(#181) Berkshire Hathaway vs. SoftBank Group
Two competing visions of capitalism & investor implications
Dear onStrategy reader,
This is the last edition of this newsletter.
In more than 180 issues and more than three years, On Strategy became my go-to place to share ideas, comment on news, and think through implications. Over time, though, it evolved into something more. What started as commentary gradually turned into a much deeper exercise: building a real portfolio, documenting decisions, and seeing those ideas perform better than I expected.
As many of you already know, I also became a certified financial consultant in the meantime. And that is why my full effort now goes into Where Is My Moat?, where I publish 4-5 analyses per week, share my portfolio, track performance, and document the full investment process in real time.
So if you liked onStrategy, I think there is a very good chance you will like Where Is My Moat? as well. Take a look and consider subscribing. You have a free month to decide whether it is for you or not. That means around 20 analyses, plus access to the full archive, except for the portfolio disclosures.
Thank you for reading On Strategy. It meant a lot to me.
We will stay in touch.
Best,
Sorin
P.S. Below you can see the analysis that the Premium subscribers from Where Is My Moat? received:
From The Economist article entitled “Only one of Berkshire Hathaway and
SoftBank can survive”:
For two firms that embody opposite visions of capitalism, Berkshire Hathawayand SoftBank Group have a great deal in common. Both are an uncategorisable mixof operating conglomerate and investment fund. Both were built by singular menrecognisable by their first names. And both receive little scrutiny from the bankanalysts who write about them and the adoring investors who own their shares.Each would also regard the other’s balance-sheet as an aberration.
Berkshire underWarren Buffett, who recently retired as chief executive (though not as chairman),amassed nearly $400bn of cash. It has little idea what to spend it on. Only a crashcomparable to that of 1929, 1987 or 2008 would vindicate this sober inaction. SoftBank,by contrast, has an excess of ideas courtesy of Masayoshi Son, its futuristic founder,but little cash. No sooner does the firm borrow money than it flies out of the door tofinance the artificial-intelligence gold rush under way in Silicon Valley. For Mr Son’sheavily indebted firm, a crash could be terminal.
Berkshire Hathaway and SoftBank Group represent the two most ideologically opposed capital-allocation philosophies in global finance. One hoards cash and the other spends it before it arrives. Here is a snapshot of each of them:
As you can see, Berkshire faces a slow erosion of relevance in a bull market, while SoftBank faces a hard solvency cliff if its AI bets misfire. The risk profiles are asymmetric, ie. different in timing, magnitude, and reversibility.
Company Philosophy & DNA
a) Berkshire Hathaway
Warren Buffett built Berkshire on a handful of core convictions that have remained virtually unchanged for 6 decades. Capital should only be deployed when price is demonstrably below intrinsic value. Businesses worth owning have defensible competitive ‘moats’ that protect returns through economic cycles. Cash is not dead weight because it is optionality, a loaded weapon awaiting the right crisis. This framework led Buffett to stay out of the dot-com boom, the early internet era, and now the AI build-out, while patiently accumulating what is now nearly $400 billion in liquidity.
The philosophy has an internal logic. If markets are frequently overvalued, then the highest-return action is often inaction. Berkshire’s cash pile, now equal to 40% of its market capitalisation, double its two-decade average, is either profound discipline or profound stagnation, depending on whether a crash materialises. The firm has not paid a dividend since 1967. It has repurchased almost no shares recently, signalling that even its own stock is considered overpriced.
The irony is that Berkshire’s largest holding, Apple, is itself one of the few technology giants not racing to spend its entire cash flow on AI data centres. Berkshire owns a slice of restraint inside the most restrained sector of the S&P 500.
b) SoftBank Group
Masayoshi Son’s operating principle is the inverse. Capital that sits still is capital wasted. Since his early bets on Yahoo and Alibaba, Son has believed that identifying transformative technology early (and owning as much of it as possible, regardless of price) produces returns that dwarf any value-investing framework. The Vision Fund, the largest technology investment vehicle ever assembled, was the fullest expression of this thesis. Its collapse and partial recovery (WeWork was a catastrophic failure; Arm was a generational win) has not changed Son’s conviction.
Today, Son has made OpenAI his defining wager. By October 2026, SoftBank will have committed $65 billion to the maker of ChatGPT, making it OpenAI’s second-largest external shareholder after Microsoft. SoftBank also contributes roughly 10% of OpenAI’s revenue through a $3 billion annual product commitment. Additionally, the firm is acquiring the robotics arm of ABB ($5bn) and data-centre operator DigitalBridge ($4bn), doubling down on physical AI infrastructure.
The structural problem is that all of this is financed by debt. SoftBank’s cash flows from operating businesses are insufficient to fund its ambitions. The firm faces a $40 billion bridge loan maturity in March 2027. Selling assets is now constrained; the family silver (Nvidia stock) is already gone, and remaining holdings in T-Mobile, Grab, and DiDi have depreciated. Its reported loan-to-value metrics are contested, omitting $28 billion borrowed against Arm shares and its Japanese telecoms operation.
Financial Profile & Capital Structure
The financial contrast between the two firms is stark and directly determines their range of survivable outcomes.
Berkshire’s $400 billion cash pile is simultaneously its greatest strength and its most visible liability. In absolute terms, it is extraordinary capital firepower. In relative terms, 40% of market cap, double the 20-year average, reflects an inability to find value at current market prices. The implicit message to shareholders is “everything is too expensive”. That judgment may be vindicated only by a crash of 1929 or 2008 proportions.
SoftBank’s situation is structurally inverted. The firm generates insufficient operating cash flow from its telecoms and platform businesses to service its investment ambitions. It funds the gap with leverage, asset sales, and periodic capital markets transactions. Its loan-to-value ratio, the metric it uses to reassure markets, excludes material debt obligations, making the true leverage profile significantly worse than disclosed.
Key Holdings & Strategic Bets
a) Berkshire Hathaway — The Portfolio
Berkshire’s investment portfolio is a roster of durable, cash-generative franchises comprising companies in insurance, energy, railroads, consumer staples, and financial services. The equity portfolio is anchored by:
Apple (largest single equity holding): One of the few mega-cap tech companies not all-in on AI capex, making it an unusual complement to Berkshire’s own caution.
Japanese trading houses ($45bn portfolio): Berkshire has been most active in Japan, accumulating stakes in Mitsubishi, Mitsui, Itochu, Sumitomo, and Marubeni, companies with deep industrial moats and attractive valuations relative to US peers.
BNSF Railway, Berkshire Hathaway Energy, GEICO insurance: Core operating subsidiaries generating the cash flows that sustain the holding company.
The portfolio is designed for longevity, not velocity. The Japan pivot is notable because it signals that Buffett (and now Abel) view US equities as expensive relative to global alternatives and are willing to look offshore for value.
b) SoftBank Group — The Bets
SoftBank’s portfolio is concentrated and momentum-driven. Three positions dominate the investment thesis:
Arm Holdings: Purchased for $31bn in 2016, now valued at ~$250bn (SoftBank retains 87%). This is SoftBank’s anchor asset and the collateral underpinning much of its debt. Arm’s centrality to AI chip design makes it a structural beneficiary of the current buildout, but its $250bn valuation implies a high multiple that could correct sharply.
OpenAI ($65bn total commitment by October 2026): SoftBank is OpenAI’s second-largest external shareholder after Microsoft. It contributes ~10% of OpenAI’s revenue. This is Son’s defining wager on AI as a civilisational shift, but OpenAI’s IPO, once anticipated, now appears increasingly uncertain, creating a liquidity trap.
ABB Robotics ($5bn) + DigitalBridge ($4bn): Physical AI infrastructure bets that extend the AI thesis into hardware and data centres, but add to an already strained balance sheet.
Notably absent or diminished: Nvidia (fully sold in October 2025), T-Mobile, Grab, and DiDi, which have all depreciated and represent declining liquidity options.
Scenario analysis
The fate of each firm is path-dependent. Below are the three most plausible macro scenarios and their implications.
Scenario A: Prolonged Bull Market (AI delivers, no crash)
This is the scenario that most punishes Berkshire and rewards SoftBank in the near term.
SoftBank: Arm holds its $250bn+ valuation. OpenAI either IPOs or achieves a valuation that justifies the $65bn stake. Debt is refinanced or retired. SoftBank emerges as the dominant AI investment vehicle globally.
Berkshire: Greg Abel (CEO) faces a shareholder revolt or is forced to return capital via dividends. The $400bn cash pile earns Treasury yields while the S&P continues to outperform. The market increasingly prices Berkshire as a cash fund with legacy businesses attached. Strategic dissolution becomes a realistic debate.
Investor implication: In this scenario, SoftBank is the high-convexity bet; Berkshire is a drag on risk-adjusted returns. Underweight Berkshire relative to the broad market.
Scenario B: Market Correction (moderate - 20–35% drawdown)
This is the scenario that vindicates Buffett’s model and stress-tests SoftBank’s balance sheet.
Berkshire: Deploys cash selectively into distressed assets. Arm’s valuation falls, but Berkshire is largely insulated. Abel has his validation moment. Buffett’s philosophy is vindicated publicly. Berkshire significantly outperforms.
SoftBank: Arm’s valuation falls 30-50%. OpenAI’s IPO is deferred further. The March 2027 bridge loan maturity becomes a genuine liquidity event. Asset sales are forced at depressed prices. Credit markets tighten. The debt spiral risk becomes acute.
Investor implication: Berkshire is a natural hedge. SoftBank faces existential balance sheet risk. Reduce SoftBank exposure ahead of rate-sensitive sell-offs; accumulate Berkshire as a recession hedge.
Scenario C: Hard Landing / Systemic Shock (>35% drawdown)
The scenario for which either firm’s extreme positioning was built.
Berkshire: The scenario Buffett was always waiting for. $400bn buys distressed assets at historic discounts. GEICO and BNSF continue generating cash through the cycle. Berkshire likely acquires entire businesses at once-in-a-decade valuations. Generational outperformance follows.
SoftBank: Debt becomes unserviceable without a functioning asset sale market. Arm’s value, while real, cannot be monetised quickly enough to meet maturities. Potential restructuring, forced Arm stake dilut’s stake, or, in extremis, insolvency of the holding company structure.
Investor implication: Berkshire is the asymmetric winner in a hard landing. SoftBank is among the most exposed large-cap holding companies globally.
Key risks
For Berkshire Hathaway:
Succession execution risk: Greg Abel has no track record managing a $900bn conglomerate through a downturn. The Buffett credibility buffer does not transfer.
Prolonged opportunity cost: Every year without a crash is a year the cash pile earns 5% while the market returns 15–20%. Compounded over five years, the cost is severe.
Forced capital return: Activist or shareholder pressure may ultimately force dividends or buybacks at valuations Buffett considered too high, hence destroying the optionality thesis.
Apple concentration: Berkshire’s largest equity holding is in a company facing tariff exposure, China revenue risk, and slowing iPhone growth. A material Apple correction would hit NAV.
Irrelevance risk: If AI continues to compound, Berkshire’s refusal to participate becomes a structural argument against the business model
For SoftBank Group:
March 2027 bridge loan maturity ($40bn): The single most important date on SoftBank’s calendar. If capital markets remain open and Arm holds value, refinancing is possible. If not, the consequences cascade.
OpenAI IPO delay: SoftBank’s investment in OpenAI is illiquid. If OpenAI does not list — and signals currently point away from an imminent IPO (ie. the $65bn commitment remains frozen capital while debt accrues)
Arm valuation compression: At $250bn, Arm trades at a demanding multiple. If AI capex slows, chip design demand plateaus, or Arm faces competitive disruption (RISC-V open-source alternatives), SoftBank’s primary asset, and its collateral, deflates.
Hidden leverage exposure: The $28bn borrowed against Arm and telecoms assets (excluded from official LTV calculations) represents a material misrepresentation of the firm’s risk profile.
Founder concentration: Son’s departure, incapacitation, or reputational damage would trigger immediate credit and equity market reactions.
Japanese retail bond investor fatigue: SoftBank has historically relied on retail investors for bond issuance. As yields rise and risk perception increases, this funding channel is becoming more expensive and less reliable.
Investor Implications
The choice between Berkshire and SoftBank is a choice between two views of economic history. Investors must answer one question before positioning: do you believe a significant market correction is coming?
If you believe the correction risk is elevated:
Overweight Berkshire Hathaway as a defensive hold with asymmetric upside in a drawdown scenario.
Reduce or exit SoftBank - the March 2027 debt wall and Arm dependency create a binary outcome that is unusually sharp.
Monitor: When does Abel use the cash? The first major acquisition under Abel’s tenure will be a defining signal about whether the Buffett model survives its creator.
If you believe AI delivers without near-term correction
SoftBank offers concentrated exposure to the AI infrastructure thesis at a premium risk level. The upside is real if Arm sustains valuation and OpenAI monetises at scale.
Berkshire is likely to underperform the market by 20–40 percentage points in a continued bull market, as has occurred over the past year.
Caveat: the return profile for SoftBank in this scenario is highly non-linear — the debt structure means gains accrue to equity holders only after creditors are satisfied, which amplifies both upside and the probability of total loss.
Portfolio construction considerations
Berkshire and SoftBank are natural hedges against each other. A position in both is effectively a bet on macro volatility rather than direction.
Neither stock is a proxy for the AI theme or for defensive value on a standalone basis: SoftBank is a leveraged AI bet, not a pure AI play; Berkshire is a cash option, not a value portfolio in the traditional sense.
Governance-focused investors should weigh SoftBank’s opacity around debt reporting, the gap between disclosed and actual leverage is a material due diligence concern.
Concluding
The Economist frames this as a binary: only one can survive. The more precise formulation for investors is that their failure modes are fundamentally different in character. Berkshire’s potential failure is slow (e.g., a decades-long drift into irrelevance, eventually resolved by breakup or forced capital return). SoftBank’s potential failure is fast (e.g., a liquidity event triggered by a debt maturity that arrives before assets can be liquidated at acceptable prices)
Slow failures create options. Fast failures do not.
Excessive temperance and excessive risk-taking are permanent features of capital markets. The question for 2026 and 2027 is which posture the macro environment rewards. That answer depends almost entirely on whether the AI buildout produces returns sufficient to service the debt load that financed it.
WIMM for companies: If you would like to offer the newsletter as a benefit to other employees in your company, please contact me directly. (sorin@whereismymoat.com)
Work with me: If you are interested in working with me on your personal portfolio, again, contact me. I’m a certified financial consultant.
Finally, “Where is my moat?” is designed for individual readers, though the occasional forward is absolutely fine.
Thanks for your support & have a wonderful day!
Felicitari ptr certificare si mult succes Where is my MOAT