The Portfolio That Kills Your Vibe

Side Note: I vibecoded a tool while writing this, as it sparked my interest, reading through reddit, youtube comments, a huge amount of ‘how do we find the actual value stocks’ came up. I’m not a stock picker, this is not financial advice, and is most likely very wrong. My portfolio is laughably small, but this tool show’s what passes the dhando method.

Value Add Is a tool to help find value stocks.

https://valueadd.extrastrength.co

Anyway, what I wanted to say was there is a particular type of person who has the Munger quote on their LinkedIn banner, a Berkshire stake in their retirement fund, and a pitch deck open in another tab for a B2B SaaS tool aimed at a TAM of 200 companies.

They are not embarrassed by this.

Buffett and Munger canon gets invoked constantly, even in my day to day circle.

‘Circle of competence.’

‘Be greedy when others are fearful.’

‘A moat.’

Everyone has read Poor Charlie's Almanack, or if you’re like me, its in the back of every zoom call, gathering dust, unopened.

Everyone has the Berkshire shareholder letters somewhere in a Notion doc they'll get back to.

The actual portfolio gets ignored.

What Berkshire owns is not a mystery.

It's publicly reported every quarter. And it reads like a list of things a 2026 YC founder would galk at and tell you to get with the times.

Where is the moat, I can hear them say.

Coca-Cola. See's Candies. GEICO. Burlington Northern Santa Fe Railroad. Dairy Queen. Pilot Flying J truck stops. Nebraska Furniture Mart. Precision Castparts. NetJets. Clayton Homes.

Not one of these is a software company.

Not one of them has an AI roadmap, a PLG motion, a community-led growth strategy, or an Acquired podcast episode.

Every single one is a boring, defensible, physical or financial business that generates cash reliably for decades.

These aren't consolation prizes.

$377 billion silence

While the rest of the market was building AI wrappers and debating which foundation model wins, Berkshire ended 2025 holding somewhere north of $377 billion in cash.

That's not a number that happens by accident. That's posture.

Buffett has been a net seller of stocks for twelve straight quarters. He's watched the S&P 500 surge around 25% in both 2023 and 2024. He's watched valuations hover at or near levels last seen in 2000 and 2021. He's watched the AI frenzy produce a particular kind of founder confidence.

And his response has been to do nothing.

Which, if you understand the framework, is ‘the move’.

WSB has me giddy with my AI gains, but I know I’m wrong.

Pabrai describes the ideal investor as someone who can "enjoy watching paint dry." He estimates well under 1% of stock-picking Americans are actually good at it.

The primary mistake smart people make is impatience. A great investment might do nothing for three to five years. Most people can't hold through that without convincing themselves something is wrong.

The corollary for operators building businesses is identical.

A great business might not make the trade press for a decade.

The business that prints cash, compounds quietly, and produces zero conference invitations might be the best business in your city.

Most people won't build it because they'd have to admit, out loud, that that's what they're doing.

What's in the box

Berkshire's five largest equity positions at end of 2024:

• Apple
• American Express
• Bank of America
• Coca-Cola
• Chevron

This accounts for 70% of the portfolio concentrated in five names.

Apple is worth pausing on, because Buffett sold most of it.

They framed it as a consumer franchise bet, not a tech bet.

He was buying the installed base, the ecosystem switching costs, the brand loyalty built over decades. Not a software company. A product that hundreds of millions of people will not stop using regardless of what gets released next.

I see you green bubbles.

When the valuation no longer matched the underlying franchise economics, he sold. Quietly. Into the rally.

The wholly owned subsidiaries are where the argument really lives.

Berkshire bought See's Candies in 1972 for $25 million. See's had $8 million in net tangible assets and $2 million in after-tax earnings at the time. 25% return on invested capital, from a California boxed chocolate company.

Buffett said he and Munger would have walked away if the seller demanded a dollar more. $25 million was the ceiling.

By 2007, See's had sent $1.35 billion in earnings back to Berkshire.

8,000% return.

Candy.

This wasn’t because ‘Disruption’ . This was because they had pricing power, loyal customers, and almost zero capital requirements to keep the business running.

Munger later said they discovered they could raise prices by 10% a year and nobody cared. He wasn’t telling the truth. What happened was they raised prices 0.8 percentage points above inflation per year for 35 years. Modest. Consistent. And that margin, compounded over decades, doubled the profit margin while unit sales also doubled.

That's compounding. Not scaling.

Burlington Northern Santa Fe. You want to move freight by rail across the western United States? BNSF is the option. There is no competing rail network covering the same geography. This is the rockefeller, play, updated. Nobody is building one because nobody can. Buffett called it an "all-in wager on the economic future of the United States." It's also a wager that the laws of physics prevent anyone from replicating or ‘disrupting’ the asset.

Berkshire has loved coke, like, Coca-Cola, since 1988. The syrup formula is almost irrelevant. The moat is that Coke distributed itself into 200 countries over 130 years and formed a taste habit in most of its customers before those customers were old enough to make a considered purchasing decision.

You don't dislodge Coke by building a better-tasting product. You'd have to change a behaviour formed at age 8, at scale, across 200 countries. This is a civilisation problem many a documentary has been created about. They proved it themselves with ‘New Coke’, ahhhh sweet, sweet, ‘disruption’.

Then GEICO was out there making money on the float between when you pay your premium and when you have an accident. Float gets invested. Moat is scale: the bigger the customer base, the more driving behaviour data, the better the model, the more competitively you can price against smaller insurers who can't afford to match you. Every new customer makes the next customer cheaper to acquire and more accurately priced.

These businesses don't need to be managed into growth.

They have structural properties that make next year slightly better than last year, by themselves, without proportional reinvestment.

Lunch Time

Mohnish Pabrai paid $433,000 for a third of a lunch with Warren Buffett.

He and Guy Spier bid at a charity auction. Pabrai was already running around $600 million and compounding at roughly 28% per year since 1999.

He'd read everything. He'd structured his fund after Buffett's original 1950s partnership model. He understood the philosophy as well as anyone alive who wasn't Warren Buffett.

He still paid six figures for the meal.

Pabrai calls himself a shameless cloner.

He means it literally.

His method is finding businesses with Buffett characteristics and buying them when the market is pricing them as if those characteristics don't exist. He points to Sam Walton as the most successful cloner in retail history.

Walton visited competitors, took notes, adopted everything that worked, executed it with more discipline. Cloning was the strategy. Most people are too proud to do it, which leaves a persistent advantage for people who aren't.

His actual filter for investments is that if you can't explain it to a ten year old in four sentences, pass.

If you need a spreadsheet to prove it works, the margin of safety is too thin.

It needs to hit you over the head like a 2x4.

He notes that Buffett doesn't use Excel. If the deal requires Excel to justify, it's the wrong deal.

That filter is also a near complete description of why most SaaS investment cases don't pass.

Detailed cohort analysis. Net revenue retention modelling. CAC payback calculations. Five-year ARR projections building toward a number that justifies what the business is being asked to be worth. You need all of it, because the business itself can't tell the story in four sentences without the spreadsheet.

Pabrai says pass.

His 2026 positioning reflects the same logic. While most of the market is in what he calls an AI frenzy, he's backing the pickaxe makers. The companies making the physical infrastructure the AI gold rush runs on.

He's also buying Turkish warehouses and, shocker, coal. Things that are "hated and unloved" by the market. Things with physical reality. Things that can't be replicated with a GitHub repo and six months of runway.

His play is simple, undeniable value at an obvious discount. Not "we think this could be a really interesting space if enterprise AI adoption accelerates over 36 months."

A moat around a pitch deck

The software industry settled on a story about marginal cost.

Build once, distribute to a million customers, serve each additional customer for almost nothing.

80% gross margins.

Revenue that scales without proportional cost increase.

This is all true at the platform level.

The marginal cost argument cuts both ways.

Near-zero marginal cost to serve also means near-zero marginal cost to replicate.

Your SaaS product costs almost nothing to copy. A well-funded competitor raises a seed round, hires four engineers, ships something comparable in six months, and undercuts you on price.

They can do this because there's no physical infrastructure to establish, no supply chain that took decades to build, no customer habit formed in childhood.

A moat built on features is a lead time.

That's all.

Coca-Cola's moat widens every year without Berkshire lifting a finger.

The more people who grew up drinking it, the more customers who default to it in the next decade, the more retailers who stock it, the more supply chain infrastructure that exists globally to serve it.

Berkshire doesn't have to spend to maintain the Coke moat. It holds.

Most SaaS moats erode the moment you stop spending on them.

Sales motion to acquire new customers.

Marketing to stay visible.

Product investment to maintain feature parity against the competitors who shipped last quarter.

Customer success to keep churn inside a number that doesn't sink the model.

Stop spending, start losing, lke a treadmill at speed, not a compounding machine.

I’m a hater, but the exceptions are real. Salesforce has genuine switching costs because it sits at the centre of how organisations run revenue operations. Workday has them because tearing out an HR system is a multi-year project that nobody in the organisation will vote for. Shopify has them because the ecosystem of integrations, apps, themes, and trained staff creates migration friction that's genuinely painful. Real moats. Also took billions in capital and more than a decade each to build. And they represent a vanishingly small slice of what gets pitched and funded every year.

The median seed stage SaaS company has a clean UI, a competitive feature set, and churn they're calling product-market fit iteration.

70%

Around 70% of SaaS startups fail within five years. Some estimates put the three-year failure rate at 92%.

We see Stripe. We see Salesforce. We see Shopify.

We don't see the 20,000 companies that raised seed rounds, got to $200K ARR, burned two years of runway trying to crack $1M, and quietly wound down. The founders wrote a thoughtful retrospective on LinkedIn about what they learned and went to go work at Series B companies. The investors marked it to zero and moved on to the next batch.

That's survivorship bias at full tilt.

The base rate for SaaS is brutal.

The base rate for a well-run regional pest control company with recurring contracts is almost the inverse.

Recurring revenue. Low churn, because switching pest control providers is genuinely the last thing anyone thinks about, or they get one too many invoices.

A customer base that stays until they move or close, and even then, they are passed onto the next occupant. Defensible territory because the economics of competing in your geography at your scale don't work for anyone who isn't already there. A business worth a steady 4 to 6x EBITDA to any buyer who understands the category.

The founder of that business is quietly wealthy and completely unknown.

So is the uniform rental operator. The regional trucking fleet. The commercial maintenance company holding the service contracts for every major building in a mid-sized city. These businesses print cash, compound on their own, and produce no content.

SaaS success produces a founder who goes on podcasts. Gets profiled. Speaks at conferences about PLG and hiring senior salespeople and the three lessons they'd tell their earlier self.

The celebrity is part of the output.

The ecosystem has confused visibility with success so thoroughly that people optimise for the wrong thing. They're building for the profile, not the outcome, and they've convinced themselves these are the same.

Pabrai has a term for what the best operators have instead. He calls it the inner scorecard.

Measuring yourself by your own standards rather than what the world thinks of you. Most people are running on an outer scorecard without knowing it.

They've optimised for the conference invite. For the profile. For the round announcement. For the thing that tells them, via external signal, that what they're doing matters.

The businesses worth building don't come with any of that.

Hello Operator

I've spent most of my career inside businesses that don't make it into business media.

B2B distribution. Foodservice supply. Office supplies at scale. The economics of moving physical product to customers who don't have time to think about who they buy from, so they stay with whoever has already earned their trust.

You already know what I'm talking about if you've worked inside any of this.

You've watched the FMCG brand everyone's heard of operate on 4% net margins while the logistics company nobody's heard of makes 18%.

You've watched the restaurant with the line out the door lose money while the company supplying their packaging to 400 restaurants across three cities makes 30% EBITDA.

You've watched software vendors obsess over TAM while the distributor serving a specific vertical in a specific geography quietly builds something that's nearly impossible to displace because the switching costs are real, the relationships are real, and the operational knowledge is years deep.

The Berkshire portfolio is a list of those businesses.

Pilot Flying J truck stops. Not the freight companies. The stops the freight companies have no choice but to pull into.

BNSF. Not the companies shipping on it. The rail network they must use.

See's Candies. Not a confectionery conglomerate. A regional California brand with customer loyalty built over 130 years that raised prices 0.8 points above inflation for 35 years and compounded into a $2 billion earnings machine.

These businesses have a Buffett investment case you can explain to a ten year old.

The SaaS pitch deck is 40 slides.

So What.

The principles the investing class worships point directly at the unsexy physical businesses the tech ecosystem is embarrassed to build.

If you're evaluating what to build next, the useful question is not "does this scale." It's "does this compound."

Scaling is a feature.

Compounding is a structural property.

Coca-Cola didn't scale in the SaaS sense. It compounded. Every decade of distribution made the next decade easier. Every generation of customers made the next generation more likely. The business got better by existing, not just by being managed into it.

The businesses that compound tend to have a few things in common. Customers who don't really think about switching. Defensible territory because the operational or relational economics of competing in your market don't work for the person who isn't already there. Economics that get marginally better over time without proportional reinvestment.

Pabrai paid $433,000 for a third of a lunch to sit across from the person who articulated this best.

The articulation is free.

It's in every Berkshire shareholder letter.

It's in the Dhandho Investor.

It's in fifty hours of Pabrai interviews.

Shout out to all the legends I've ripped info from for this piece:

Berkshire Hathaway 10-Q | GuruFocus on See's Candies | SBO Financial Teardown | Morningstar on See's pricing | Quartr on Pabrai | NBC on Pabrai Lunch | Investing.com on Berkshire Cash | Britannica on Berkshire | SaaSy Brands on failure rates | Kiplinger on Berkshire portfolio