The Madman Who Speaks the Truth
Software Is Eating the World
The dominant technology investment thesis of the past fifteen years has been that software companies would disrupt and ultimately dominate nearly every major industry by leveraging structural advantages over incumbents.
Software scales at near-zero marginal cost. It iterates quickly. It benefits from data network effects. And it operates through asset-light models that require little physical infrastructure. These characteristics allow successful platforms to grow rapidly while expanding margins over time.
Over the past decade, investors increasingly viewed enterprise software not simply as a high-growth segment, but as a new form of digital utility. Subscription-based SaaS models transformed lumpy, license-driven revenue into recurring, highly predictable cash flows with strong visibility. Mission-critical systems, such as CRM, payroll, cloud infrastructure, and cybersecurity became embedded in daily operations, creating high switching costs and low churn. With gross margins often in the 70–90% range and net revenue retention exceeding 110% in the strongest businesses, software companies exhibited the stability of a utility but with far greater scalability.
This combination of durability and growth helps explain the elevated price-to-sales multiples that characterized the 2015–2021 cycle. Traditional utilities offer stable cash flows but limited growth; software appeared to offer both recurring revenue and 30–50% annual expansion, with significant operating leverage as companies matured. Investors were willing to underwrite long-duration cash flows and assume that today’s revenue base would translate into dominant future free cash flow streams.
The durability of the software model attracted enormous flows from private equity and venture capital. In peak years, software represented roughly one quarter of global private equity deal volume and an even larger share of total buyout value in certain vintages. Recurring revenue supported leverage, high margins reduced cyclicality risk, and secular growth justified premium purchase multiples. Venture capital similarly concentrated capital in enterprise and vertical SaaS, underwriting large addressable markets under the “software as utility” framework.
The financing backdrop reinforced the boom. In the aftermath of the Global Financial Crisis, bank retrenchment and persistently low interest rates fueled the rapid expansion of the private credit industry, which increasingly became the primary lender to sponsor-backed transactions. Direct lenders were comfortable underwriting software businesses given the visibility of subscription cash flows. Abundant private credit capital lowered financing friction, supported higher purchase multiples, and became a structural enabler of the software buyout cycle.
In August 2011, venture capitalist Marc Andreessen published an essay titled “Software Is Eating the World.” It became the intellectual anchor of Silicon Valley’s bull run. The thesis proved remarkably prescient, more so than even its strongest advocates anticipated. Yet 15 years later, we are witnessing a significant twist to the narrative: artificial intelligence, itself the most powerful evolution of software to date, may now begin to eat software itself.
Short Horses
“If you had told me in 1900 that the automobile was coming”, Warren Buffett once remarked, “I would have said, ‘Short horses.’” When transformative technology emerges, the better trade is often betting against the incumbents rather than trying to pick the ultimate winner.
In our July 2025 letter, we discussed the transformational implications of agentic artificial intelligence and the risk it poses to traditional software models:
The ability to perform high-quality work at minimal cost lowers the barrier to entry for new competitors and enables incumbents to undercut each other on pricing. The startup cost for AI-native businesses is low, often relying on off-the-shelf models or repackaged capabilities built atop existing large language models. In some cases, there is outright replacement risk. We are particularly focused on software and software-adjacent services—sectors that form a core component of the $5 trillion global IT spending base. AI agents can be understood as a fundamentally new kind of software: more dynamic, goal-oriented, and self-adjusting than traditional programs. While traditional software follows static instructions, AI agents represent a shift toward software that adapts, learns, and acts autonomously in changing environments.
It proved to be a timely warning. As the year progressed, the market shifted from trying to identify the primary beneficiaries of AI to becoming intensely focused on AI-driven replacement risk. While the private valuations of AI leaders such as Anthropic, OpenAI, and xAI collectively increased by nearly $1 trillion through successive funding rounds, we began to see the counterweight emerge: the equity value of many listed software companies contracted sharply. The S&P Software & Services index declined more than 20% in just two months, with several leading software and IT services firms, including Salesforce, Intuit, Adobe, and Accenture, down 30–50% (1). These moves do not reflect deterioration in current revenue trends but a significant compression in forward valuation multiples i.e., the market is signaling increased uncertainty about the durability of long-term cash flows.
The ripple effects extend beyond public equities though. If public software multiples reset, privately held software companies owned by private equity funds must eventually be marked lower as well. That has implications for the debt that financed those acquisitions, as the equity cushion beneath the loans compresses. The second domino therefore includes asset managers with significant exposure to private credit, including firms such as Blue Owl, Ares, and Blackstone, along with related BDCs and CLO equity vehicles.
As we have noted in prior letters, we have been cautious on private lending for some time. The post-GFC shift of credit risk from deposit-funded banks to asset managers was structurally sensible: losses are borne by investors who intentionally allocate risk capital. However, after an exceptionally long credit cycle and a surge of capital into what became a highly lucrative asset class, it was highly probable that certain participants were mispricing risk, or, as Jamie Dimon recently put it, “doing some dumb things”(2).
First, leverage in the system is often significantly higher than it appears. Many private lenders hold portfolios of loans to highly leveraged, PE-backed companies acquired at elevated EBITDA multiples. On top of that, lenders frequently add fund leverage, borrowing against their loan portfolios to originate additional loans. In stable conditions, returns look attractive—high-single-digit yields with low volatility—but in a default cycle, losses can escalate quickly.
Second, a growing share of private credit exposure has migrated into retail channels through RIAs, BDCs, interval funds, and other semi-liquid vehicles. Investors are drawn to recurring distributions and perceived stability, often without fully appreciating the underlying illiquidity and structural leverage, or the fact that, in credit, stability tends to persist until it doesn’t. Years of steady profits can be wiped out quickly in a major default cycle. If defaults were to rise, liquidity mismatches could amplify the stress, potentially forcing certain vehicles to gate or suspend redemptions.
Equity corrections, particularly those concentrated in specific sectors or themes, are relatively common. While often painful for investors, their impact on the broader real economy is typically contained. By contrast, corrections that involve severe credit losses strike at the plumbing of the financial system itself, impairing balance sheets and restricting liquidity. Historically, episodes of credit stress are far more likely to spill over into the real economy. This is the third domino and from a risk perspective, the one that warrants the closest attention.
The Bulletproofs
In prior letters, we discussed the early beneficiaries of AI, namely companies benefiting from the surge in AI-related capital expenditure and participants in the data center ecosystem: hardware manufacturers, semiconductor companies, engineering firms, power distributors, and electrical equipment providers. These sectors experienced a broad-based rally in 2025, but performance has recently become more idiosyncratic. Investors are beginning to differentiate more carefully, concerned that: (a) the bulk of infrastructure spending is concentrated among a small number of hyperscale players, and any pause or strategic shift by them could have outsized consequences for suppliers; (b) leverage and operational interconnectedness across the ecosystem may be higher than appreciated; and (c) even in a successful buildout scenario, infrastructure demand could ultimately prove cyclical and project-based rather than the recurring, sticky revenue profile that commands premium valuations.
A newer winning category this year is what I would call the “Bulletproofs.” This includes more traditional and often less glamorous sectors such as trucking, mining, and heavy machinery, which have come back into favor as capital rotates out of certain technology and services names. Both the Dow Jones Transportation Index and the U.S. Industrial Machinery Index have outperformed the S&P 500 by nearly 20% over the past three months (3). As the narrative has shifted from identifying AI beneficiaries to assessing AI vulnerability, investors are re-evaluating the durability of competitive advantages and assigning a premium to assets that are difficult to replicate and less exposed to rapid technological obsolescence.
It is always prudent not to over-interpret moves driven by flows rather than fundamentals. These sectors remain cyclical, and ultimately economic direction will have a significant influence on valuation. At the same time, the uncertainty surrounding AI’s long-term impact on businesses is likely to persist for years. As Derek Thompson recently wrote, “nobody knows anything” (4). This structural uncertainty is likely to compress valuation multiples for industries directly exposed to technological disruption, while providing support for companies perceived as more insulated.
The increasing dominance of passive investing further amplifies this dynamic. By some estimates, nearly two-thirds of U.S. domestic equity fund assets are now indexed. Meanwhile, sector weights within major indices have shifted meaningfully over the past two decades, with Industrials, Energy, and Materials seeing their combined weight in the S&P 500 decline by more than ten percentage points. Reintroducing industrial exposure to create greater portfolio balance makes strategic sense, particularly against the backdrop of U.S. policy efforts aimed at fostering an industrial renaissance.
A few names illustrate our increased focus in the industrial sector. Century Aluminum, the largest domestic primary aluminum producer, is a direct beneficiary of the administration’s 50% Section 232 tariffs on imported aluminum — the company is restarting idled capacity at its South Carolina smelter and recently announced plans to build the first new U.S. smelter in nearly fifty years. Boeing, after years of well-documented manufacturing failures and a prolonged FAA-imposed production cap, is finally ramping its 737 MAX line from 38 toward 47 aircraft per month, converting a multi-thousand plane order book into accelerating cash flows. Kirby Corporation operates the largest inland tank barge fleet in the country, a structurally scarce asset with utilization already in the low 90% range and earnings at record levels heading into 2026. Valero Energy, the world’s largest independent refiner, is perhaps the most direct beneficiary of the reopening of Venezuelan oil flows: its Gulf Coast refineries were engineered to process exactly the heavy, sour crude Venezuela produces, and the company is now securing those barrels at discounts, driving a material improvement in refining margins. Finally, MSC Industrial Supply is a leading distributor of metalworking products whose revenue tracks the U.S. Industrial Production Index closely, a relatively pure play on a domestic manufacturing recovery (5).
The Madman Who Speaks the Truth
As I was finishing this letter, Block Inc., a $40 billion digital payments company, released its Q4 earnings. Included was a letter from its founder, Jack Dorsey, explaining a decision to reduce headcount by nearly half. He wrote:
Today we shared a difficult decision with our team. We’re reducing Block by nearly half, from over 10,000 people to just under 6,000, which means that over 4,000 people are being asked to leave or entering into consultation. I want to use this letter to explain why I believe this is the right path for our company, and what Block looks like going forward.
2025 was a strong year for us. Gross profit growth more than doubled from the first quarter to the fourth quarter….So why are we changing how we operate going forward? The core thesis is simple. Intelligence tools have changed what it means to build and run a company. We’re already seeing it internally. A significantly smaller team, using the tools we’re building, can do more and do it better. And intelligence tool capabilities are compounding faster every week. I don’t think we’re early to this realization. I think most companies are late. Within the next year, I believe the majority of companies will reach the same conclusion and make similar structural changes. I’d rather get there honestly and on our own terms than be forced into it reactively (6).
Many rushed to dismiss the potentially seismic implications of these comments, arguing that AI is merely a convenient justification for cost-cutting, and that the company was simply correcting post-pandemic over-hiring. Others pointed to the founder’s eccentric reputation to question the broader relevance of his remarks. I would be less dismissive. In Greece, we have a saying: “From children and madmen, you hear the truth.”
My view is that there is a genuine existential risk for companies whose revenue is primarily derived from monetizing structured human cognition rather than physical assets, proprietary platforms, or scarce infrastructure. The threat can emerge in two ways: customers internalize those tasks using readily available AI tools, or AI-native competitors, operating with structurally lower cost bases, undercut pricing or deliver superior output and take share. The progression can be swift. It often begins with slowing organic growth as new client additions decelerate, followed by pricing pressure and rising churn as switching costs decline and elasticity increases.
Some firms will adapt, and a few may ultimately strengthen their positions. But the window for reinvention is narrow. Survival requires repositioning not as incumbents defending legacy models, but as trusted partners deploying and managing AI tools to improve client outcomes. At the same time, companies must embed AI internally to raise productivity and materially reduce costs, passing some of the efficiency gains to clients. This transition will almost certainly involve material workforce reductions.
Which leads to an emerging Prisoner’s Dilemma. In game theory, the Prisoner’s Dilemma describes a situation in which individually rational decisions produce collectively suboptimal outcomes. It may be rational for one firm to aggressively cut headcount and automate ahead of competitors. But if every firm does the same simultaneously, the cumulative impact on employment and aggregate demand could ultimately weaken the very markets those companies depend on. What is individually prudent may, at scale, become economically destabilizing.
It’s a difficult puzzle to solve and the societal implications are profound. What happens to AI-displaced employees whose six-figure incomes disappear overnight, forcing them into lower-skilled roles at a fraction of prior pay? Not all companies face the same constraints. A large institution such as JPMorgan must move incrementally; the political and reputational consequences of an aggressive AI-driven workforce reduction would be significant. Smaller software and services firms, however, particularly those reliant on stock-based compensation and facing shrinking addressable markets, may find themselves in a “who blinks first” dynamic. The temptation to execute a Dorsey-style reset will be high.
Translating these conclusions into practical investment implications, my sense is that the key metric across many sectors will shift from revenue growth to earnings growth. AI may compress markets in some cases while simultaneously enabling companies to operate at a fraction of their previous cost base. The winners will not necessarily be those who grow fastest but those who adapt fastest, and the place that adaptation will show up first is in the progression of margins and earnings, not the top line.
This material is distributed for informational purposes only. Investment Advisory services offered through Chesapeake Asset Management, an investment adviser registered with the U.S. Securities and Exchange Commission (“SEC”). The views expressed are for informational purposes only and do not take into account any individual’s personal, financial, or tax considerations. Opinions expressed are subject to change without notice and are not intended as investment advice. The information contained herein is not intended to be personal legal, investment or tax advice, or a solicitation to buy or sell any security or engage in a particular investment strategy. Nothing herein should be relied upon as such, and there is no guarantee that any claims made will come to pass. In addition, certain information contained herein or utilized to draw the conclusions contained herein has been provided by, or obtained from, third party sources. All material presented is compiled from sources believed reliable, but the accuracy and completeness, and the opinions based thereon, are not guaranteed. While Chesapeake Asset Management believes that such sources are reliable, it cannot guarantee the accuracy of any such information and does not represent that such information is accurate or complete.
All materials and information are provided “as is” without any express or implied warranties. Opinions expressed are subject to change without notice and are not intended as investment advice or to predict future performance. Past performance is no guarantee of future results. All investments involve risk including the loss of principal. Any security can experience a loss. Please see Chesapeake Asset Management’s Form ADV Part 2A and Form CRS for additional information.
Registration with the SEC does not imply a certain level of skill or training. Neither the SEC nor any other regulatory authority has approved, passed on or endorsed the merits of any investment opportunity that may be described herein. No regulatory authority or other third party has confirmed the accuracy or determined the adequacy of this presentation. Any representation to the contrary is a criminal offense.
The letter and its contents are proprietary information of Chesapeake Asset Management, LLC, (“CAM”) and any reproduction of this information, in whole or in part, without prior written consent of CAM is prohibited. Additional information is available from CAM upon request. Neither CAM nor its affiliates is acting as your advisor or agent.
Sources:
- Based on Bloomberg data—refers to period from Nov. 2025 to Feb. 2026.
- Remarks made at JP Morgan Chase’s Annual Investor Day.
- Based on Bloomberg Data.
- https://www.derekthompson.org/p/nobody-knows-anything.
- Names are for illustrative purposes and may not appear in all client portfolios, given the customized nature of our discretionary investment offering.
- Full letter can be found on: https://investors.block.xyz/overview/default.aspx.