The Bull Case for Software in the AI Era

December 4th, 2025

Friends,

We are excited to share the edited excerpts of our October 9th Ideas & Networking Conference fireside chat with Alok Agrawal, the founder of Bloom Tree Partners.  Bloom Tree was a long/short hedge fund seeded by Tiger Management, reaching peak assets of $2 billion.  Alok began his investment career at Bessemer Venture Partners, where he led investments into companies such as Skype.

I found my discussion with Alok to be one of my most valuable to date.  We explored Alok’s background, including stories and lessons from Julian Robertson.  We also discussed the merits and challenges associated with the current massive boom in AI capex, looked at the potential for out-of-favor SaaS companies, explored short selling in the current market, and reviewed Alok’s healthcare AI startup.

The interview with Alok includes numerous actionable investment ideas as well as a fun lightning round format at the end.  I am confident you will find value in this interview.  I also encourage you to follow Alok on Twitter, he has tremendous insights: @AlokANYC

Have a great holiday season.  Enjoy!

Best Regards,

William C.  Martin

Topics in this Issue of An Entrepreneur’s Perspective:

**

Interview with Alok Agrawal, Founder Bloom Tree Partners: The Bull Case for Software in the AI Era

This fireside chat originally occurred on October 9, 2025.   ChatGPT was used to format and lightly edit the original discussion, using a “formal conversation” prompt. 

Welcome, Alok!  Tell us a bit about yourself.

It’s been a journey defined largely by serendipity.  I actually grew up in Zambia, spending the first eight years of my life in the rural Copperbelt region before returning to India after the copper market collapsed and crime surged in the late 1970s.  I completed my undergraduate studies in computer science in India, intending to pursue research in computational geometry — a highly specialized field.

I applied to three U.S. universities, each hosting a professor I wanted to work with: Princeton, UC Berkeley, and the University of Illinois at Urbana-Champaign.  I was admitted to all three and chose Berkeley because it was ranked highest in U.S. News & World Report.

The weather probably didn’t hurt.

I honestly didn’t know anything about the weather — it was the pre-internet era.  I began working toward my PhD, but shortly afterward, my advisor left for Germany.  I shifted my focus to computer animated design for VLSI, which is technology used to design semiconductors, and worked with Alberto Sangiovanni-Vincentelli, a renowned professor at Berkeley.

Halfway through my PhD, I realized teaching wasn’t for me.  I taught one class and found it exhausting, and the students certainly didn’t appreciate my strict grading.  That marked the end of my academic ambitions.  I took a leave of absence, joined Oracle, helped build several products, later attended business school, and eventually moved into finance.

You then worked at Bessemer in Boston, right?

Yes.  My plan was originally to return to Silicon Valley and start a tech company — I had even written a business plan.  But 2002 was not a favorable time for raising startup capital — the market was still reeling from the dot-com crash.

Around that time, Bessemer Venture Partners reached out through the HBS network, and I joined them.  It was a great way to stay connected to the startup ecosystem and continue exploring ideas.  I worked there for about three years.  And, as I often say, my life has been a series of serendipitous turns.  One of those turns came when I had a chance meeting with Julian Robertson — the legendary founder of Tiger Management, one of the most well-known hedge funds in the world.

How did you have a chance meeting with Julian Robertson?

It’s actually quite a story.  A former classmate from business school — someone I wasn’t particularly close to — had worked at the original Tiger Management before it was wound down in 1999.  That was the year Julian experienced a significant drawdown, losing about -30% due to being short dot-com stocks.  In some ways, today’s AI bubble reminds me of that period, though there are important differences.

After that loss, Julian transformed Tiger into a family office.  My classmate believed I would make a good investor and persistently encouraged me to meet Julian.  Initially, I had no idea who he was.  Eventually, once I understood his reputation, I agreed to a meeting.  One thing led to another, and I was offered a position on his core team.

I ended up covering all of his private deal flow — though it was fairly limited — so the majority of my time, about 90%, was spent in public markets.  Eventually, Julian seeded me to start Bloom Tree.  In other words, I became a hedge fund manager almost entirely by accident.  And now I’ve come full circle, returning to the technology sector.

That’s remarkable.  Julian was quite a character, and there are many stories about him.  Are there any that stand out to you?

There are numerous stories.  He was indeed a colorful personality.  I could share lessons I learned from him or simply anecdotes.  One key trait of Julian’s was his meticulous attention to management teams and the individuals we partnered with.

For instance, I once owned a title insurer — FNF.  Julian discovered this and called me immediately.  He said, “Do you know Foley wronged me on a deal?” I had no idea what he meant.  It turned out that Julian and Bill Foley had a falling out years earlier over a vineyard transaction in New Zealand, and Julian had placed him on a permanent blacklist.  That exemplifies his insistence on integrity: if he suspected even a slight issue, he avoided involvement.

That aligns with a story I heard recently at a John Griffin event — about a CFO who cheated at golf.  Julian never conducted business with him again.  What, in your view, made Julian such an exceptional investor?

I believe Julian’s skill lay in knowing when to size positions and when to allow them to grow.  In hedge fund investing, your batting average — how often you are correct — is important, but the real driver of returns is your “slugging percentage” — the magnitude of gains when you are right.  Julian mastered that.

He also had an extraordinary ability to read markets.  One example illustrates this perfectly.  After he seeded my fund, we were technically independent, but shared the same building.  For nearly 18 years, I worked on the same floor as Julian, seeing him frequently, including at the gym.  Eventually, he instituted quarterly or semi-annual breakfast meetings with all the funds he had seeded, which my team and I attended.

During one such meeting, an analyst pitched a healthcare company called Envision Healthcare.  Julian remained silent during the pitch but later summoned me to his office, saying, “This stock is trading like death.  Why do you own it?” I didn’t fully understand at the time, but I trusted his judgment.  I instructed my team to sell immediately, stating we would revisit the position later.  Upon further analysis, I discovered a Yale University study indicating Envision had been overbilling managed care companies.  A few months later, this was confirmed publicly, and the stock declined by -50%.

Julian had identified the issue simply by observing how the stock was trading.  That intuition, combined with disciplined portfolio management — he frequently held 10% positions and maintained gross exposures exceeding 300% — made him exceptional.  He knew when to let winners run and managed risk with remarkable precision.

You launched Bloom Tree and ran it for 15 years.  Can you discuss the strategy you applied there and the types of investments you focused on?

We ended up running what is known as a low-net, or market-neutral, fund.  That wasn’t necessarily my original intention — it was more an artifact of the circumstances.  Julian would typically seed only one, sometimes two, funds per year.  Being selected as a new Tiger seed carried tremendous scarcity value, and capital could grow quickly — from zero to $25 million, and then to $1 billion, in a short period.

I launched Bloom Tree in May 2008, which, in hindsight, was right at the top of the market that year — talk about timing! I had a very strong track record with Julian, having managed a book for him in 2007.  By the time I got all the fund documentation in order, it was May, and I was fully invested on day one.  It worked out well initially, but just a few months later, the fundraising market essentially shut down due to the financial crisis.

Julian had just come off the two best years of his career, 2007 and 2008, but he had taken his own capital down considerably after a heart attack in 2007.  He managed only a few hundred million at that point and was looking to build a seeding business.  By the time I was ready to market my fund, there were already 20+ “Tiger Cubs” seeking capital, making it difficult to stand out.

I was skilled at short selling and focused on not losing money.  Despite the market experiencing a -30% drawdown in 2008, Bloom Tree was actually up slightly over the peak-to-trough period.  That consistent focus on alpha positioned us as a low-net fund.  Our portfolio maintained around 20% net exposure on a dollar basis.  We were generally long large-cap companies and short smaller-cap firms — not because I preferred shorting small caps, but because those companies often presented fundamental issues.  This approach created a negative beta on the short side, resulting in a beta-adjusted, market-neutral portfolio.

You decided to wind down Bloom Tree a few years ago.   Tell us about that?

Yes.  I ran the fund for 15 years.  The first really challenging period came in early 2021, which wasn’t easy.  We were caught in several significant short squeezes.  One investor joked there should be a term for it — the “Cathie Wood factor” — because we were short six of her companies, and I hadn’t fully realized it at the time.  We were also short three of Bill Hwang’s companies.  Between those nine positions, many of them tripled in value over just three months, with some, like GSX, eventually going to zero.  It was a painful period.

I also had personal reasons to take a step back.  By 2022, I spent a lot of time considering what I wanted to do next.  As a single portfolio manager, the only way to take a break was to return outside capital, which is what I did.

Following up on the short-selling side — since the Tiger approach emphasizes it and takes pride in it — do you think the landscape has structurally changed with the rise of retail investors, zero commissions, and trading apps, particularly from a risk management perspective?

Absolutely.  One of the simplest ways to avoid these short squeezes is to avoid certain types of short positions altogether.  Conceptually, short selling is inherently difficult.  The cost of capital works against you.  I always reminded my team — and communicated to investors — that if you own a collection of strong businesses, you earn the underlying equity cost of capital over time.  That’s a principle I learned from Julian.  I’ll never forget something he told me when I started with him: If you buy 20 good businesses and short 20 bad businesses, and you haven’t made money in two years, you should do something else — investing isn’t your forte.

It’s a simple statement, but it reflects a profound concept.  Equities have a cost of capital, which is essentially the underlying compounding rate.  That’s why the S&P total return index compounds at roughly 8.5% annually over the long term.  If you own companies that can deliver 8–9% compounded over time, you’ll do very well.  Shorts, on the other hand, are costly: you lose the equity cost of capital – 9% to 10% a year on short small-cap positions, plus borrowing costs, which can be substantial, particularly in recent years when interest rates were near zero.  There are, of course, different strategies for short selling, but it’s inherently challenging.

The takeaway is…don’t short sell?

Short selling is difficult! Capital compounds over time, and that’s the key.  For certain investors, including mine, the objective was to deliver an uncorrelated, low-beta return stream.  They would leverage that by combining multiple managers like me into a broader portfolio, which helped manage overall risk.

Since you’ve closed your fund, have you continued shorting stocks?

No, I don’t see the point.  Short selling requires an enormous amount of work.  At Bloom Tree, we managed a book with 25 longs and 40 shorts.  Each short required roughly twice the effort of finding a good long position.  Longs are relatively easier — you just need to find strong businesses.

Guest Q&A: Earlier, you mentioned you would avoid certain kinds of shorts.  Could you expand on that?

I don’t short “stories” or hope, and I don’t short great businesses.  That was a lesson I learned from Julian Robertson.  My shorting framework has always been centered on terminal value — you need a differentiated view of what the business will eventually be worth.  Then you need a realistic catalyst, and for most shorts, that catalyst ends up being earnings.

We focused on three categories for shorting: new competition that could severely compress a company’s gross margins; frauds; and fads.  For example, we were among the first to short Wirecard, though we actually lost money initially.  Beyond that, shorting “hope” is extremely difficult.  Time works against you.  Every year a short position is held, you’re losing 8–10% to the equity cost of capital, and that makes shorting a high-risk proposition.

How active are you on the long side personally? How does your personal strategy compare to your previous fund?

Mostly, my personal portfolio is allocated to ETFs.  I needed a break from active stock picking, and one way to achieve that is to step back from individual names.  That said, I do occasionally buy individual stocks when they become extremely cheap.  For example, I recently purchased UnitedHealth Group (NYSE: UNH), and over the past couple of weeks, I’ve also bought several software companies.

I tend to buy when valuations are absurdly low.  Of course, there are tax implications — selling an ETF in New York City can be expensive.  But if you see a compelling risk-reward setup with potential for a meaningful return over a few years, it’s worth it.  That was the case with UnitedHealth, and it’s the same rationale behind my recent software and SaaS purchases — a sector that’s currently out of favor but presents attractive opportunities.

I’d love to hear more about that.   Also, last year at these event you shared some bearish ideas on AI capex.   Capex is up a bit since then.

I was skeptical last year, and I still am to an extent.  NVIDIA (NASDAQ: NVDA), for example, was trading around $135 back then — it’s $185 now.  My thinking is grounded in simple math, which I’ve updated given Jensen Huang’s strong execution.

Last year, my calculation was that for every dollar spent on GPUs, you need roughly another dollar for the other components of a data center — DRAM, networking equipment, and so forth –depreciated over five years.  That’s already $2 in total costs, and I’m not even including the real estate for the facilities.  Historically, the infrastructure layer, like Azure, runs at a roughly 60% gross margin — a 2.5x markup on cost of goods sold.  Then, the SaaS or software layer runs at roughly 75% gross margin, a 4x markup.  Multiply those together, and you’re looking at roughly a 20x revenue multiple across layers.

If you adjust for other AI use cases — internal AI products at Facebook (NASDAQ: META), which is an 80% gross margin business — you get a blended multiplier of roughly 10–12x.  To put this in context, AMD (NASDAQ: AMD) and Intel’s (NASDAQ: INTC) data center CPU sales over the past 6–7 years averaged about $65 billion per year and supported a $2 trillion-plus software and e-commerce ecosystem.  With AI, I think the multiple should be lower.  Even with a conservative haircut, my estimate is a 10x multiplier.

Looking forward, NVIDIA’s forecasted GPU sales are about $160 billion; adding in TPUs, AMD, and others brings us to roughly $200 billion.  If that grows 20% annually for the next three years, a 10x multiplier implies $2 trillion in application-level revenue is required to justify the investment.

I see a path to $400–500 billion in application revenue from ROI improvements alone at companies like Facebook.  Their capex-to-revenue ratio has gone from 20% to 43%, and historically, Facebook has been an 80% gross margin business.  As capital spending scales, revenue growth needs to follow, or they have to replace operating expenses with capital expenses, essentially reducing headcount and automating tasks with AI.  Some of that will happen, but the bottom line remains: $2 trillion in application-level revenue is required.  This doesn’t include revenue at the infrastructure level from OpenAI or Anthropic.  So while people get excited about $13 billion at OpenAI or $4 billion at Anthropic, that alone won’t move the needle.

So how do you play this view?

Instead of shorting NVIDIA, I ask myself: what can benefit from all this capital spending? That’s where SaaS comes in.

SaaS has been under pressure due to AI narratives — the idea that seat-based pricing will vanish and AI will replace many applications.  But if you look at the fundamentals, SaaS is extremely sticky.  I was an early investor in SaaS — I remember owning NetSuite (now part of Oracle (NASDAQ: ORCL)) and Salesforce (NASDAQ: CRM) when most others were short.  The good SaaS products are deeply embedded in workflows and serve as the primary repositories of a company’s critical data.  Replacing them is extremely difficult.

The best metric to assess this is gross retention.  The difference between 70% and 97% gross retention is massive: 70% implies a three-and-a-half-year customer life, 97% implies a 30-year customer life.  For example, Workday (NASDAQ: WDAY) has 97% gross retention — it holds core HR and financial data — making replacement nearly impossible.

So, I believe a portion of the $2 trillion in potential application-level revenue will come from new AI startups.  There are clever companies using AI in transformative ways — much like what I’m building — which wouldn’t be possible without AI infrastructure, but there’s still a large gap between capex spending and realistic application-level revenue.

I believe there are certain software companies with exceptionally sticky products that will benefit from AI in two distinct ways.  The market isn’t fully pricing this in yet.  Of course, the key is to focus on companies with strong, capable management teams — those are the ones most likely to capitalize on these opportunities successfully.

You mentioned Workday — are there any other names you like?

There are several.  I recently bought UiPath (NASDAQ: PATH), which has been a somewhat controversial pick — the stock went from $80 down to $12, and I bought it at $12.  UiPath has 97% gross retention.  If you compare it to the little demo Sam Altman showed recently about connecting agents, it’s essentially child’s play compared to what UiPath’s platform can actually do.  The company has been building graphical workflows for years, and that level of sophistication is hard to replicate.

Another company I like is HubSpot (NASDAQ: HUBS).  Its attrition is slightly higher, but I really like the management team — they’re aggressive and capable.  These companies benefit in two ways that the market doesn’t seem to be pricing correctly.  First, they can create entirely new products and solutions.  And second, distribution remains the critical factor.  Building basic software is one thing, but integrating it, distributing it, and tying it into customers’ workflows takes decades of expertise.

That’s a compelling thesis.  Do you own Bill Holdings (NASDAQ: BILL) as well?

Yes, but for a different reason.  My investment in Bill relates more to Ramp and how well it’s performing.  Bill acquired Divvy, and while I’m not particularly bullish on their core business — it’s not as sticky as Workday — it seems they have a clear opportunity to scale Divvy much faster.  I bought it in the mid-$40s.

Many of these companies are trading at what I’d call “run-off” valuations.  Take Workday: at today’s valuation, you could literally run the company for cash and do fine.  That implies that either their R&D spending is inefficient or it’s going to generate significant returns.  I believe the latter, given the quality of their management.

Another factor the Street isn’t fully modeling is AI-driven productivity.  I’ve been experimenting with these tools myself — after 25 years, I started programming again for fun.  I knew nothing about Python, but using tools like Cursor, I’ve written 50,000 lines of code in the last six months.  The capabilities are astounding.  For a SaaS company, with 75% gross margins and typical spend on R&D and sales and marketing, there are enormous opportunities to cut costs, accelerate growth, and create new products for existing customers.

Guest Q&A: Other than NVIDIA, among the trillion-dollar market cap companies, would you pick one long and one short?

Among the trillion-dollar companies? I wouldn’t pick a short — philosophically, I don’t short great companies.  Valuation shorts are particularly difficult.  Historically, I’ve been a contrarian on Google (NASDAQ: GOOG), but more because I own a lot personally.  I’ve held Google shares since the IPO, and my wife worked there for eight years — I refused to let her sell any shares.

Google occupies a very unique position.  It’s not just their distribution advantage.  With Gemini 3.0 reportedly coming soon, it could be a game changer.  In my view, top LLMs are converging to a similar capability level, and the models themselves will be commoditized over time — high fixed cost, low marginal cost, much like telecom networks.  The real value will be in the application layer on top of LLMs, because there’s no effective way to price discriminate at the base model level.  A $20/month ChatGPT plan serves both a parent asking for bedtime stories and a PhD researcher generating high-value output.  The base LLM is the same in both cases.

That’s why I’ve been very bullish on Google Cloud over the past five or six years.  They have TPUs, and the price-performance of their hardware is exceptional.  While the market believes the value lies in hardware, companies like NVIDIA have commoditized the infrastructure layer.  CoreWeave (NASDAQ: CRWV) operates at roughly 20% gross margin — maybe even lower — and Oracle is at 16%, according to recent leaks.

Google, however, makes its own TPUs and benefits from a huge cost advantage.  There’s essentially an embedded “NVIDIA” inside Google that the market isn’t valuing.  When you consider the economics of free search, low-cost infrastructure is critical.  That’s why Google became an infrastructure company in the first place.  By contrast, if OpenAI tries to monetize LLMs on top of NVIDIA’s chips with 80% gross margins, the math simply doesn’t work.  The cost structure is just too high.

Tell us about the healthcare AI startup you’re working on.

I’m building a product that I think could be a game changer, though I’ll admit it’s a very challenging idea to execute.  If it works, it has enormous potential.

The problem I’m addressing relates to managed care companies.  Despite their name, they don’t really manage care — they manage spend.  There’s no comprehensive solution in the market today to help individuals actually manage their own healthcare journeys, or the health journeys of those they care for.  In the U.S., there are 50 to 60 million caregivers — people taking care of aging parents, children, or family members with chronic illnesses.  Yet, when was the last time anyone actually used their insurance company’s app? Rarely, if ever.

The business model of companies like UnitedHealth or Cigna (NYSE: CI) is fundamentally about payments and cost management, not health management.  That gap motivated me to think about a product that could truly support people in navigating healthcare.  I’d been considering several ideas after stepping away from my fund, and this one struck me as having the largest upside.

If you look at major consumer categories, everyone has at least a $100 billion-plus company.  Healthcare is the exception, despite being the largest consumer spend category in the U.S.  — around $4 trillion annually.  HIPAA and privacy regulations make building consumer-facing products challenging, but recent legislation in 2021 now allows individuals access to their electronic health records via FHIR standards.  That opens the door for innovation.

What we’re building is essentially an AI agent — initially called a medical coordinator, now a medical copilot.  It’s software designed to help you manage your own health journey, or that of someone you care for.  Think about all the tasks a caregiver handles: post-surgical monitoring, medications, symptom tracking — today, it’s all manual.  Our product automates much of this by integrating a person’s medical history and providing actionable guidance.

The plan is to offer the product for free to maximize adoption.  Distribution is a key focus — we’re in early discussions with a smaller Medicare Advantage player to explore co-branded distribution.  Monetization isn’t the challenge; the real goal is widespread usage.  Once people are engaged, there are many potential ways to generate revenue.

Let’s do a quick one- or two-minute lightning round:

  • Tim Cook: Should go.
  • Amazon Alexa: Dead.
  • Zuckerberg’s AI Strategy: Smart, with multi-faceted AI potential; could succeed if investments are focused.
  • OpenAI vs.  Gemini: Mostly using Gemini — one-tenth the cost, comparable performance.
  • SoftBank: Haven’t looked.
  • Grok: Hard to bet against Musk.
  • Tesla: Core business now about self-driving.
  • India’s Modi: Likes him.
  • Ukraine War: Should never have happened.
  • NYC Mayoral Election: Depressing; shelved apartment plans.
  • Congressional Midterms: Optimistic; administration making right moves.
  • Favorite Trip: Malaysia, Taman Negara rainforest.
  • Favorite NYC Restaurant: Per Se.

Amazing interview, thank you Alok.   Good luck!

**

Conference Highlights: Our Favorite Investment Ideas

At our Ideas & Networking Conference on October 9, 2025, we were pleased to share several new investment ideas that we wanted to pass along.

To set the stage, we shared a tempered view of the current U.S. equity market.  We highlighted ebullient valuations (Shiller CAPE at 40x+), record index concentration, and high levels equity ownership, with stocks accounting for 45% of household assets and foreigners owning a record 18% of U.S. stocks.  U.S. stocks now account for 65% of global equity indexes – the largest share since the 1950s – and nine of the world’s top ten largest market-cap companies are American.  We also noted the 14% annualized returns of the S&P 500 over the past decade compared to an 8.8% return for foreign equities.

Other American assets beyond equities tell a similar story.  Residential real estate, private equity, precious metals, and crypto all trade near or at historical highs.  As I wrote in late 2024, “virtually every USD asset class is expensive… time to seek non-USD assets.”  Further, amid this asset inflation, the United States is saddled with a deteriorating fiscal outlook and a projected federal debt load that will exceed 120 percent of GDP before the decade ends.

With investors “full up” on U.S. equity exposure, we continue to favor international stocks.   That trend has only become more compelling in 2025, even with most international markets outperforming the U.S. on a local and USD currency basis so far this year.  We think this will be a multi-year trend.

We are also focused, which is not unusual for us, on identifying idiosyncratic and off-the-beaten path investment ideas, both in the U.S. and abroad.  Below is a recap of the ideas we shared, both at our 2024 and 2025 meetings:

September 2024 Ideas

  • Alphawave (AWE.L) was acquired by Qualcomm (QCOM), rising 35% since last year’s conference and gaining 120% off the lows.
  • Everspin (MRAM) gained more than 50% (we have sold)
  • Warner Brothers Discovery (WBD) surged more than 125% (we are currently boxed/hedged on the position)
  • DMC Global (BOOM) was the one disappointment, down roughly 33%; In late November 2025, we bought a few shares and believe it likely resolves positively over time.
  • Five Point Holdings (FPH) and Enerflex (EFXT), two prior positions that we reiterated our strong conviction on — each rose more than 80%.

The 2024 basket was an excellent reminder of why differentiated, non-index exposures and patient capital can still generate outstanding returns even when markets appear expensive.

 

October 2025 Ideas

Sinclair Broadcasting (NASDAQ: SBGI) – Hidden Assets, Cyclical Tailwinds, & Strategic Optionality

At roughly $15.50 per share (it was $13.60 as of October 9th), Sinclair Broadcasting remains one of the most misunderstood companies in the public markets.  The headlines tend to focus on the company’s leverage — especially within the traditional TV broadcasting business — but that narrative ignores the enormous asset value and asymmetric optionality embedded in Sinclair’s “Ventures” portfolio.

Key points:

  • Sinclair Ventures owns unencumbered assets with a combined value that likely meets or exceeds SBGI’s entire ~$1 billion market cap.
  • The TV business is indeed levered at more than 7x, but the regulatory changes underway — especially those related to FTC agency oversight — could significantly change industry economics and drive consolidation.
  • The 2026 election cycle provides a powerful tailwind for political advertising revenue.
  • There is real potential for a break-up, spin-off, or other value-unlocking event.


Fairfax India (OTC: FFXDF) – A Compounding Machine Hiding in Plain Sight

Fairfax India is a closed-end investment vehicle trading around $16 per share, or roughly a -25% discount to its reported December 2024 book value. The firm has compounded that book at approximately 8% annually since inception (based on what’s likely an understated book value), consistently buying back shares and recycling capital into high-quality Indian businesses.

The real gem is its 64% ownership of Bangalore International Airport, India’s third-largest airport and the fastest-growing by passenger volume.  Today, that stake represents roughly half of Fairfax India’s book value — but it is held at only 10 times normalized free cash flow (according to company estimates), a stark discount to global airport comparables.  We believe the value of the Bangalore airport stake will be better recognized over the next few years.  Fairfax India also owns interesting assets across the finance, chemical, manufacturing and logistics sectors.

 

Sprinklr (NYSE: CXM) – A Fallen SaaS Name

Sprinklr has been left for dead by investors following a period of management turmoil and slowing growth.  At around $7.00 per share, it trades at roughly 2x enterprise value to revenue, despite generating real cash flow and maintaining a strong balance sheet.

Importantly, Sprinklr operates in a sector — customer experience management — where scale, data, and integrations matter.  While growth decelerated, the product remains sticky and the customer base loyal.  With a new management team in place for the past year, Sprinklr has two potential outcomes: operational turnaround or a sale to a strategic or private equity buyer.  We think either path would generate significant upside from these depressed levels.

 

Bolloré SE (BOL.PA / BOIVF) – A Deep-Value European Conglomerate With a Fortress Balance Sheet

Bolloré is controlled by French industrialist Vincent Bolloré, one of Europe’s most influential business figures. The complexity of Bolloré’s ownership structure has long deterred investors, but that complexity also creates opportunity.  Based on Muddy Waters’ estimate of roughly 1.2 billion shares outstanding, the stock at $5.50 per share trades far below its intrinsic value.

Asset highlights:

  • Approximately $5 per share in net cash.
  • An 18.5% stake in Universal Music Group worth an estimated $6–7 per share.
  • Additional holdings in Vivendi, Lagardère, Gameloft, Canal+, Hachette, and Havas.
  • Regular share repurchases in recent years.

Our estimated fair value is likely $15+ per share, with the caveat that Bolloré is a family-controlled business where catalysts may be long-dated.  Still, patient investors are rewarded with a deeply discounted, high-quality portfolio of irreplaceable media assets.

Finally, we did pitch a short idea at our conference: Oracle (NASDAQ: ORCL)!!  With the stock around $300 per share, we noted the company’s extraordinary valuation, weak balance sheet, and AI-cloud bookings that appear to be low quality revenues.  The stock has since pulled back to around $200 per share and we would not be initiating new shorts at these levels, as we believe fair value is around $175-200 per share.  We would add to the short at $270 or higher.

Those are some of our ideas for the coming year — knock on wood!

**

Favorite Books & Media

Rebel Oxford Philosopher Declares War on Universities

A thoughtful and provocative interview on education and innovation with Michael Gibson (@William_Blake) of the 1517 Fund, where I’ve been a long-time investor.  Gibson was also the co-founder of the Thiel Fellowship and we interviewed him in July 2022, “Reinventing Early-Stage Venture Investing”: https://ragingcapitalventures.com/an-entrepreneurs-perspective-interview-with-michael-gibson/

Gibson thinks that unsolved problems should be posted at the front door of universities to get people focused on solving problems, and thinks that getting educational credentials takes too long to get.  He thinks there is a “shelf life” to creativity, with many of the great math and physics insights coming to folks in their 20s and early 30s.  The podcast also has an interesting discussion about Plato, and Gibson argues that even Plato believed that you needed more than just rigorous and rational thought to improve and change societies: you also need folks who are quirky and different.

https://www.youtube.com/watch?v=YfLj1pHGfT4

 

Barry Diller – Building an Entertainment Empire

A media and internet lion, Barry Diller credits his career to curiosity, luck and a willingness to start from the bottom — beginning in the William Morris mailroom.  He has always chased good ideas by first hearing plenty of bad ones and likes to put smart, creative people in a room and let them argue until something genuine emerges. In this podcast, he shares stories from Paramount, Fox and the early internet era.  Though Sam Altman once showed him an early ChatGPT demo, Diller says he doesn’t have the energy or curiosity to fully engage with the current AI cycle.

Diller said he admires John Malone’s strategic clarity, remains bullish on MGM’s entertainment and hospitality model, and is skeptical of today’s GenAI spending until real revenue appears. His cautionary note: don’t let sophistication turn into cynicism — keeping a bit of naivety is essential for good decision-making.

https://podcasts.apple.com/us/podcast/barry-diller-building-an-entertainment-empire/id1154105909?i=1000728015329

 

Why Reasoning Agents Change Everything in Defense

This podcast with the CEO of Vannevar Labs, Brett Granberg, focuses on “reasoning agents” and is worth a listen for those looking to get in the weeds on potential Agentic applications for LLMs. It is super interesting and in-depth.

Vannevar thinks there is an imminent opportunity to use Agents to massively enhance the productivity of workers at the Department of Defense, thus driving significant savings by reducing the usage of outsourced consultants from traditional government-focused consulting companies.  Finally, it is worth keeping an eye on Vannevar, which is an emerging defense tech unicorn with $80mm in ARR.  We were seed investors in the company in 2019.

https://podcasts.apple.com/us/podcast/why-reasoning-agents-change-everything-in-defense-in/id1831808567?i=1000729403318

**

A Selection of Recent Tweets from @RagingVentures:

 

 

**

“Humphrey’s Executor… is … six quick pages devoid of textual or historical precedent.” – Justice Antonin Scalia, in his lone dissenting opinion in the 1988 Morrison v. Olson case

Published By