The Capital Cycle Perspective on SaaS and AI 

April 2025 

Market Mayhem

Markets have been on a wild ride lately, with volatility levels that feel more like echoes of past crises than normal trading. At the center of the storm is a wave of uncertainty driven by shifting global trade policies—particularly President Trump’s escalating tariffs, which have rattled investor confidence.

This past week alone saw some of the most dramatic market swings in decades. On Monday, the S&P 500 reversed a steep 4.7% drop to close up 3.4%. Tuesday saw a full reversal again, from a strong open to a sharp decline. Then came Wednesday, when the index surged 10% in a single day—its biggest gain since the 2008 financial crisis—after news that some tariffs would be delayed.

All three sessions rank among the most volatile days since 1982, with Wednesday’s rally in the top five. For context, the S&P typically moves about 1.2% a day—making this level of turbulence striking. Underneath the headline moves, the pain is real: U.S. equities have entered correction territory, and small caps have plunged into a bear market. It’s the worst downturn since the COVID shock, and investors feel the pressure.

What’s driving the selloff is a potent mix: harsher-than-expected tariffs raising business costs, and a broader policy environment that feels increasingly unpredictable. The market's fear gauge—the VIX—has surged past 40, a level that signals deep anxiety.

Yet amid the chaos, one corner of the market is starting to look interesting again: SaaS. After years of elevated valuations, the sector has seen a steep reset. According to the BVP Nasdaq Emerging Cloud Index, SaaS companies are trading around 5x run-rate revenues—down from long-term averages near 8x and well below the 10–12x peaks of the 2020–2021 boom. This raises an important question: is the market overcorrecting, or is this a sign of a deeper reset?

The numbers are telling. The median EV/Revenue multiple for public SaaS companies has fallen from 15.3x in early 2021 to just over 5x today. The BVP Cloud Index itself is down roughly 70% from its peak. Even top-tier companies like Datadog, Snowflake, and MongoDB—still growing 30%+ year over year—have seen their valuation multiples slashed by 60–70%, despite hitting revenue targets. This suggests we’re seeing a fundamental re-rating of the sector, not just a reaction to individual company performance.

For long-term investors, that context matters. Most sophisticated market participants know that short-term earnings noise isn’t what drives value—long-term cash flow generation is. Equity represents ownership in a company’s future, not just its next quarter. Valuation models like NPV (Net Present Value) highlight that even sharp, temporary downturns have minimal impact on a company’s intrinsic value—unless earnings are permanently impaired.

What often creates opportunity in moments like these are market overreactions—amplified by common behavioral missteps. Investors tend to extrapolate current trends too far into the future, become overconfident in short-term predictions, or overreact to headline noise and earnings surprises. These biases lead to dislocations—exactly the kind of environment where thoughtful, disciplined investors can find value.

Capital Cycles

In technology investing—where breakthroughs often blur with bubbles—success requires more than a sharp eye for trends; it calls for a clear framework to navigate the noise. One of the most powerful tools in an investor's arsenal is the capital cycle: a model that explains how the flow of capital not only shapes industries and markets, but also reflects deeper forces of human behavior.

The capital cycle is a powerful framework for understanding how markets allocate resources over time—and how those allocations inevitably sow the seeds of their own reversal. While the concept encompasses multiple factors including regulatory environments, technological constraints, and geopolitical considerations, at its heart lies a fundamental insight: capital tends to flow toward areas generating high returns.

When an industry begins generating high returns on investment, capital pours in. Investors, lured by strong performance and profitability, direct new funding into expansion projects, acquisitions, and innovation. This flood of capital results in increased capacity and production. However, due to lags in supply response and overly optimistic assumptions about future demand, this investment often leads to overcapacity.

As supply outpaces demand, competition intensifies. Prices fall, margins shrink, and returns on capital begin to decline—frequently dropping below the cost of capital. At this point, capital begins to exit. Companies scale back investment, write down assets, consolidate, or, in some cases, declare bankruptcy. This capital withdrawal phase resets the cycle, paving the way for future profitability once supply and demand re-align.

Investors often view this cycle through two lenses. The first, shaped by Bridgewater Associates, focuses on macro forces—debt, interest rates, and demand—arguing that we’re now entering a deleveraging phase reminiscent of the 1930s. The second, from Marathon Asset Management, zooms in on industry-level supply and competition, showing how human tendencies—herd behavior, short-termism—drive overcapacity and mispricing. Both views matter, especially now. Ignoring the capital cycle can lead to costly mistakes: mistaking hype for sustainable growth, or fleeing just as recovery begins. But those who understand its phases can spot turning points early—when industries are unloved, capital-starved, and ripe for reinvention.

This entire cycle closely mirrors Joseph Schumpeter’s idea of creative destruction—the notion that periods of prosperity often lead to misallocated capital, while downturns clear the path for more efficient economic activity. Not every capital cycle, however, turns into a bubble. Bubbles emerge when short-term feedback loops form, reinforcing investment simply because asset prices are rising. That kind of reflexivity—where enthusiasm feeds on itself—is what separates a speculative mania from a typical boom. The absence of such a public equity feedback loop is why the current AI surge, despite its intensity, isn’t being called a bubble—at least not yet.

Looking back, the capital cycle has played out in strikingly similar ways across history. In the mid-19th century, the railroad boom was driven by a belief in territorial expansion. Railways unlocked the value of land, and land speculation often preceded the actual laying of tracks—feeding a cycle of investment based on anticipated returns.

Similarly, in the 1990s, the telecom and internet boom was fueled by the meme that "internet traffic doubles every 100 days." This narrative justified an enormous infrastructure buildout. Network expansion announcements pumped up stock prices, which in turn allowed companies to raise more equity and debt, accelerating the buildout even further. Shadow leverage—through vendor financing and favorable banking terms—added fuel to the fire.

Today’s AI buildout has a similar feel to those historical cycles. The dominant belief—“more compute plus more data equals better models”—is fueling a massive push into AI infrastructure. OpenAI's own technical leaders have confirmed that model quality continues to improve with increased capital input. From that standpoint, AI looks a lot like the early stages of the railroad or telecom booms. But what sets it apart is the absence, so far, of the classic bubble dynamic where soaring stock prices drive a frenzy of equity issuance and risk-taking. In that sense, the cycle may be maturing without the speculative excess—at least not yet.

One of the most valuable insights from the capital cycle framework is its predictive power. When a sector is underinvested—starved of capital for years—it often sets the stage for strong future returns. Conversely, when too much capital floods into an industry, future returns tend to fall. This understanding allows investors not just to explain past booms and busts, but to anticipate future turning points—looking ahead rather than reacting after the fact.

Bridgewater’s Understanding Debt and Capital Cycles paper underscores how the capital cycle on the supply side complements the debt cycle on the demand side. When central banks lower interest rates, borrowing becomes cheap, triggering a surge in credit expansion and investment. This influx of capital is then deployed—sometimes wisely, often recklessly—into assets and projects.

Market tops typically form when a few conditions converge: rising interest rates increase the cost of capital, overcapacity begins to erode returns, and euphoric sentiment inflates expectations, encouraging mispricing and risky behavior. The collision of rising costs and falling returns often signals a turning point, not just in asset prices but in broader economic activity.

One of the capital cycle’s most powerful applications is its ability to help investors avoid value traps—companies that seem attractively priced but are structurally impaired. Tesco serves as a cautionary tale: once considered a value opportunity, its heavy capital expenditures outstripped profitability for years, ultimately leading to significant erosion in returns. The key takeaway is that low valuations in sectors plagued by overcapacity are often misleading. A useful indicator here is the Capex-to-Depreciation Ratio. When this consistently exceeds 200% over a sustained period, it often points to overinvestment—a red flag for future underperformance.

SaaS, AI and Capital Cycle

That brings us to two of the most critical questions shaping investor sentiment today. The first centers on the SaaS market, which has seen a dramatic reset in valuations—currently trading at just 20–30% of the highs reached over the last decade. This collapse prompts a pressing dilemma: is this a compelling opportunity for mean reversion, where fundamentals reassert themselves and valuations rebound?

The SaaS business model offers several characteristics that potentially enhance resilience during economic uncertainty:

1. Recurring revenue streams that provide visibility and stability

2. High gross margins (typically 70-80%) that allow for operational flexibility

3. Mission-critical applications that customers are reluctant to abandon

4. Scalable infrastructure that can adapt to changing demand

However, the sector also faces challenges in the current environment:

1. Rising interest rates that disproportionately impact growth stocks

2. Increased scrutiny of software spending in corporate budgets

3. Competitive pressure from new AI-native applications 

4. Extended sales cycles as customers become more cautious

The key question for investors is whether the current valuation compression represents a cyclical downturn or or is it a classic value trap, where declining prices reflect deeper structural reassessment of SaaS business models in light of emerging technologies like AI that cheap valuations alone cannot fix?

To answer that, it’s essential to examine the second, more dynamic force now reshaping the landscape: the emergence of a new capital cycle driven by Artificial Intelligence. This AI buildout is not just another tech trend—it’s a systemic reordering of capital allocation, innovation pathways, and enterprise value creation. And in many ways, it casts a long shadow over SaaS, challenging old business models while creating new ones. Understanding where we are in this AI-driven capital cycle—early optimism, aggressive infrastructure buildout, and the beginnings of commercialization—helps frame whether SaaS is merely out of favor or being structurally displaced. Only by viewing SaaS through the lens of this broader shift can investors begin to distinguish between companies within the SaaS segment poised for recovery and those likely to be left behind.

The Current AI Capital Cycle

The AI investment boom is well underway, echoing classic capital cycles—marked by surging enthusiasm, aggressive spending, and rapid infrastructure buildouts. But unlike past bubbles, like the dot-com era, this cycle lacks the public equity feedback loop where announcements instantly inflate valuations. So while exuberance is high, we’re not yet in speculative territory.

Capital is flowing heavily into infrastructure: compute, networking, power, and data centers—mirroring past dual-layer cycles like railroads and land, or internet and telecom. Nvidia alone has added over $2 trillion in market cap in a year, while Microsoft, Google, and Meta have committed over $100 billion to AI. VC funding hit $50 billion in 2023, and AI data center construction is expected to exceed $30 billion in 2024.

However, bottlenecks like power supply and data center availability are emerging, with long lead times and early signs of overordering. It’s still early in the cycle, and uncertainty around returns is driving volatility.

We're firmly in the extractive phase, where foundational players—those owning silicon, models, and data—are capturing most of the value. Transformer architectures are nearing their efficiency limits, leading capital to consolidate around incumbents or flow to novel algorithmic approaches. In private markets, funding is skewed toward infrastructure and model builders, though Series A and B rounds have tightened due to a glut of weak startups and a crunch for high-quality ones.

Yet, signs of the distributive phase are emerging. Founders at OpenAI, Inflection, and Adept are stepping away as the focus shifts from R&D to commercialization. As models become commoditized, capital is shifting up the stack—to applications, agents, and services. Microsoft, for example, has already realized over $10 billion in AI revenue in 2024 alone.

Investment in AI today resembles a barbell: relatively safe bets on infrastructure and services on either end, with a crowded and chaotic middle where startups build co-pilots, tools, or middleware—some of which may thrive, many won’t.

AI is also transforming software architecture. Rigid workflows tied to systems of record are being replaced by AI-powered systems that adapt to users—think Gmail search instead of folder hierarchies. The opportunity now lies in compressing complex workflows. Today’s agents struggle with anything beyond 10 steps, but the real prize lies in automating the 100-step tasks that dominate enterprise operations.

Finally, as we move toward AGI, many current startup opportunities could be cannibalized by progress in foundation models. Tools like vector databases or LLM infrastructure platforms may be obsoleted by the very technologies they depend on. In this fast-moving cycle, the real challenge—and opportunity—is knowing what will endure and what will vanish.

And What About SaaS?

The rise of AI as a new computing paradigm is already reshaping the future of SaaS—sometimes quietly, sometimes dramatically. For investors, this shift isn’t just about buzz or headlines; it’s about understanding how core assumptions in the software business are changing.

A major shift is underway—from traditional SaaS to AI-native platforms powered by intelligent agents. Though still in its early stages, this transformation points to a future where clunky, interface-heavy applications are replaced by fluid, AI-driven systems that deliver outcomes directly from unified data sources. The linear workflows and rigid APIs that defined the traditional SaaS model are giving way to adaptive, learning-based architectures designed for seamless, personalized experiences. At the core of this evolution lies a harmonized data foundation—a single source of truth—without which AI agents cannot operate effectively. It’s not just a technical upgrade; it’s a fundamental rethinking of what enterprise software can and should be. 

Some SaaS products—especially those built around data analysis, content creation, or routine decision-making—are at real risk of being replaced by AI-native applications. Others stand to gain. Platforms that successfully integrate AI can repurpose their tech stack, redefine their value proposition, offer smarter automation, and gain pricing power. But doing so comes with trade-offs: AI workloads require different infrastructure and compute needs than traditional SaaS, potentially altering the economics of cloud delivery. At the same time, engineering talent and product focus are shifting—often away from traditional SaaS development and toward AI-first initiatives.

All of this helps explain why SaaS valuations have dropped so sharply. Many companies are now trading at just 20–30% of their 10-year averages. That collapse begs a fundamental question: is this an opportunity for mean reversion—or a sign of deeper structural change?

Seen through a capital cycle lens, much of the SaaS sector looks late-stage capital cycle asset class. Growth has slowed, customer acquisition costs are rising, and the competitive moat around many workflow-specific tools is eroding. Meanwhile, AI-native challengers are bypassing the legacy model altogether—offering more fluid, user-adaptive experiences that are no longer tied to rigid systems of record. Where software once shaped workflows, AI is enabling workflows to shape the software.

This convergence of falling valuations, rising disruption risk, and surging AI investment makes the SaaS landscape both confusing and compelling. One helpful way to think about it is a two-by-two framework: on one axis, a company’s exposure to AI disruption; on the other, its capacity—financial, operational and management ability—to transform. Companies that are both exposed and well-capitalized may be mispriced right now. Take Salesforce: with a massive global customer base and deep resources, it’s well-positioned to absorb AI into its product suite and unlock meaningful value. UiPath is another example—it already straddles the line between automation and AI, and with the right product evolution, its upside could be considerable.

But not everyone will make that leap. Smaller or mid-sized SaaS companies, especially those held in private equity portfolios with limited flexibility or high leverage, may struggle to reinvent themselves. Transformation isn’t just costly for these businesses—it might be infeasible. In such cases, lower valuations aren’t a bargain; they’re a reflection of deeper risk.

AI has shifted the ground beneath SaaS. The sector is no longer the go-to growth play it once was. That doesn’t mean it’s lost its potential—but it does mean investors need a more selective, methodical approach. The capital cycle teaches us that markets swing between excess and despair—and both extremes eventually correct. The real challenge is knowing whether a company is out of favor for the moment, or structurally out of step with where the market is heading.

In this environment, valuation compression isn’t a buy signal—it’s an invitation to look deeper.