Cam Invests

i don't know anything about ai, or investing
don't follow me

[Cam Invests @camfbell]

PHOTONICS MIGHT BE ONE OF THE MOST INTERESTING AI SUBSECTORS RIGHT NOW.

AI is obviously a huge industry, but the deeper I go into it, the more I realise just how important the optical layer is becoming.

More GPUs, bigger clusters and faster networking means an insane amount of data has to move between chips, racks and data centres.

That’s where companies like $COHR, $LITE, $AAOI, $MTSI, $MRVL, $AVGO, $GLW and $APH start getting interesting.

I spent the weekend trying to map out who actually does what, where each company sits in the photonics stack, and how they all connect.

This is my attempt at putting the whole US AI photonics supply chain onto one page.

Still plenty for me to learn here, but hopefully this helps anyone else trying to understand the space.

Feel free to message me and I can send over the high-res version.

[Cam Invests @camfbell]

HUGE DAY FOR AI STOCKS TODAY!

1. AI stocks cooled off a little after last week’s run.

The Nasdaq fell 0.3% and the S&P 500 slipped 0.1% as oil jumped around 5% on renewed Strait of Hormuz concerns.

That matters because higher oil feeds straight back into the inflation/rates conversation, exactly when the market is waiting on US CPI later this week.

So we had a bit of risk coming off despite some pretty bullish AI news underneath.

2. Nvidia $NVDA is now helping solve arguably the biggest bottleneck in AI after power... money.

Nvidia $NVDA announced partnerships with Apollo $APO, BlackRock $BLK, Blackstone $BX, Brookfield $BAM, Goldman Sachs $GS and KKR $KKR to create financing platforms targeting more than $500B of third-party capital for AI infrastructure.

That number is nuts.

Big Tech AI spending is already expected to exceed $730B this year, but Nvidia clearly understands that selling more GPUs eventually requires someone to finance the data centres around them.

AI infrastructure is becoming an asset-finance business as much as a technology business.

3. And there was a pretty important warning hiding inside the data-centre boom.

Banks are apparently getting much more serious about permitting, local opposition, water and power availability before financing new projects.

At least 75 US data-centre projects worth around $130B faced local opposition in Q1 alone.

Interesting part for me... developers are increasingly looking at onsite generation as one way around some of these issues.

Demand for compute still looks insane. But getting land + power + permits + finance together is becoming the actual bottleneck.

This probably matters more and more for the whole data-centre infrastructure trade.

4. Microsoft $MSFT might be about to seriously ramp its own AI chips.

According to The Information, Microsoft $MSFT could unveil its Maia 300 as soon as September and has reportedly been discussing manufacturing capacity with TSMC $TSM for more than 300,000 chips in 2027.

Longer term it reportedly wants capacity for 1M+ units.

Still unconfirmed numbers, and Microsoft says the reported figures don't reflect the full scale of the program.

But the direction is pretty obvious.

Hyperscalers don't want to stop buying Nvidia $NVDA GPUs. They want multiple ways to build compute and push down the cost per token.

5. Intel $INTC is taking advantage of its huge rally.

Intel $INTC announced a $15B common-stock offering and shares dropped around 4%.

The money will help fund capex and its expensive foundry buildout. Intel already raised 2026 capex guidance from $18B to $20B and is targeting high-volume production on 14A in 2028.

Dilution obviously isn't fun.

But this is another reminder of just how capital intensive the next stage of AI infrastructure is becoming. Even chipmakers are raising massive amounts of money to keep building capacity.

6. A much smaller one I've been watching...

Navitas Semiconductor $NVTS sued Renesas Electronics $RNECY claiming its GaN chips infringe four Navitas patents.

Why do I care about a patent lawsuit?

Because GaN is moving into the AI data-centre power stack as rack power gets denser and more efficient conversion becomes increasingly important.

Navitas $NVTS is only around a $3B company, so protecting the IP around that transition could matter quite a bit if GaN really does become a bigger part of AI power architecture.

Shares fell roughly 5% today though, so the market wasn't exactly celebrating it.

7. Another sign of how much capital wants into data centres...

CyrusOne is reportedly preparing for a possible 2027 IPO that could raise around $5B.

The company is owned by KKR $KKR and BlackRock $BLK and operates more than 60 data-centre campuses across the US, Europe and Japan.

It's not alone either. There is now a growing pipeline of giant AI/data-centre businesses heading toward public markets.

To me that's worth watching closely.

[CleanSpark Inc. @CleanSpark_Inc]

CleanSpark fully supports Governor Abbott's directive to audit every data center project in Texas's interconnection queue. The transparency requirements on power, water, tax incentives, and ownership are exactly the standard responsible developers should be held to. We welcome it.

From day one, our operational model has focused on reliability and accountability: we actively support the grid, utilize efficient closed-loop cooling to minimize water consumption, and maintain the rigorous audit and disclosure standards expected of a Nasdaq-listed company.

Importantly, we add significantly to local tax bases and keep economic benefits in the communities that host us. Texas deserves developers it can trust, and we're committed to upholding that standard.

[Cam Invests @camfbell]

HUGE DAY FOR AI STOCKS TODAY!

1. Probably the biggest driver today was actually the US jobs report.

July payrolls fell by 23,000 vs +80,000 expected. Unemployment came in at 4.1% and the odds of a September Fed rate hike dropped to around 44%.

The S&P 500 closed at a record and the Nasdaq gained 1.3%.

Pretty simple read for AI stocks. Softer economy = less pressure for higher rates = more breathing room for expensive growth names.

2. SK Hynix $SKHY just approved another massive bet on AI memory.

The board approved 54.3 trillion won, roughly US$38.3B, of investment through 2031.

35.2T won goes into the next Yongin fab focused on advanced DRAM/HBM, with another 19.1T won going into a new Cheongju fab focused on NAND and AI server storage.

To me this is one of the cleaner reads on where the industry thinks demand is going. You don't commit nearly $40B to new memory capacity if you think this AI buildout disappears next year.

3. Cloudflare $NET ripped around 16% after earnings and this one caught my attention.

Q2 revenue came in at $696.1M, up 36% YoY, with adjusted EPS of $0.29. They also lifted full-year revenue guidance to $2.86B-$2.87B.

The interesting bit isn't just the beat.

Cloudflare is seeing AI agents drive more machine-to-machine internet traffic, which means the AI spend is spreading beyond GPUs into networking, security and the infrastructure sitting between all these agents and applications.

That second-order AI trade keeps getting broader.

4. Atlassian $TEAM was one of the biggest software movers, up roughly 35%.

Revenue hit $1.77B, up 28% YoY, cloud revenue grew 31% to around $1.2B and adjusted EPS came in at $1.87 vs $1.50 expected.

Its AI product Rovo is now being used by more than 80% of Fortune 500 companies.

This is interesting because one of the big fears around software has been that AI destroys traditional SaaS. Atlassian is starting to show the other side of that trade. Existing software platforms might actually become more valuable if AI gets deeply embedded into how customers use them.

5. And a completely different part of the AI infrastructure story.

Oklo $OKLO announced its Groves reactor reached first criticality less than a year after groundbreaking, meaning it achieved a controlled, self-sustaining nuclear chain reaction.

Important distinction though. Groves is an isotope test reactor, NOT one of Oklo's future commercial power plants feeding data centres.

So I wouldn't overread it.

But execution is probably the biggest question around these advanced nuclear names. Actually building, commissioning and operating a reactor gives Oklo a real-world proof point while AI keeps making power availability a much bigger part of the data centre equation.

Overall this was a pretty bullish day for the AI trade.

Lower rate pressure helped the whole sector, but underneath that we're still seeing real money going into memory, networking, software and eventually power. The AI buildout continues to spread further away from just the GPU.

[Cam Invests @camfbell]

Great call. Love the company and the management team. Am invested.
He didn't say no dilution for sandserville though.
He said they should get 90% PF, and they are aiming not to dilute.
They will end up issuing shares, they can't afford not to.
But they will announce that the same time they announced $meta to balance it out 😅

[Cam Invests @camfbell]

@asklivermore Very good reminder in this time.

[Cam Invests @camfbell]

@babyfolio $CLSK not sure if it's an earnings play or a longer play for me yet

[Cam Invests @camfbell]

@daniel_koss Haha. Good thing musk is never off about his timing or forecasts

[Cam Invests @camfbell]

I entered a position on $CLSK $13.77.
I actually expect the stock to dip on earnings as they will be unlikely to name the georgia contract.

My thesis is once this contract gets more realised.
The market will start to price is their much larger Texas site.

I will write an article on it Soon.

Thanks @michaelsikand for your write up.

[Cam Invests @camfbell]

@Sell_4_Profit @kevinu @ChrisCamillo I don't know if we can blame them for btc not getting to where everyone predicted haha.

Private Jet is pretty crazy.

I see them as moving away from btc now, using their energy and sites for compute power. Which I am more bullish on that btc.

[Cam Invests @camfbell]

My conviction in $CLSK is growing

I first heard about it through @KevinU

One thing I’ve picked up from @ChrisCamillo is to go back through old earnings calls and try to really get a feel for management

so that’s what I did this weekend

My first concern was pretty simple. They’re a BTC miner... are they really sophisticated enough to make the jump into AI?

But the more I looked into the team and the company history, the more that view changed

They weren’t really “bitcoin people”

They were business people who saw an opportunity in mining, entered later than most, and executed well

Now they’re moving into AI infra

The big catalyst is the unnamed contract everyone thinks could be Meta. No idea if that rumour is true

But the contract itself looks big enough to seriously change the value of the business

And the part I really like is the balance sheet

$CLSK has a huge BTC treasury, so they may be able to fund a lot of the capex without smashing shareholders with another raise

still digging. conviction keeps growing

[Cam Invests @camfbell]

@Sell_4_Profit @kevinu @ChrisCamillo Very different business now from 2023.
Are you more or less bullish on the new business model?

[Cam Invests @camfbell]

@CompoundingLab Did you sell prior to earnings as you think it will be a negative result?

[Cam Invests @camfbell]

Huge fan of FSD.
But why does it decide to turn a random corner when I'm driving with no destination?
Common sense should default to straight on no?
@elonmusk @larsmoravy @woodhaus2 @aelluswamy @Tesla_AI @Tesla

[Cam Invests @camfbell]

Anyways to play the valuation difference between SK hynix Korea and USA?
@babyfolio this is your speciality 👀

[Cam Invests @camfbell]

@kevinxu Has to be $AMZN .
Big upside from the AI boom.
With almost 0 risk of loosing it all.

[Cam Invests @camfbell]

@TheBigBerbowski Malta?

[Cam Invests @camfbell]

It's a strange time when -1.23 red seems like a good day!

Kospi

[Cam Invests @camfbell]

@vaderrrrrrrrr

[Cam Invests @camfbell]

Interestingly $trt has been largely unaffected by the recent AI sell off

[Cam Invests @camfbell]

Holy
Wtf is going in in Korea