Alex A.C. (@AnalysisOp) header
Alex A.C. (@AnalysisOp) profile picture

M. Eng. Electronic ⚡ 〰️ Financial data & analysis. Value investing. Growth? at a reasonable price 🦔 🧔🏻

EU
Joined February 2023
Photo by Alex A.C. (@AnalysisOp)Photo by Alex A.C. (@AnalysisOp)Photo by Alex A.C. (@AnalysisOp)Photo by Alex A.C. (@AnalysisOp)Photo by Alex A.C. (@AnalysisOp)Photo by Alex A.C. (@AnalysisOp)Photo by Alex A.C. (@AnalysisOp)Photo by Alex A.C. (@AnalysisOp)Photo by Alex A.C. (@AnalysisOp)Photo by Alex A.C. (@AnalysisOp)
Alex A.C. retweeted
SEMI has joined a Samsung, SK Hynix and Micron early-stage working group to set a charging standard for mobile robots in semiconductor fabs, media report, including batteries, charging-station communications, connectors and battery-management systems, aiming to reduce maintenance burdens and dependence on individual robot vendors. $SKHY $MU $SSNLF #Samsung #SEMI en.sedaily.com/technology/20…
4
11
62
6,049
Esto va a reventar. Es imposible defender un sistema en el que tú pagas para que se beneficien extranjeros que nunca han pagado un € en España. O para que tu voto valga igual que el de alguien que no ha pisado España. Es la prioridad nacional al revés.
90
2,642
40
7,742
87,959
With $NBIS reporting Thursday, November 12, and so much fear and volatility across AI infrastructure, we’ve made our latest Nebius 2027 report free to read in full, including the valuation and every price target our Premium members received in October. We think the market is underestimating how much business Nebius can deliver next year. In our October 26 report, we forecast $21.86B of 2027 revenue against the $12.24B Street estimate we reviewed, a gap of 78.5%. Our EBITDA forecast is $12.49B against $6.53B, or 91.3% higher. Most of that difference comes from the amount of capacity we expect to earn revenue during 2027. We worked through the sites, hardware deliveries, customer commitments and commissioning schedules quarter by quarter. At our expected fleet yield, the Street revenue estimate would require roughly 513 average billable MW. Our model reaches about 938 MW. Our targets for December 31, 2027: Bear: $340.31 Base: $471.44 Bull: $628.75 Probability-weighted: $479.29 We weight those cases 30%, 40% and 30%. Even our Bear case produces $19.07B of revenue, more than 50% above the Street estimate in the report. Our disagreement with consensus runs through the operating forecast, well before we put a valuation multiple on the business. The build accelerates sharply toward the end of 2027. Our Base case closes the year with 1,613 MW of accepted IT capacity, but much of it earns revenue for only part of the year. Pennsylvania and Independence arrive late in our schedule. We expect a much larger business by December, while recognizing that a delayed commissioning date would remove months of revenue from the forecast. We also follow who is paying for that capacity and on what terms. Longer contracts bring customer advances that help fund construction and equipment. Shorter contracts give Nebius more exposure to the prices customers are willing to pay for scarce compute. We expect that mix to support strong growth, with the financing burden increasing if prices weaken or customers contribute less cash upfront. Our Base case includes $58.77B of cash capex and $12.46B of new equity during 2027. Those new shares are already included in our targets. We remain bullish after allowing for that financing because we expect the additional capacity to support far more revenue and earnings than the Street estimates in our report. Thursday’s disclosures will help us judge whether the sites, contracts and financing are progressing as we expect. After earnings, we’ll refresh the data and publish separate Premium models for each individual year: 2027, 2028, 2029 and 2030. Members will be able to follow how the expansion develops year by year, including the revenue, funding needs and dilution along the way. Our members received this research first, and their support allows us to open selected reports to the wider community. We hope this one gives people something useful to work through ahead of Thursday. The full report is linked in the first reply. Thanks again for supporting independent research.
🤖 Made with AI
11
26
5
198
30,286
Alex A.C. retweeted
China is turning its political system into a strategic advantage by concentrating national resources on science and technology. The U.S., Europe, and other East Asian economies need to rethink how they compete. Free-market competition is excellent at fostering diversity, creativity, and disruptive innovation. But it is not always the most efficient way to mobilize limited time, capital, and resources. This matters enormously in AI, semiconductors, quantum computing, and other capital-intensive industries. Technological supremacy is no longer just about who invents the best technology. It is also about who can mobilize capital, talent, infrastructure, and industrial capacity the fastest. We should preserve the strengths of free markets without ignoring the strategic advantages of coordinated national investment. The real competition is not simply technology versus technology. It is increasingly system versus system.
2
3
2
30
11,547
Alex A.C. retweeted
$LITE CEO Michael Hurlston addressed the 45% selloff today. On demand, he said “There’s a massive disconnect between what we’re seeing in our demand book and what the markets are showing. Our demand couldn’t be stronger.” Asked whether the market reaction gives him pause on $MRVL as the buyer, he said “We really like Marvell. We think they’ve got a lot of upside from where they are, and that’s part of the reason we took all stock.” He admitted “the timing was bad” but didn’t back off. “Our board of directors never considered anything but taking all stock,” citing diligence on Marvell’s books and design wins. On getting the deal closed, targeted for 1H28: “It’s going to be a bumpy one.” He named regulatory approval as the biggest hurdle left. $LITE $MRVL
50
50
19
600
216,927
Alex A.C. retweeted
Mohnish Pabrai: "If you've got a spectacular winner, it's going to take care of itself even with a small position size — as long as you don't trim." "We don't need to go all-in with a big conviction [from the start]. The winners are going to get there. No matter where they start out, they'll get there. Just be relaxed and patient and they'll be fine." (h/t @TIP_Network)
2
12
1
120
10,724
Charlie Munger: "It takes character to sit with all that cash and to do nothing. I didn't get to where I am by going after mediocre opportunities." Great opportunities are rare. FOMO makes mediocre ones look great & tempts you into moves you'll regret. FOMO is temporary. Regret is forever.
2
54
6
428
22,126
Alex A.C. retweeted
One detail from our latest RL work that I love: The team nearly doubled inference-engine throughput in a benchmark… and the RL training step didn’t get any faster. The time saved on model calls was lost in the sandboxes. In another job, fixing our own CPU serving code saved 444 seconds per training step. These are the details that make systems engineering so interesting. The next big gain can be hiding in request handling, sandbox setup, or how stages overlap, and every successful change can move the bottleneck somewhere else. Our team built an internal auto-tuner to make that process repeatable: profile, forecast, test, measure again. Across four models and two workloads, the best configurations delivered 1 to 5x faster RL steps on fixed GPU budgets, with a median speed-up of about 3.5x @kingsquare2, @Mhr1036 and I wrote up how it works, where the gains came from, and the experiments that surprised us. Really glad we’re sharing the engineering behind this.
Inside Nebius’s RL auto-tuner: how optimizing training, inference and sandboxes together delivered up to 5× faster steps on fixed GPU budgets. And five surprising lessons from 210 runs, including why faster inference can leave RL just as slow.
6
5
74
6,327
Alex A.C. retweeted
When it comes to growth stocks... the pendulum always swings too far in both directions. Back in April should $AAOI have been trading at 12x NTM revs? guess not (see chart below) Fast forward 7 months to today/tonight... should $AAOI be trading at 1x NTM revs? probably not Keep this in mind... investors are always too bullish at the highs and too bearish at the lows. This is starting to remind me of 2022/2023 when so many growth stocks became really disconnected from their fundamentals... they got oversold... they got washed out... investors gave up on them and many missed the ensuing rallies. $APP was up ~600% in the 18 months after it's December 2022 lows $PLTR was up ~300% in the 18 months after it's December 2022 lows $UBER was up ~200% in the 18 months after it's June 2022 lows $NVDA was up ~700% in the 18 months after it's October 2022 lows $HIMS was up ~200% in the 18 months after it's May 2022 lows. $CVNA was up ~2,200% in the 18 months after it's December 2022 lows $META was up ~400% in the 18 months after it's October 2022 lows $SMCI was up ~700% in the 18 months after it's April 2022 lows $CRWD was up ~300% in the 18 months after it's January 2023 lows $NFLX was up ~200% in the 18 months after it's May 2022 lows $RKLB was up ~1,500% in the 18 months after it's April 2024 lows $CELH was up ~300% in the 18 months after it's May 2022 lows $TSLA was up ~150% in the 18 months after it's January 2023 lows and the winner is $ASTS which was up ~4,000% in the 18 months after it's April 2024 lows. Reality check: - was this a perfect earnings report from $AAOI? no - are there minor things to dislike in the report? yes - did they guide to 200% YoY revenue growth in Q4? yes - did the guide to 500k units per month by end of 2026? yes - did they guide to 1M units per month by end of 2027? yes - does this mean they can do 250-300% revenue growth in 2027? yes - should the stock be trading at 1x 2027 revs? of course not - is this market a flaming pile of shit? yes - do fundamentals and valuations matter right now? no - will fundamentals and valuations matter again some day? hope so Have a good night. Let's hope tomorrow is better than today.
The price action on $AAOI is almost comical because it just shows how short sighted investors can be. If you take a step back... look at the reported numbers, look at the guidance and read through the press release it's quite obvious there's a lot to be excited about. $AAOI reported Q3 revs of $269.9M which was a tiny beat vs consensus of $268.1M but still up +128% YoY. The stock is probably down AH because the midpoint of Q4 guidance is slightly below consensus... they said Q4 revs will be $385M to $420M with the sell side at $420M. The midpoint on their Q4 guidance still implies +200% YoY revenue growth... is this market so fucked up that +200% growth is now disappointing? That's just absurd. They also said Q4 net income margins will be 12-14% which is fine by me because it puts $AAOI on track to hit 17-18% in 2027. If you're a longer term investor... and trying to find the best asymmetric opportunities in the market with the stocks that have become completely dislocated from the fundamentals... well then I think $AAOI is near the top of that list... with this in mind the Q3 numbers really don't matter and same for Q4 although I'd still love to see them eclipse the high end of their guidance. The biggest catalyst right now is their production expansion ramp for 800G and 1.6T transceivers. If you read through the press release they expect 500,000+ units per month by end of 2026 and 1,000,000+ units per month by end of 2027. This means $AAOI could be doing $250-300M per month (not including their CATV business) by end of 2026 and $500-600M per month by end of 2027. I believe $AAOI is still on track to do $4B+ revs in 2027 and then exit the year at $6-7B ARR. Using the midpoint of their Q4 guidance, $AAOI would do ~$1.015B revenues in 2026 which is up +123% YoY from 2025. If they hit my $4.0B revenue target in 2027 (base case)... it puts revenue growth at ~300% for the full year... which also means the stock is now trading AH at less than 1x 2027 revs. If I'm right and $AAOI can do $4.0B revs in 2027 with 17-18% net income margins, it implies the stock is trading AH at ~6x 2027 eps... using fully diluted 87.8M shares and assuming they do 1-2 more offerings over the next 12 months ...hopefully at much higher prices unless they used debt which might make more sense depending on the terms. They also announced a signed LTA with a hyperscaler customer that gives them visibility.
23
24
3
328
61,630
Alex A.C. retweeted
$AAOI earnings call — live notes 4: Several CPO customers gave positive preliminary results on AOI laser performance this quarter. Full laser qualification ongoing; ELSFP modules next.
$AAOI $COHR $LITE AAOI's Q3 2026 revenue and adjusted EPS beat expectations, but its Q4 revenue guide is below consensus. Datacenter sales are accelerating, while the timing of usable manufacturing capacity has become a more important constraint on the revenue ramp. Q3 revenue reached $269.9 million, up 127.5% year over year and 40.6% sequentially. That was 0.7% above the $268.11 million consensus and within the company's $255 million to $290 million guidance. Non-GAAP EPS was $0.17 against $0.14 expected. GAAP net income turned positive at $6.3 million. The datacenter business provides the clearest demand signal: revenue reached $155.6 million, up 254.2% year over year and 44.5% sequentially. CATV also contributed materially, with revenue of $110.9 million, up 37.6% sequentially. It accounted for roughly 39% of the company's sequential revenue increase, so the total-company growth rate includes a substantial broadband equipment contribution. The new multi-year strategic agreement with a Tier 1 hyperscale customer is another positive development. It adds company-specific support to the demand outlook. The release does not disclose the customer's identity, contract value or delivery schedule, so the agreement's contribution to near-term revenue still needs to be established. The capacity update is the main execution issue. Management says current manufacturing equipment capacity is approaching 330,000 units per month. It now expects enough equipment installed by the end of 2026 to produce more than 500,000 monthly units of 800G and 1.6T products, compared with the previous plan of around 650,000. For US expansion in 2027, obtaining electrical power and other construction issues are taking longer than expected. The end-2027 target is now above one million monthly units, versus roughly 930,000 previously, but a larger eventual target does not resolve the timing of when capacity becomes available. Q4 revenue guidance is $385 million to $420 million. The $402.5 million midpoint implies another 49.1% sequential increase, yet sits 5.8% below the $427.26 million consensus. Q4 non-GAAP EPS guidance of $0.48 to $0.63 has a $0.555 midpoint, essentially in line with the $0.56 expected. Adding the revenue guide to the first nine months implies full-year revenue of about $998 million to $1.033 billion, below the previous target of around $1.1 billion. The business is growing rapidly, but the implied revenue path has moved lower. Profitability improved unevenly. Non-GAAP gross margin was 28.7%, below the previous 29% to 30.5% guidance and down from 29.8% in Q2. GAAP gross margin rose to 28.2% from 27.7%, so the two measures moved differently. Q4's non-GAAP gross-margin guide of 29.5% to 31% points to recovery, with the causes of the Q3 margin shortfall still to be explained. There was meaningful operating improvement despite that margin pressure. The supplemental financial tables show non-GAAP operating expenses rose only 4.0% sequentially to $70.3 million, against 40.6% revenue growth. Non-GAAP operating income moved from a $10.3 million loss to a $7.2 million profit as revenue grew faster than expenses. GAAP operating income remained negative at $1.4 million; $10.1 million of net non-operating income helped the company report its GAAP net profit. The industry comparison remains constructive. Coherent's latest Datacenter & Communications revenue rose 79.1% year over year to $1.95 billion. Lumentum's quarterly revenue rose 136.5% to $1.26 billion, led by pump lasers and optical circuit switches, with cloud-module margins also improving. Together with AAOI's datacenter growth, these results show spending reaching multiple parts of the optical network. Their different product mixes also explain why strong industry demand cannot, by itself, establish AAOI's margin trajectory. Customers' spending, each supplier's delivery capacity and the profits it retains need to be assessed separately. Funding and execution also need to be read together. AAOI ended September with $621.5 million in cash, cash equivalents and restricted cash. Its October 30 credit agreement provides $170 million of equipment-related term-loan commitments and a $30 million revolving facility. Financing supports equipment purchases, but power availability and construction schedules still govern how quickly that equipment can become productive. The release does not include a cash-flow statement, so a higher cash balance alone cannot establish operating cash generation. My assessment at the release stage: the evidence for AI optics demand has strengthened, and AAOI has delivered substantial datacenter growth. Its Q4 revenue guide is below market consensus, and Q3 adjusted gross margin fell below its own prior guidance. The call now needs to clarify 800G and 1.6T shipment timing, how the new agreement feeds into deliveries, and the schedule for bringing the delayed US capacity into production.
2
1
11
2,185
Alex A.C. retweeted
The discount rate only comes down as the uncertainty resolves. The demand is not in question but I don't think the broader market truly understands why yet.
AMD CEO @LisaSu: "The market can be short-sighted, but it is what it is." Long-term, AMD's CEO believes in AI demand and that computing and intelligence will affect every aspect of our daily lives.
2
6
53
8,155
$ARM CEO on the potential slowdown in US data center expansion: "I do believe strongly that if the US hesitates in any way, shape or form in terms of data center build out, the rest of the world is is ready and willing. There is just huge demand for this capacity" [CNBC]
$META President on WSJ on building data centers after the mid-term elections in the US: "So I think it's going to be harder. I think it has been hard, I should say. But I also think that it's becoming clear what models work, and the states want those models that work because they want the ability to say, 'We've invested, we have these jobs,' but there has to be local support. And so we are in 23 different states. Five of them will have one-gigawatt projects and larger, and Texas is one of those."
6
4
2
62
23,121
We’re going to be deep diving heavily back into the AI infra buildout here soon. Core model updates coming to $NBIS $CRWV $IREN to start. Looking to build year by year through 2030 with an emphasis of separating fact from FUD regarding where these names stand today. We remain very constructive on the sector as a whole with special credit to those with a healthy balance sheet, plenty of financing flexibility for the buildout ahead, and those creating full AI native integrated stacks. Overall, we believe the current market dynamics are extremely sentiment driven and could flip “overnight” at any time. We aren’t predictors of short term technicals, but we believe the flip will come now or over the next several months. In the meantime, we see a multitude of opportunities across the sector and continue to average into the most high quality names. We view this is a generational buying opportunity for the next leaders in Cloud. The fundamental picture has not worsened and if anything valuations are more attractive across tech as a whole than they have been over many years. We are buyers and will continue to be at these prices, unless the thesis is fundamentally proven broken.
2
4
41
2,531
Alex A.C. retweeted
InnoLight starts 200mm NPO production and tapes out 300mm designs - On Nov. 3, InnoLight said its first near-package optics device had entered early production on Tower’s ($TSEM) 200mm PH18M process in California. - InnoLight also taped out several follow-on NPO products on Tower’s 300mm silicon photonics platform. > InnoLight, an established pluggable module supplier, is designing optical engines that sit closer to the package. Watch shipment volumes for its first 200mm product and yields on the 300mm follow-ons. Lumentum secures more than $5B in purchase commitments on its second major NPO/CPO win - On Nov. 2, Lumentum ($LITE) said its second major NPO/CPO win carried more than $5B in customer purchase commitments. - The same day, Marvell ($MRVL) agreed to acquire Lumentum in an all-stock deal. The companies expect it to close in the first half of 2028. > If the deal closes, Marvell could co-design its DSPs with Lumentum’s lasers. Watch how quickly the $5B-plus commitment converts into quarterly shipments and revenue. Siltronic sees AI demand squeezing chip supply for PCs and smartphones - On Nov. 4, Siltronic’s CEO said chipmakers were prioritizing data centers, limiting growth in parts of the PC and smartphone markets. > When PC or smartphone shipments soften, separate end demand from chip availability. The same shipment decline can point to different inventory levels and follow-on orders.
4
5
2
30
10,980
Alex A.C. retweeted
Some details from the $COHR call that stood out: 1) Almost every recently signed TLA includes some form of upfront customer funding, such as prepayments or development payments. As Jim Anderson put it, customers have “skin in the game” on the capacity being built for them. 2) Internal InP output doubled, but the internal versus external sourcing mix “hasn't changed much in a year.” Transceiver demand grew fast enough to absorb the additional supply. 3) Scale-across revenue exceeded $200M and nearly quadrupled YoY, “ahead of MultiRail.” That product’s revenue ramp is expected to begin in the March quarter. Coherent is spending heavily to expand, but customers are helping fund capacity.
$COHR In Q1, we signed another significant PhotonLink LTA with a major AI company. The total revenue value of this new LTA is approximately $20 billion over 5 years.
1
3
1
49
11,772
Alex A.C. retweeted
Bazinga! $AMD now upping guidance next year to 80%. It's conceivable, a stretch but possible, they do $100b in 2027. "Last quarter, we said we expected annual server revenue to grow more than 70% and data center segment revenue to more than double for the year. Since then, customers have expanded their Epic and Instinct deployment plans and we have secured more supply to support that growth. We now expect server revenue to double year-over-year in 2027 and data center AI revenue to grow even faster than we projected 90 days ago. As a result, we expect data center to be more than 80% of our total revenue in 2027 and to drive the majority of our revenue and earnings growth."
1
8
96
6,302
Alex A.C. retweeted
Tohoku University and Toyota-supplier Aisin built a 16 nm AI chip (SoC) with 32 MB of embedded STT-MRAM that holds both a compact Linux boot image and Tiny YOLOv3 (object detection AI model) weights on-chip. No external DRAM or SSD is needed for startup or inference. It boots in 65.7 milliseconds (39× faster) and runs object detection at ~50.7 mJ/frame with big energy savings. They used weight sharing and hybrid 8/16-bit math to fit the model. The chip also includes on-chip SRAM (used for feature maps and other working data) and a DDR interface, so external memory can still be attached when larger models or more capacity are needed. TSMC manufactures the chip. Nice real-silicon demo of near-memory computing for edge AI — useful for automotive cameras, sensor fusion, and always-ready devices that need instant-on and low standby power. Not revolutionary: Others – Hailo, Ambarella, Renesas, etc. – already use large on-chip SRAM, embedded MRAM, or in-package DRAM, but it’s a good example of the gains when OS + model live in non-volatile on-chip memory. $TSM $AMBA #Hailo #Renesas #semiconductors
19
54
4
342
20,057
Alex A.C. retweeted
Replying to @modal
@modal, a serverless AI compute platform, increased many on-demand compute rates today. Modal surpassed $300M in annualized revenue in May and crossed $1B in October. The price increases look like a positive signal for GPU compute demand and pricing.
This tweet about someone begging to rent their B300s at 4.6/hr has raised some eyebrows. Individual price quotes are just one of many signals in the GPU rental market. It is more important to track medium to long-term pricing across on-demand, reserved, and capacity block offerings. Spot prices are also heavily influenced by timing, provider, and region. OP clarified that the offer was in India. We don't see a clear inflection based on the GPU pricing data we’re currently tracking here: funda.ai/play/gpu-spot-price… We elaborate below.
5
4
1
30
7,997
$AAOI just secured a massive five-year semiconductor supply agreement. Yunnan Germanium’s subsidiary, Yunnan Xinyao, has signed an agreement to supply Applied Optoelectronics with 900,000 4-inch indium phosphide (InP) substrates through 2031. The numbers are significant: • 180,000 substrates annually • $194.4M contract value for 2027 alone • $1,080 implied price per substrate • Five-year supply framework through 2031 Here’s why I find this particularly bullish. AAOI is aggressively expanding its 800G and 1.6T optical transceiver manufacturing capacity, targeting over 930,000 units per month by the end of 2027. InP substrates are a critical upstream material for manufacturing the lasers used in high-speed optical communications. Securing a five-year supply arrangement of this magnitude suggests AAOI is preparing its upstream supply chain for a substantially larger manufacturing operation. You don’t build a supply chain of this scale without serious production ambitions. Deliveries remain conditional on Chinese dual-use export licenses, creating some geopolitical supply risk. Nevertheless, this is another meaningful piece of AAOI’s capacity expansion story. gu.qq.com/resources/shy/news…
19
15
6
233
32,125
Alex Karp says Palantir has a "moral imperative" to take the whole AI market “And our somewhat crazy aspiration to take the whole market doesn’t look that crazy right now. And we believe at this company, and many of the people watching it, we do not only have a financial reason for taking the whole market, which Glazer is salivating at." "We, on the mission side, we believe we have a mission reason for taking the market. And that mission reason is we support your alpha. We support your sovereignty. We’ve done this for over 20 years. We support the warfighter. We are delighted that we can support the warfighter even more because we’re making so much money. But we believe." "We have a moral imperative to be the dominant player in our industry broadly defined, including all AI applications and the use of model open and closed, especially training open-weight models. And that moral imperative drives every part of our business.”
Palantir CTO reveals AI researchers are taking up smoking to signal to other researchers that they believe AI will cure cancer--and calls it a “mass psychosis” Shyam Sankar (CTO): “But if you understand that it’s actually a theological critique, then you can kind of weigh it against the empirical facts of what is actually happening out there. “I think there’s a profound truth to Pascal’s saying that every human has a God-shaped hole in their heart. And it’s important to understand that many of these folks in Silicon Valley who are building this technology have filled that hole with the machine god of AGI. “And I mean, there’s so many interesting stories around this. For example, so many AI researchers have started smoking as a way of signaling to other researchers how much they believe in AI.” Margaret Brennan: “What? That makes no sense. What do you think?” Shyam Sankar: “I believe so much in AI that it’s going to cure cancer. So I’m showing you my chips are all in. So there is kind of this mass psychosis, social delusion around it.”
2
6
1
30
8,332