100Baggers.club

Invest in exceptional businesses with durable moats at the right time, and compound alongside them through cycles.

[100Baggers.club @Miltons2Cents]

800V may become one of the next big themes in AI infrastructure.

The first wave of AI infrastructure was easy to see:

GPUs,
HBM,
optical modules,
servers,
advanced packaging,
semiconductor equipment.

Many of the biggest winners came from those layers.

But after NVIDIA’s next-generation Rubin platform, if AI rack power keeps moving from tens of kilowatts toward hundreds of kilowatts — and potentially close to 1MW — the bottleneck is no longer just faster chips or more memory.

The question becomes:

How do you deliver that much power into the rack?
How do you distribute it?
How do you protect it?
How do you cool it?
How do you test it?
How do you make the whole system reliable enough for production?

That is why 800V matters.

It is not just a new voltage spec.

It is a signal that AI datacenters are moving from server rooms toward high-density power factories.

At higher rack power, if voltage stays too low, current becomes enormous. That means thicker copper, more loss, more heat, more space taken by power delivery, and more difficult protection.

Raising voltage helps reduce current and forces the whole power chain to be redesigned.

But from an investing perspective, the key is not simply asking:

“Who does 800V?”

The real question is:

Whose solution enters customer designs?
Whose components get listed in the final customer-approved procurement list?
Who passes certification?
Who can scale production?
Who can recognize revenue?
And who can keep margin and cash flow?

This chain could create opportunities in several groups:

1. Datacenter power infrastructure
Names like Vertiv and Eaton are already exposed to AI datacenter buildouts.

2. System-level 800V delivery
Companies like Flex may benefit if they can turn 800V power racks into deliverable systems.

3. Grid and site-level electrical bottlenecks
GE Vernova and similar companies may matter as datacenters need transformers, switchgear, grid interfaces, and site power infrastructure.

4. Power semiconductors and controllers
Names like Texas Instruments, onsemi, Navitas, and AOSL may benefit if their devices move from reference design into real customer production.

The key lesson:

800V is not a single-stock theme.

It is a new infrastructure chain.

The winners will not be determined by who has the best story.

They will be determined by the revenue conversion sequence:

Design win → customer approval → production → revenue → margin → free cash flow.

The next big AI infrastructure winners may come from this sequence.

Full report:
https://t.co/cCFD3YKcRB

#AIInfrastructure #800V #NVIDIA #Rubin #DataCenters #PowerSemiconductors #Vertiv #Eaton #GEV #TXN #ON #NVTS #AOSL #Investing #StockMarket

[100Baggers.club @Miltons2Cents]

The AI power trade is not one single trade.

VST, CEG, and TLN all benefit from the same big idea:

AI datacenters need reliable power.

But investors should not put them in the same bucket.

They are three different risk/reward structures.

1. VST = the balanced cash-flow validation story

VST is not a regulated utility, and it is not just an AI power concept stock.

It is a merchant reliable-power platform with nuclear, gas, retail load, hedges, long-term PPAs, and buybacks.

The key question is:

Can higher PJM capacity prices, Meta/AWS contracts, Cogentrix, and existing reliable generation assets turn into ordinary free cash flow per share?

VST is the “asset revaluation → FCF validation” name.

Full VST report:
https://t.co/s61w5gBAAW

2. CEG = the quality premium story

CEG is the cleaner, higher-quality platform.

It has a stronger nuclear base, and after Calpine, it also has more gas dispatch capability and commercial customer exposure.

The market already recognizes that quality.

So the key question is not whether CEG is a good company.

It is whether ordinary free cash flow per share can keep proving the valuation premium.

CEG is the “quality premium → per-share cash flow proof” name.

Full CEG report:
https://t.co/3itKim0Xwq

3. TLN = the high-beta execution story

TLN is sharper.

Its value driver is concentrated around Susquehanna, AWS, PJM capacity prices, and the Cornerstone acquisition.

Susquehanna is less than 20% of TLN’s equity capacity, but it generated more than 40% of total company generation in 2025.

That is why the asset matters.

TLN also has more open forward power exposure, which means more upside if PJM prices, AWS execution, and Susquehanna availability all go right.

But the downside is also more direct.

TLN is the “high beta → high execution hurdle” name.

Full TLN report:
https://t.co/UazNH9DMs5

My simple framework:

VST is the balanced name.
CEG is the quality name.
TLN is the high-beta name.

The mistake is saying:

“AI needs more power, so all power stocks are the same.”

They are not.

The real work is asking:

How much capacity revenue is retained?
How much contract value is actually disclosed?
How much EBITDA becomes ordinary FCF?
How much FCF becomes FCF per share?
How much good news is already priced in?

Reliable power is becoming scarce.

But stock returns will depend on who can turn that scarcity into per-share cash flow.

#VST #CEG #TLN #AIPower #AIInfrastructure #NuclearEnergy #PowerStocks #DataCenters #Investing #StockMarket

[100Baggers.club @Miltons2Cents]

https://t.co/8BjmQzm1Eg
Tech stocks are getting crowded. That’s usually when I like to look at areas the market has ignored, hated, or left for dead.

We’ve talked a lot about SaaS, AI infrastructure, and semis recently.

Today, let’s put that aside and talk about U.S. consumer stocks.

The real story starts with prices.

A couple of years ago, eggs were the crazy one. Anyone who went grocery shopping in the U.S. remembers it. A carton of eggs felt absurdly expensive. Breakfast, baking, family grocery baskets — everything got hit.

Now eggs have finally cooled down.

But just as consumers got some relief there, beef started moving higher again.

Beef is different from eggs. The supply cycle is much slower. You don’t fix cattle supply in a few weeks. It takes time — breeding, feeding, processing, supply rebuilding. So when beef prices rise, that pressure can stick around.

It shows up in grocery aisles, fast-food menus, restaurant margins, and eventually company earnings.

That’s the real state of U.S. consumption right now:

Consumers haven’t stopped spending.

But every dollar is being reallocated.

Rent, mortgages, car payments, insurance, food, healthcare, and credit card interest take the first bite out of the household budget. What’s left is where consumers make choices:

Costco bulk pack or full-price brand?
Walmart value aisle or specialty store?
McDonald’s value meal or a higher-ticket restaurant?
New sneakers now or wait?
Home improvement project now or delay?
So when people say U.S. consumption is strong, they’re not wrong.
And when people say U.S. consumption is weak, they’re also not wrong.
The better question is:
Where is the money flowing?

That’s why consumer stocks are getting interesting again.

Walmart and Costco are not just winning because they are cheap. They are becoming default choices for households trying to stretch budgets. Groceries, household goods, pharmacy, bulk purchases, membership value — all in one place.

When budgets get tight, consumers may actually rely on these places more, not less.

McDonald’s is similar. Beef inflation can pressure margins, but value meals still matter when consumers are watching every dollar. People don’t stop eating. They just become much more careful about what a meal is worth.

On the other side, companies like Nike, Lululemon, and Home Depot are in a different position.

They are not bad companies. They may still be great long-term brands.

But buying full-price sneakers, yoga pants, a sofa, new appliances, or starting a kitchen renovation is easier to delay.

When budgets tighten, those purchases get pushed out.

So you can’t just say, “This consumer stock is down, so it’s cheap.”

A lot of stocks are down for a reason.

You still have to ask:

Has inventory really cleared?
Are margins stabilizing?
Is operating cash flow improving?
Is free cash flow per share actually coming back?

A lower stock price is not the same thing as a bottom.

It just earns the right to be studied.

The food chain is also fascinating.

Falling egg prices help consumers and grocery retailers. A cheaper grocery basket helps Walmart, Costco, Kroger, and anyone trying to reinforce a value perception.
But for egg producers, falling egg prices mean the extraordinary profits from the last cycle are normalizing.
Beef inflation is the opposite. It hurts consumers and pressures restaurants with beef exposure. If menus get too expensive, traffic can suffer.
Chicken has been more stable, which can help certain low-price menus and chicken-focused chains.

Same food inflation story. Completely different stock implications.

That’s one of the easiest mistakes in consumer investing:

You see one macro variable.

But by the time it hits company earnings, the impact can be totally different depending on the business model.

My framework for consumer stocks right now is simple.

First: great companies still need the right price.

Walmart, Costco, Visa, Mastercard, McDonald’s — these are strong businesses. But the market already knows that. If the price has already discounted years of good news, the return may not be attractive.

Second: cheap companies still need data.

Nike, Lululemon, Home Depot, and parts of beverages and alcohol may look more interesting after the selloff. But low valuation is not enough. I want to see inventory, margins, and cash flow improve together.

Third: cyclical companies need normalized earnings.

Eggs, beef, chicken — these are cycles. You can’t value egg producers using peak egg-price profits. And you can’t permanently punish them using trough conditions either.

So the next opportunity in consumer stocks is probably not about betting on a broad U.S. consumer rebound.

It’s about dispersion.

Who captures pressured household budgets?
Who converts revenue into margin and cash flow?
Whose stock has not already priced in the good news?
Who looks cheap but is not actually fixed yet?

The hotter tech gets, the more the market may start looking back at ignored areas where cash flow still exists, share is stable, and valuation has become reasonable again.

U.S. consumption has not collapsed.

But consumers are getting more selective.

The opportunity is not in buying “consumer” as one category.

It is in finding where the budget is migrating.

Great companies need the right price.
Cheap companies need better data.
Cyclical companies need normalized cash flow.

#ConsumerStocks #Investing #Stocks #Retail #Walmart #Costco #McDonalds #Nike #Lululemon #HomeDepot #Inflation #FoodInflation #StockMarket #ValueInvesting #USConsumer

[100Baggers.club @Miltons2Cents]

https://t.co/uEWMonytOb

This HBM cycle is fundamentally very strong. SK hynix’s 2026 Q1 revenue and margins were extraordinary, and AI datacenter demand for HBM, server DRAM, and enterprise SSDs is absolutely real.

But investors need to separate HBM from the rest of memory.

HBM, traditional DRAM, NAND, and consumer SSDs are not the same trade.

HBM behaves more like a core bottleneck next to AI compute. It has stronger pricing power and better economics. A lot of the price increases in traditional DRAM, NAND, and consumer SSDs are happening because manufacturers are prioritizing AI-related supply first, which squeezes conventional memory supply.

Further down the chain, many companies may be benefiting more from inventory revaluation than from truly durable cash flow.

That’s the key risk in this cycle:
not fake demand, but real demand + inventory profits + leveraged buying + retail speculation all reinforcing each other at the same time.
What matters from here isn’t who is up limit-up every day. It’s three signals:

Contract pricing is still very strong, but spot pricing is already starting to soften around the edges — especially DDR4, NAND wafers, and consumer SSDs. If contracts keep rising while spot prices weaken, that’s usually a dangerous late-cycle signal.

Watch cash flow, especially for high-inventory companies. Strong net income alone is not enough. Operating cash flow needs to follow, and inventories cannot keep expanding forever.

Watch search behavior and market language. When people stop researching what HBM is and start talking about leveraged ETFs, borrowing money to buy stocks, crashes, and margin calls, the market has shifted from industry analysis into leverage and speculation.

What the market is really betting on is whether memory can evolve from a traditional cyclical commodity into part of AI infrastructure.

That idea is not crazy. HBM, server DRAM, and enterprise SSDs really are becoming part of AI datacenter efficiency.

But the story still needs to be proven with data:
HBM yields and certifications,
long-term customer commitments,
margin durability,
free cash flow after capex,
and whether traditional DRAM and NAND are seeing structural demand or just temporary supply squeezes.

The market is betting that a cyclical industry is becoming infrastructure.
Our job is to keep checking whether that transition is actually happening — or whether the market is once again treating a cyclical peak like a permanent new normal.
#HBM #AI #Semiconductors #Memory #DRAM #NAND #SKHynix #Micron #Samsung #AIInfrastructure #DataCenters #EnterpriseSSD #Investing #Stocks #TechStocks #SemiconductorCycle #MarketCycles #AIStocks #KoreaStocks

[100Baggers.club @Miltons2Cents]

https://t.co/xsfqOHKepj
The biggest thing retail investors should pay attention to with Cerebras isn’t whether the technology is real. It’s the IPO structure.

The IPO was priced at $185.
The next day, the stock opened near $350 and closed at $311.
That means the people who actually got shares at $185 were IPO allocators and insiders. By the time most retail investors entered in the secondary market, they were already buying a stock that had nearly doubled.

And the cost basis gap is even more important.

The weighted average exercise price of vested employee options was around $5.
Series G / pre-IPO secondary transactions were around $36.
Series H was around $89.

For those holders, even $185 was already an enormous gain. Even after the recent pullback, this is still a very attractive exit window.
That’s the part many retail investors miss when they only focus on AI, OpenAI, chips, or the 750MW contract:
Retail is buying the story.
Early investors and employees may be selling years of ultra-low-cost inventory.

And the future supply pressure is not small either.

The IPO itself issued 30M Class A shares.
Post-IPO basic shares outstanding are ~215M.
The greenshoe can add another 4.5M shares.
Non-executive employees could unlock up to 2.5M shares on Day 1, and another 2.5M shares on Day 2 if the stock closed above 133% of the IPO price.

133% of $185 is $246.
The stock closed at $311 on Day 1, so that trigger was already met.
The bigger pressure comes later.

The prospectus says up to ~171M shares may be released in stages during the lockup period, including up to ~15M shares held by directors and executives. On top of that, there are RSU tax sales, vested employee options, OpenAI warrants, the 2026 equity incentive plan, and employee stock purchase plans.

So from here, Cerebras doesn’t just need to prove the technology and OpenAI contract are real.
It also needs to prove the market can continuously absorb wave after wave of low-cost shares coming into circulation.
For retail investors, the real question isn’t whether Cerebras is a good company.
The real question is:

Are you buying a future platform…
or providing liquidity for early investors and employees at a very high valuation, low float, and ahead of a massive unlock schedule?

[100Baggers.club @Miltons2Cents]

AI is real. The AI bubble is also real.

I want to define the problem clearly: saying AI is in a bubble is not the same as denying AI.

A technology can be extremely powerful, genuinely improve how people live and work, and still produce a massive investment bubble along the way.

Railroads, the telegraph, the telephone, and the internet all did this. They raised productivity, changed how society was organized, and improved living standards. AI will very likely do the same.

Anyone who has actually used Claude Code, ChatGPT, Cursor, Gemini, or similar tools can feel the productivity gain. Coding, writing, research, information organization — many tasks are now several times faster, sometimes orders of magnitude faster.

So I do not doubt the long-term value of AI.

Over a 5- to 10-year horizon, AI will almost certainly change how many people work and live. But that is not the same as saying today’s capital spending and valuations are reasonable. Real productivity gains do not automatically mean the entire AI value chain will capture those gains as durable profits.

At the beginning of the year, I was skeptical of the AI bubble argument.

Since Claude Code’s capability jump last year, I and many people around me have felt a real productivity step-change. From Silicon Valley giants to small teams of a few dozen people, AI has allowed companies to do with fewer people what previously required much larger teams.

In a sense, AI is converting part of opex labor cost into cheaper and more scalable compute/tooling cost.

Over the past year, we have also seen many small teams produce surprisingly high revenue. Some of those stories are clearly distorted by mixing up revenue and profit. Some are marketing theater. But I also believe many of them are real.

The problem is that local truth is not the same as system-wide truth.

The more real some of these local cases are, the easier it becomes for the market to capitalize a much broader profit story before it has actually appeared.

Take corporate cost-cutting in the US.

Over the past few months, I have talked with many friends about how AI is changing their day-to-day work. They work in healthcare, finance, consulting, and tech. The sample size is small and not statistically meaningful, but the pattern is surprisingly consistent.

Many companies are cutting headcount or slowing hiring while saying publicly that they are all in on AI and using AI to improve efficiency.

But the reality is often more complicated.

Some companies are using AI as a convenient narrative to correct over-hiring from the previous cycle. Some are moving work to lower-cost regions like India or relying more on contractors. Others are simply not backfilling roles when people leave.

On paper, headcount comes down. The remaining employees do use AI to improve productivity in some tasks. But many of them feel busier, more stretched, and under more pressure.

So yes, efficiency gains are real.

But that does not mean new revenue and new profit are already appearing at scale.

For large public companies, it is very hard to separate how much of the margin improvement truly comes from AI, and how much comes from layoffs, hiring freezes, offshore labor, contractor usage, or general cost control.

At the startup level, in the limited data I have seen, I also have not yet seen a broad, large-scale jump in revenue and profit.

Now look at the companies that are clearly making money.

AWS, Azure, and Google Cloud are all growing impressively. Google Cloud is already huge, yet it is still showing strong revenue growth and improving margins. That is very real.

Cloud providers, GPUs, semiconductors, servers, cooling, power equipment, and many second-order AI infrastructure companies have all benefited from AI capex.

But there is a key issue.

The money first comes from enormous hyperscaler capital spending.

Cloud providers are expanding, but they are still middle layers. They sell compute, storage, model services, and cloud infrastructure. The true end customers are the companies buying those services to create value in their own businesses.

The AI value chain ultimately needs to close like this:

End companies use AI to make more money or save sustainable costs.

They keep buying model and cloud services.

Cloud and model companies earn strong enough margins.

That demand then flows to GPUs, servers, power, cooling, semiconductor equipment, and the broader supply chain.

Right now, the first part of that loop is still not clear enough.

We have seen killer apps.

NotebookLM is a great example. Claude Code is a great example. These products are excellent, and they clearly improve user productivity.

But what supports the long-term valuation of the entire AI value chain cannot just be killer apps.

It has to be killer businesses.

A killer app proves that users want to use it.

A killer business proves that end customers can make enough money from it, and are willing to share that profit with the AI value chain on a durable basis.

Those are very different things.

What capital markets and Big Tech are betting on right now is a race against time.

Before capital spending peaks, before market sentiment turns, a true killer business has to appear at scale.

Long term, I believe it will happen. AI is too powerful, and the application surface is too large. Society will absorb this capability into production systems.

But in the short term, two forces are racing each other.

On one side: capex, data centers, power, GPUs, model training, and inference costs are rising very quickly.

On the other side: truly verifiable incremental profit at the end-customer level has not yet appeared at comparable scale.

If killer businesses arrive early enough, the AI bull market can extend smoothly. Today’s investments will be caught by future cash flows.

If they arrive too late, the market may realize that it paid upfront for a distant vision rather than a closed profit loop.

In that case, the technology does not have to fail.

Asset prices can still reprice violently.

So my current view is simple:

AI’s power is real.

The AI bubble is also real.

The real debate is not whether AI is useful. It obviously is.

The real debate is whether this round of capex and valuation has already pulled forward too much of the future killer business.

For individual investors, this is a psychological test.

If you are too bearish too early, you may miss a massive technology bull market.

If you chase too late, you may end up bag-holding at the hottest point of the bubble.

The hardest part is not deciding whether AI is real.

The hardest part is deciding when real technological progress turns into real, large, durable profits that can support the entire value chain.

[100Baggers.club @Miltons2Cents]

AI hardware stocks are all cyclical — but not the same kind of cycle.

Equipment stocks follow a slow, capex-driven cycle.
AI demand → cloud capex → GPU / ASIC / HBM orders → fab / memory / packaging expansion → finally equipment orders and revenue (ASML, KLA, Lam, AMAT).
This creates a big lag.

We’ve seen this play out:
• 2020–2021: chip shortage + cloud demand → equipment orders surged
• 2022–2023: PCs/phones/servers weakened → end demand slowed, but equipment revenue held up due to backlog
• 2024–2026: AI capex + HBM + advanced packaging → cycle re-accelerates

The trap: when revenue, margins, and orders all look great, you may already be in the mid-to-late cycle.
So don’t just buy “AI beneficiaries.” Watch:
• order intake
• backlog trends
• customer capex signals
• DIO / DSO
ASML / KLA = structural control points
Lam / AMAT / TEL = higher memory beta → more cyclical
Bottom line: equipment isn’t less cyclical — just slower and lagged.
The edge is buying control points when the market worries about capex, not when everything looks perfect.

Memory is a completely different beast — a fast, amplified cycle.

It’s not just demand. It’s:
real demand + price expectations + customer inventory + channel inventory + spot/contract pricing + capex reflex.
That’s why it’s so deceptive.
Classic cycle:
• 2021: demand strong → customers double-order, channels build inventory → prices spike
• 2022–2023: demand weakens → inventory unwinds → prices collapse → earnings crash
Micron is the textbook case:
FY22: high revenue, ~45% gross margin
FY23: revenue halved, margins went negative
This is the key trap:
When profits peak, margins peak, and PE looks low — it often isn’t cheap.
It’s peak earnings.

Because:
high profits → aggressive capex → supply catches up → price reverses → earnings collapse
HBM improves the cycle, but doesn’t eliminate it.
It delays it.

If supply ramps in 2027–2028 and demand doesn’t keep up, the cycle still turns.
So memory must be read counter-cyclically:
Low PE + high margins + high capex = warning sign
Better entries come when:
• earnings look terrible
• capex is cut
• inventory clears
• prices stabilize

Different cycles → different playbooks.

Equipment: buy control points on weakness
Memory: buy cycle bottoms, not peak profits
AI is real.
Cycles are also real.

Understanding both is the edge.

#AI #Semiconductors #Investing #Stocks #TechStocks #AIInfrastructure #Memory #HBM #NVDA #ASML #KLA #AMAT #LAM #Micron #TSMC #StockMarket #LongTermInvesting

[100Baggers.club @Miltons2Cents]

Microsoft 56% operating cash margin: is AI investment really “effortless”?

Microsoft FY2026 Q3 shows some very strong numbers:

CFO / revenue: 56.3%
Productivity & Business Processes margin: ~60%

At first glance, this doesn’t look like a company burning cash in an AI cycle.
It looks more like a legacy software cash machine funding a new AI wave.
But the real question is:

👉 How much of that operating cash actually turns into shareholder free cash flow?

So what’s the right lens?

A. 56.3% operating cash flow margin means AI investment is basically pressure-free — Microsoft can confidently front-load CapEx.

B. PBP margin near 60% shows Office / enterprise software profits are still deep — valuation should focus on revenue growth and Cloud / RPO.

C. Operating cash flow is strong, but AI CapEx is taking cash first — the key isn’t whether Microsoft earns money, but how much free cash remains per dollar of revenue after CapEx.

D. High CapEx is normal in early AI platform cycles — as long as enterprise entry points are stable, short-term FCF pressure shouldn’t matter.

#Investing #Stocks #Microsoft #MSFT #AI #Cloud #FCF #CapEx #Valuation #TechStocks

[100Baggers.club @Miltons2Cents]

Open Duolingo (DUOL) and you’ll see a very tempting number: ~12–13x PE.

That’s… unusual.

A company with:

high user engagement
strong brand
subscription revenue
and an AI education narrative

…trading at low-teens PE?

The first instinct for many investors is:

👉 “Is this a mispriced high-quality growth stock?”

So what’s your reaction?

A. The market may be discounting AI tutor risk — if the learning entry point fragments, a low multiple could be justified.

B. For high cash-conversion subscription businesses, PE isn’t the best lens — I’d look at EV/FCF and FCF per share first.

C. Low PE on a high-quality grower at least signals improving asymmetry — it belongs on a rerating watchlist.

D. This could be a tax illusion — if one-off tax gains are treated as recurring earnings, normalized PE may actually be 30x+.

#Investing #Stocks #Equities #GrowthStocks #ValueInvesting #AI #EdTech #Duolingo #FCF #Valuation

[100Baggers.club @Miltons2Cents]

Bloomberg published an article ranking companies that announced buybacks this year by dollar amount, so I converted those figures into percentages

[100Baggers.club @Miltons2Cents]

April was a standout month for Alphabet — up ~30%, one of its strongest in years. But this re-rating isn’t just about Gemini hype. It’s because multiple lines actually delivered in Q1 2026: $109.9B revenue (+22% YoY), Search +19%, Cloud +63% with ~32.9% operating margin, and backlog above $460B.

At this point, Google is no longer a simple “will AI disrupt Search?” story. It’s three engines running together: a dominant ads cash cow, a rapidly scaling Cloud business, and an emerging AI infrastructure platform. What’s hardest to replicate is how vertically integrated it is — from TPUs and data centers at the bottom, to Gemini and Cloud in the middle, to distribution via Search, YouTube, Workspace, and Android on top, all the way through monetization via its ads system. In terms of stack completeness, it’s arguably the most complete AI platform among big tech.

But the real question now isn’t whether Google is strong — it clearly is. The question is whether that strength translates into high-quality shareholder returns. AI products are easy to overestimate on the demand side; strong usage doesn’t automatically mean strong profits.

Q1 looked great, but free cash flow was only around $10.1B (~9.2% margin). That reflects a shift toward a more capital-intensive model — AI data centers, TPUs, training, and inference infrastructure. Put differently: if products like Gemini, AI Overviews, and AI Mode don’t ultimately translate into higher ad yield, incremental Workspace monetization, or meaningful Cloud workloads, then more usage could actually mean higher inference costs and pressure on FCF.

So my current view: Google is still one of the strongest platform companies in the AI stack — no question. But at this price, “great company” alone isn’t enough reason to aggressively add. It’s more about tracking and validating: Search monetization under AI, Cloud margins, returns on capex, and incremental ROIC.

Google’s dominance historically came from Search being not just high-demand, but high-margin and effectively monopolistic. AI, by contrast, is a much more competitive and structurally lower-margin game. Whether Google can convert its full-stack advantage into similarly high profitability — that’s still something the market needs to see.

#Google #Alphabet #GOOGL #AI #Cloud #BigTech #Investing #TechStocks #GenerativeAI #Valuation

[100Baggers.club @Miltons2Cents]

SaaS Isn’t Dead. Feature-Only Software Is.

I’ve seen a lot of people claiming that Naval said “SaaS is dead.”

I looked up what he actually said. That is a major misread.

Naval’s point was: “pure software is uninvestable.”

That does not mean “SaaS is dead.”

The key distinction is between pure software and responsibility software.

If a software company’s value is just features, UI, dashboards, automation scripts, or lightweight collaboration, then yes — AI has already crushed the cost of building it and reduced its differentiation.

That kind of software will have a very hard time sustaining long-term investment value.

But the real value of durable SaaS is not software features.

It is control over high-responsibility workflows, write-back rights, systems of record, compliance chains, and budget entry points.

For those companies, AI cannot simply bypass the existing stack.

To actually execute tasks, AI needs access to the customer’s data, permissions, workflows, audit trails, and responsibility systems.

So AI may not replace these companies.

It may make them more important.

The valuation logic for SaaS has to shift from software functionality to responsibility systems.

If you are just selling tools, you are in danger.

If you control real-world state changes, you may become even more valuable.

By that standard, two of the most interesting examples to me are VEEV and INTU.

VEEV controls R&D, clinical, quality, compliance, and commercial workflows inside life sciences.

These are not ordinary software features.

They are production-grade, audit-grade, responsibility-grade processes for pharma companies.

For a VEEV customer, switching systems does not mean changing an interface.

It means redoing validation, permissions, audits, data models, and the entire compliance responsibility chain.

AI is unlikely to bypass VEEV.

It is more likely to be embedded into Vault, quality, clinical, regulatory, and industry data workflows — reducing the cost of documentation, approvals, validation, and process execution.

VEEV also has a very strong balance sheet.

Cash and short-term investments are around $6.6B.

The company has historically been conservative with capital, but earlier this year it approved a $2B two-year buyback plan and has already repurchased about $180M.

INTU is another example.

It controls tax filing, accounting books, payroll, payments, cash flow, and loan matching — financial states that are verifiable and carry real consequences.

AI does not just make INTU’s software easier to use.

It can move INTU from helping users operate financial software toward helping users complete financial outcomes.

TurboTax Live, QuickBooks, Payroll, and Credit Karma are all moving closer to that responsibility layer.

INTU has around $3B in cash and short-term investments.

It has already repurchased about $1.8B recently, and in March said it would substantially accelerate buybacks, with up to $3.5B of remaining authorization available.

That matters.

Buffett’s framework on buybacks is simple:

A repurchase only creates value when a company buys its own stock below intrinsic value.

That is the key point.

A buyback is not just returning capital.

When done at attractive prices by a high-quality business, it is a direct signal that management sees its own stock as one of the best uses of capital.

Now compare that with many AI-related stocks today.

Prices are going up, but insiders and companies are often selling stock to the market through dilution, issuance, or insider sales.

The message is simple:

They think selling stock at today’s price is attractive.

On one side, you have companies using real cash to buy back shares.

On the other side, you have companies and insiders selling stock into market enthusiasm.

Which side deserves more serious research is pretty obvious to me.

More analysis: https://t.co/nkvFPU9FGC

$VEEV $INTU #SaaS #AI #Software #Investing #Buybacks #QualityStocks #LongTermInvesting

[100Baggers.club @Miltons2Cents]

Just finished my tax filing earlier this month. My CPA noted an interesting shift: the number of people hiring professionals is dropping significantly. More taxpayers are pivoting to TurboTax + AI or DIY + AI.

The market currently has a superficial take on $INTU, blindly lumping it into the "SaaS stocks disrupted by AI" category. Wall Street treats every strong earnings report like a funeral because the narrative is stuck on "AI automation" and "IRS Direct File" threats.

But Intuit’s real value was never just a standalone tax tool.

They sit at a critical intersection that makes them fundamentally different from traditional "tool-only" SaaS companies:

The Integrated Loop: QuickBooks handles the year-round bookkeeping, which creates a natural funnel into TurboTax, while Credit Karma manages financial distribution across the entire year.

The Responsibility Structure: If an issue arises—like an error or an audit—Intuit doesn't just leave the user stranded. They provide a level of accountability and audit support that a simple software tool cannot replicate.

I no longer view $INTU as a "tax season" play. It is a Small Business Financial Operating System with a built-in entry point, distribution network, and liability structure.

From this perspective, AI is an amplifier rather than a disruptor. Intuit already controls the tax, accounting, and credit data, alongside a massive network of pros. These are resources that cannot be easily copied by a single AI model. The real metric to watch is not whether AI can help file a return, but whether this structural moat can continue to lock in cash flow and customer stickiness.

My partner and I recently did a systematic comparison of this logic against $ADBE, $ADSK, and $PTC.

Read the full analysis here: https://t.co/nkvFPU9FGC

[100Baggers.club @Miltons2Cents]

$INTC After Q1, Intel’s story is no longer as simple as “legacy x86 keeps losing relevance.”

DCAI rebounded. Xeon 6 was selected for NVIDIA DGX Rubin NVL8. Google Cloud continues to use Xeon 6. And agentic AI may increase the importance of CPU-side orchestration: tool calls, I/O, runtime management, code execution, sandboxing, and multi-step workflow coordination.

That matters.

If AI workloads move from single model inference to complex task execution, the CPU may not remain just a low-value host sitting next to the GPU. It could become an important bottleneck for system throughput and end-to-end latency.

So the old one-way bear case — that server CPUs will simply be squeezed by AMD, ARM, and NVIDIA’s internal platforms — needs to be revised.

But that does not automatically make Intel a high-conviction buy.

The real issue is not whether Intel has good news. It does.

The issue is that the market may already be pricing several very different things at once: DCAI recovery, an agentic-AI CPU option, government support, Foundry success, 18A catch-up, and potential restructuring.

Those layers have very different levels of validation.

DCAI at least has early financial evidence. But Foundry external revenue is still small, Foundry losses remain heavy, and ROIC has not yet returned above the cost of capital.

To me, Intel is now a high-controversy asset that deserves close tracking — not a stock where one strong quarter is enough to settle the debate.

The key questions from here:

Can DCAI sustain strong growth for several quarters?

Can Xeon attach rate and ASP improve in AI servers?

Can Foundry external customer revenue scale meaningfully?

Will Intel disclose hard 18A yield evidence?

If those pieces start to validate, the thesis can be upgraded.

Until then, I respect the change in the story — but I would not pay too much upfront for evidence that has not yet arrived. $INTC

More analysis: https://t.co/wTcADIUiTe

[100Baggers.club @Miltons2Cents]

$FDX reported Q1 2026 revenue of $24.00B and generated $1.99B in operating cash flow, anchored by a rapid 2-day cash conversion cycle amidst volume headwinds. Net income landed at $1.06B, producing $4.41 in diluted EPS. Capital efficiency metrics remain intact with TTM ROIC at 11.81% and ROE at 33.58%, supported by $1.04B in free cash flow. Trailing twelve-month margins show gross at 24.41%, operating at 6.54%, and net at 4.88%.

The ongoing Network 2.0 integration of the Express and Ground segments is actively compressing operational overlap. The company closed over 200 redundant stations this quarter, tracking toward management's guidance of shuttering 475 locations by 2027 to capture $2.00B in structural savings. At the Memphis hub, the new AI-driven Project Hercules system is now sorting 56,000 packages per hour, reducing manual intervention by 30% and directly impacting the $17.77B in quarterly operating costs.

Segment data indicates Express utilized higher package yields to offset softer macro freight volumes, further aided by a 20% year-over-year drop in fuel costs. Moving forward, management is targeting June 1, 2026, for the completion of the Freight segment spin-off. The company is also directing its $955M quarterly CapEx toward the fdx data platform, planning to expand AI optimization to 65% of its sites as it shifts from physical delivery execution to broader supply chain visibility services.

Full analysis: https://t.co/4sHQEq3aU6

[100Baggers.club @Miltons2Cents]

$OKLO closed Q4 2025 with $0 in revenue but armed with $788.45 million in cash and equivalents to fund its extended regulatory runway. The pre-revenue modular reactor developer reported a net loss of $41.45 million and negative free cash flow of $60.38 million. Operating expenses hit $57.10 million, reflecting heavy capital deployment toward engineering design and regulatory compliance.

Operating cash outflow was measured at $33.43 million, while capital expenditures reached $26.95 million as the company funds initial site preparation and foundation work in Idaho. An ROIC of -128.03 percent maps directly to the structural latency between current infrastructure investments and future power generation. A recent $295.61 million net inflow from financing activities expanded total assets to $1.53 billion, providing the required liquidity to bridge this multi-year commercialization gap while operating with zero debt.

The operational model depends entirely on crossing the regulatory threshold, with management targeting 2027 for its initial commercial reactor deployment. The firm is actively advancing its Custom Combined License Application with the NRC to transition from the laboratory phase to commercial construction. Pipeline development is currently anchored by non-binding intent agreements for up to 500MW of capacity, directly targeting hyperscale data centers and AI infrastructure providers requiring uninterrupted, carbon-free baseload power. Future revenue visibility hinges on converting these capacity intents into binding, long-term Power Purchase Agreements.

Full analysis: https://t.co/n3rlewAFkK

[100Baggers.club @Miltons2Cents]

$AVAV Q4 revenue printed at -$10.80M due to contract accounting adjustments, pulling quarterly net income down to -$156.55M. While total 12-month revenue reached $1.19B, the recent quarter highlights a deep structural transition. Gross margin for the trailing twelve months compressed to 8.78%, and R&D expenses consumed 171.22% of gross profit. This front-loaded cost pressure pushed trailing ROIC down to -8.60%.

A 72.52% year-over-year expansion in outstanding shares reflects the heavy dilution from the BlueHalo acquisition. This move spiked total assets to $5.45B, with goodwill now making up 45.14% of the balance sheet. Operating cash flow came in at -$5.11M, absorbing non-cash hits like an $18.4M goodwill impairment. SG&A also remains elevated at 31.24%.

Operationally, management is focused on bridging the gap between R&D and manufacturing scale. The core initiative is accelerating Switchblade 600 monthly production from 40 units to 240, eventually targeting 1200+ systems. This hardware ramp aligns with the new FreedomWerx facility in Salt Lake City slated for late 2026. Simultaneously, the company is rolling out the AVACORE AI architecture to standardize autonomous flight capabilities. With a cash conversion cycle sitting at 79 days, working capital friction persists as the business attempts to pivot from lab-scale pilot programs to mass defense procurement.

Full analysis: https://t.co/LO2prsOVpE

[100Baggers.club @Miltons2Cents]

$ULTA generated $1.19B in operating cash flow over the 13 weeks ending January 31, significantly outpacing its $356.68M net income. Revenue reached $3.90B, an 11.8% YoY increase. Comparable store sales grew 5.8%, heavily weighted toward price adjustments with average ticket up 4.2% while transactions rose just 1.6%. Diluted EPS came in at $8.01.

The cost of this top-line expansion materialized in profitability metrics. Operating margin compressed to 12.2% from 14.8% last year, and gross margin ticked down to 38.1% due to increased promotional frequency and fixed cost deleverage.

Operationally, the company is shifting its customer acquisition channels. The recent integration with TikTok Shop and the Space NK acquisition, which added 86 UK and Ireland locations, reflect a push beyond physical US stores. Inventory levels increased 10.8% to $2.18B to support these new brand launches and the rollout of Market Fulfillment Centers aimed at sub-24-hour delivery.

Looking ahead, management is prioritizing digital retention among its 46.7M active members as the Target partnership winds down in 2026. The PRISM personalized AI skincare recommendation system is slated for full app integration in 2025. Capital efficiency remains intact, highlighted by a 28.13% ROIC and a 4.11% TTM net buyback rate.

Full analysis: https://t.co/YzHfWv3VzR

[100Baggers.club @Miltons2Cents]

$CPRT generated $1.12B in quarterly revenue and a 33.76% net margin, translating to $350.73M in net income despite a 1.9% dip in global insurance volume. Higher average vehicle selling prices successfully offset this volume decline across the global auction network.

The underlying unit economics show an 8-day cash conversion cycle and $127.50M in operating cash flow. After accounting for $69.62M in capital expenditures for physical yard network expansion, free cash flow landed at $57.88M. Asset efficiency remains high with ROIC hitting 29.11%, while gross margins hold at 45.34%.

Operations highlight a dual focus on physical capacity and digital integration. While physical land expansion continues, management is rapidly scaling the Total Loss Express 360 AI damage estimation system, planning deployment in the UK and Germany by June 2026.

With modern vehicle complexity driving the US total loss frequency up to 22.6%, this AI rollout allows the company to capture value earlier in the insurance claims process rather than relying solely on backend auction commissions. The balance sheet provides absolute flexibility for this ongoing infrastructure buildout, featuring $5.10B in cash and equivalents and a 10.06 current ratio, enabling uninterrupted global logistics expansion without requiring external leverage.

Full analysis: https://t.co/GOqogZzJKz

[100Baggers.club @Miltons2Cents]

$ORCL reported a massive divergence between its record $523B in remaining performance obligations and near-term liquidity, driven by an aggressive AI infrastructure build-out. Total revenue reached $16.06B for the quarter ended Nov 30, with cloud now accounting for 50% of the top line. Net income came in at $6.13B, which includes a $2.7B non-recurring pre-tax gain from an Ampere stake sale. Gross margin held at 70.70%, with Non-GAAP operating margin at 42%.

The underlying business mix is shifting rapidly. IaaS revenue jumped 68% YoY, fueled by a 177% surge in GPU-related consumption. Multicloud consumption skyrocketed 817% YoY as the company expanded its footprint across 34 active regions, integrating database instances with AWS, Azure, and Google Cloud. Conversely, traditional software licensing declined 3% YoY to $5.9B.

This physical scale-out requires heavy upfront capital. Capital expenditures hit $12.03B for the quarter, significantly outpacing $2.07B in operating cash flow and pushing free cash flow to -$9.97B. Looking ahead, management raised FY26 capex guidance to $50B, up from initial $35B estimates, to deploy over 20 new data centers and expand superclusters. The core dynamic remains the temporal lag between heavy upfront fixed-cost GPU investments and the eventual realization of consumption-based cloud revenue from their contract backlog.

Full analysis: https://t.co/to57aRSDO9