I find it interesting that the degens on $RDDT are starting a viral campaign to save Wendy’s ( $WEN ).
The US fast food burger chain.
And the stock price is now up 20% overnight.
Just for background: Wendy’s is a popular community meme, where people work behind the dumpsters after their portfolios go to 0 from 0DTE options.
(No positions, just found it amusing)
I wonder if it’s going to work?
Don’t quite think “siphoned off” is the correct term.
It’s capex for massive revenue increase or margin increase down the line.
$AMZN is probably my favorite hyperscaler right now and example to give.
Amazon’s headcount is absurd, like ~1.57M. If the capex goes into automating their workforce with LLMs.
Then transitioning into physical AI:
- things from self driving (deliveries)
- robotics (Amazon warehouses, shipping automation).
+ revenue increase from building out AWS compute with Trainium and possibly selling chips too with the Neocloud strat.
It’s probably the clearest path forward compared to every hyperscaler out there.
$TSLA optimus use case targets is extremely broad as a pitch, but Amazon already has a specific reason to scale robotics for internal opex optimization.
As for $GOOGL, probably 2nd right now, AI capex was necessary for defending its Google Search moat Gemini from ChatGPT
They also have Google Cloud revenue with efficient TPUs + can sell TPUs like Nvidia GPUs.
Gemini user volumes keep going up (despite the lack of contention in frontier benchmarks); and AI strategy to be working for ad optimization too.
But there’s less clear paths with physical AI stuff ig?
Microsoft and Meta are still trying to convince the market why capex is necessary, (we’re kinda seeing that in effect with Meta’s 30%+ Y/Y revenue growth), but doesn’t look like they’re convinced.
As for market narratives, Microsoft Maia seems to be behind, their AI development was stunted from OpenAi investments, so sentiment is kinda in the ground.
But think that will change down the road like the 180 with Google.
I’m sure all the hyperscalers are seeing the leader effect right now:
If you have the leading LLM, people will keep using it. That LLM gets smarter from all the training data; and that gap might be structural.
Which is why everyone is kinda rushing the buildout right now, but for some the immediate incentives seem obvious.
OpenLight (private) seems to get bigger and bigger every time I look at it.
If you’re curious about their public ecosystem outside of Advantest:
$JBL ( $SIVE partner ) for scaling PICs
$MRVL and $MXL for DSPs
$TSEM for the foundry.
$300394.SZ / TFC Optical (OSAT/subassembly)
Can often get exposure into private growth through public equities if you’re not an accredited investor.
Since optical players look very interconnected.
Fun new information discovery from Poet OSINT community:
Seems likely that $POET / $SIVE are going to power a Top-3 hyperscaler (either Amazon, Microsoft, Google).
Given a Linkedin update from Ankur Singla (CEO of Lumilens).
Who stated their customer is one of the top 3 hyperscalers with their post focusing on CPO/NPO.
With that clue, seems more likely the Sivers CW DFB light source path over other EML suppliers given it's CPO Scale Out/NPO.
If you don't remember, Sivers is the laser supplier to Poet. And Poet has purchase agreements with Lumilens.
Always fun to find major potential breadcrumbs in the wild before they're officially confirmed. (Disclosure, long Sive)
Nancy Pelosi just disclosed:
- $1 million - $5 million of $INTC March 2027 calls
- 200 $UBER March 19, 2027 call option contracts with a $50 strike price
This was done last month May 29th. (her husband is the active trader executing this). Just for the people who like following along Pelosi family trades.
Always amazing looking at $MU earnings:
Revenue: $41.46B vs. $35.8B est.
EPS: $25.11 vs. $20.78 est.
Forecasts:
Revenue: $49B to $51B, vs $43.24B est
EPS: $30.00 to $32.00, vs. $25.31.
“Micron said on Wednesday that it has signed 16 long-term agreements”
"When completed, we expect approximately half or more of our company revenue to be under these"
Looks like memory demand has become structural…
But great earnings to show up the AI trade is continuing to ramp up.
Today, Anthropic has directly accused the $BABA Qwen AI lab of distilling its frontier AI models.
By creating thousands of fake accounts and over 28.8 million exchanges.
Feels like this is kind of known by now... but there's been no real penalties enforced yet.
We'll see what happens.
Think so. But in the meantime, I call my strategy: Diversified Losses.
With $AXTI, $SOI, $AAOI, and many others.
Had a massive drawdown recently,
CPO exposure was hit the hardest (Foci, Msscorp, etc) and adjacent names, feels bad.
Probably lesson personally, I had too much concentration in photonics vs. memory/other sectors without weighting/hedging properly.
With Soitec, there's been a few negative institutional reports that I'd disagree with.
Think AXT was hit harder in specific just because of float expansion/dilution concerns. AAOI, probably just brought down with the theme.
I can't give advice on buying, so completely up to you to make for cost averaging or entering positions.
But I do think we're still early in the Supercycle with photonics, there's bound to be corrections/crashes along the way up.
If my personal thesis is correct though, many of these names will have a major inflection point in midway through 2027 scaling up to 2028.
Markets don't typically wait to price things in advance, but some ideas might be a tad early or in the buildout given it's H2 2026 now.
Which is why it's important to build your own conviction.
Very interesting statement today: $MU CEO predicts a multi-decade memory demand cycle driven by humanoid robots.
"Humanoid robots, he says, will require roughly ten times more memory than today’s Level 2+ autonomous vehicles."
"And that demand wave is set to begin before the decade is out."
Something as well as was "Over time, we expect the value of on-device AI combined with pent-up unit replacement demand to drive memory demand growth"
Which is also another trend (Apple Intelligence is currently dog, but I'm sure we'll see innovations with localized/edge AI).
Feels like all the industry leaders from $TSM Chairman, $TSLA Elon Musk, to $MU CEO see humanoids as the next major trend so physical AI is probably next.
I wonder if the world is going to have enough memory. Or if we'll see enough breakthroughs to shrink memory usage.
Global markets in a nutshell:
Japanese Company: Active monopoly over random chokepoints in the hyperscaler AI buildout.
Valuation: $150m-$350M
Silicon Valley: Here's a $200m for a "Seed Round".
The $WEN meme traders made it to Global Media from Japan to US!
The burger fast food chain is now up ~50%, which is pretty impressive.
Feels bad to be sidelined and see my AI memory/optical portfolio perform worse than a $RDDT memestock tho.
Did anyone take positions?
Feels like the only thing that hasn’t crashed…
Is memory like $MU, indexes, or large cap semis like Intel so far.
- Photonics from $AXTI to $SIVE down 40%.
- Space from $ASTS and $RKLB down 40% 1M.
- Popular AI names like $PLTR is down ~35% YTD.
- Software like $CRM down -40%.
- Bitcoin sub <60k, Ethereum sub <$16k.
Not a fun time with a hawkish fed narrative and potential rate hikes.
However this does sorta feel overshot due to margin liquidations on less liquid assets compared to mega caps.
But we’ll see what happens, usually fundamentals override liquidity shock in the longer run.
I’m still personally bullish on the AI buildout + upstream AI capex beneficiaries, but 1-2 potential rate hikes certainly don’t help.
oh i think u misunderstand... I'm actually more bullish than ever as prices go down.
$SIVE at ~$1.9B MC you have:
- $GFS reference laser
- laser for Ayar and others in $NVDA NVLink ecosystem for CPO scale up
- $POET and others for CPO scale out
- $JBL and others for 1.6T+ pluggable optical transcivers
- O-Net for ELS mass production
So you have all these hyperscaler suppliers trying to create as much as possible with Sivers as the laser bottleneck...
And it's just a matter of how much Win Semi + others partners can make, with $SIVE receiving ~60% gross margins + optionality to TAM expand downward.
The 15% recent share expansion that Swedish media have been bearposting is for NASDAQ listing float + M&A (and it's authorization).
You just have potential short sellers running illegal bot farms, Swedish media bearposting a company Swedish exchange, on top of a general macro drop.
The ~$140M convertible note is real, but it's pennies to US institutions. We'll likely see more institutional ownership when data comes out.
But this is also why it's good for $SIVE to prioritize NASDAQ listing so they don't need to deal with this noise and local media.
With $AAOI at $10B, you have:
- A end-to-end US laser/design/assembly player projecting $471m month revenue H1 entering H2 2027.
- With other hyperscalers like $AMD apparently discussing LTAs.
And they probably are sitting on a ton of cash after running ~$1.4B in ATMs (400M + 400m+ 600m).
Kinda all that needs to be said with those insane revenue projections as long as management doesn't BS.
This just reminds me of when $NBIS crashed to $70 last year even while projecting $7-9B ARR, and as timelines got closer it recovered to $250+. Feels like dejavu.
I'm just waiting for both volume ramps to hit. Photonics are generally more volatile than the rest.
Nah, gonna fill out the job application for $WEN and set up camp behind the dumpsters.
Even $MU and $SNDK ridiculous performance today couldn't save my port.
Cheering on Japan though, hoping Sweden gets mogged with a 2-0 to represent the 20% drop.
Looks like Power Semis are already starting price hikes.
Which is bullish thematically for US power semi trade from $AOSL to $POWI, before the 800V DC shift fully hits.
China's Yangjie Technology, announced a 2nd price increase this year, raising prices across its full product range by 10%-15%, July 1.
Chongqing China Resources Micro, Silan Microelectronics, and NCE Power also price hiked.
“Demand is playing a larger role than costs in the latest round of price increases” from AI DCs, to energy systems, and new EVs.
Another potential beneficiary is the upstream material companies that make these:
- nonferrous metals like copper and tin.
-Plastic molding compounds
- chemicals, and packaging consumables
“Nearly all major Chinese power semiconductor makers have raised prices in 2026.” (Digitimes)
So positive read through on power semi adjacent (reduce China-driven price pressure + better pricing + demand validation)
There's a global correction right now, no clue when it stops.
Kospi is down -8.18% (Sk hynix/Samsung)
Nikkei is down -4.8%
TWSE Index is down -3.82%
As you've seen with $SOI or $RKLB dropping 30-40%.
From personal experience, high beta stocks get hit a lot harder, but usually frontrun index drops.
And they typically recover earlier?
Aside from Korea which is typically volatile, when major indexes start correcting 3-4% a day, usually not a fun time.
“What matters is not the eggs, it is the Goose itself”
SoftBank probably down -13.5% today
after markets saw this investor presentation today.
Also the OpenAI IPO possible 2027 delay cause Altman wanted a $1T valuation might have been a small cause…
Just Elon Musk casually sounding the alarm.
On the massive demand + price hikes for $MU / SK Hynix / Samsung memory relative to supply.
He probably has higher visibility into the memory bottleneck than others…
OFC I'm aware. But I'm personally sleeping comfortably since I have conviction in my hyperscaler mapping research with $SIVE.
And yes, I still have my million+ share position.
Not sure if people realize this: but I'm only here to share my thoughts/ideas.
I don't control market volatility, what decisions you all make, or how markets react to new information synthesis.
It's much safer for analysts to just reactively tag along Morgan Stanley/JP Morgan/Goldman Sachs research whenever it's created and just summarize.
Rather than coming up with new ideas from OSINT mapping and waiting them get validated.
Because when you discover a new angle:
Everyone keeps heatedly debating topics of 4-6 inch InP fabs, employee count, who their hyperscaler customers are, volume ramp timelines, etc to try and play devils advocate with a thesis.
Then actively monitoring every single 5-20% price movement.
I'm forced to stay on this topic more since it's less validated + there's always heated discussions. Just like $EWY in Feb, which I did memory projections on + Helium/LNG/Oil analysis.
But months later everyone sees memory looks structural with Micron's 16+ LTAs and LNG isn't taking down SK Hynix margins.
Or $NBIS from last year in terms of sum-of-parts / dilution structures vs $IREN.
And now it's close to ATHs and listed on $QQQ.
I'm personally just waiting Sivers to volume ramp in 2027 + listing on NASDAQ to support their M&A efforts.
So they can walk down the same path as $LITE when they scaled from $3B to $60B+.
just throwing out a random thought, the UBS report is anecdotally true + and it's interesting there's more Chinese reports on Anthropic distillation.
Many US startups/companies using AI:
Are using Chinese models like DeepSeek since they’re much cheaper than Gemini/OpenAI/Anthropic reasoning models for cost per task.
Kinda a catch 22 when US is built on capitalism and everyone goes for the cheaper options.
I think the US just has to get good and out innovate here:
- both with heavy KYC-auth frontier models for US only like Mythos given those are being distilled by competitors. and amybe staggered tiers for allied partners so US is always ahead
Maybe something like dedicated AI Google auth like FaceID + Persona would do the trick before access/key creation for the newer models, then short lived scopes.
At least addressing those simpler fake/stolen account + pool allegations, requiring biometrics before use.
Then US Gov drafting new regulations making it illegal for sharing access similar to identity forgery with bank accounts.
- then the most obvious one is just heavily optimized reasoning models that mogs DeepSeek and others.
Kinda good Trump admin paused Fable/Mythos access, imo not quite sure why a trillion dollar company keeps letting itself get distilled over and over again.
There should be friction added if you’re accessing something close to ASI.
TLDR: US just needs to good at cheap inference and add bank-grade authentication to AI model access. Easier said than done tho.