... with Iran tensions + Oil + Helium + LNG fears/pricing. And... few months ago before that. I'm sure everyone remembers Deepseek back in like 2025 Each one had genuine risks, with rate hikes/margin liquidations to oil prices + supply disruption. And now financing/credit markets. Not trying to downplay the risks but haven't seen it materialize in the companies I'm tracking (eg. optical sector) given counterparty types + sector visibility + architectural shifts in 2027-2028. For example in this $LITE CEO interview, the guy said there hasn't been any order cancellation/delays given rate hikes. But obviously neocloud sector that uses heavy debt leverage might be different. I've seen that the general sentiment was a recovery after midterms in 2 weeks, I'm not the type to give predictions like that. We'll see Directionally bullish on certain sectors of AI supply chain like memory/photonics, but everything was sold off indiscriminately, even if they're not exactly the same trade.

@aleabitoreddit Oct 19

Serenity on X

I can't give advice on what others should do but from what I'm seeing from $AAOI revenue ramp, it looks something like this from disclosed capacity. -> $858M Q1 2027 -> $1.29B Q2 2027 -> $1.62B Q3 2027 -> $1.89B Q4 2027 -> $2.14B Q1 2028 They're sitting on $6.8B MC right now. I'd be very impressed if they do $7.5B exit annualized revenue run rate in 2027 off 40% gross margins and markets assigned them the same valuation. And judging from $LITE making capacity agreements into 2030-2031... I don't quite think demand stalls out after 2027-2028. Anyway, lot of execution risk obviously, but H1 2027 inflection point is pretty close, so I'm personally waiting.

@aleabitoreddit Oct 19 3 clicks

Serenity on X

$LITE CEO interview on demand, counterparty + financing risk: - "CFO of $GOOGL came on and reported capex going up Y/Y, we expect hyperscalers for report an increase in capex in earnings. Based on what we see, we see their spend going up." - "The purchase agreements we have, to a large degree, are noncancellable." - The difference between Lumentum and companies like $AVGO, "we are not as dependent on AI datacenter buildout going forward" There's a conversion between electrical products to optical products" creating structural demand. Q: When do you see able to see yourself having more inventory? (catching up to surge in demand) - When we look at demand, there is "no end in sight" TLDR: Financing risk and demand aren't equally distributed across AI supply chains. But markets are just selling everything off together. As for photonics in specific, I said this before, but increasing optical content per XPU + CPO/NPO adoption keeps going up so we'll likely see sector spend go up, even if aggregate capex goes down. There's the usual "no end to demand in sight" commentary, but $LITE does have capacity commitments into 2029 now... and are negotiating for 2030. (with new information that most of which are likely hyperscalers + non-cancellable) So doesn't quite sound like it's all over for the AI trade, especially photonics.