NextFin News - U.S. software stocks are living through a year of violent whipsaws, and the swings are far from over. The S&P 500 software and services index has fallen more than 33% from its October record, rebounded 33% during earnings season, sold off again, and still trades down more than 3% for the year — all while a rapid expansion of leveraged ETFs and momentum-driven trading is magnifying every move.
The roller coaster is usually blamed on artificial intelligence. A January product release from Anthropic accelerated a selloff as portfolio managers worried AI would render many software products obsolete. But the intensity of the swings points to something deeper than a single competitive threat. The same sector posted second-quarter aggregated earnings growth of 24.4%, well above the 16.4% expected before reporting began, according to earnings research data — a fundamental picture that does not match a sector in existential decline.
That mismatch between the earnings reality and the price volatility is the story. Software is caught between a cyclical wave of AI anxiety that is already mean-reverting and a structural change in how the market trades that is not going anywhere. The second force is the one investors are underpricing.
The Index That Cannot Decide What It Is
The S&P 500 software and services index hit a record high on October 28. By April 10 it had given up more than a third of its value. It then rebounded 33% through a mostly solid first-quarter earnings season, sold off again, and partly recovered during a largely healthy second quarter. As of late August the index is down more than 3% for the year and more than 12% below its record.
That path is not the chart of a sector receiving a steady stream of bad news. It is the chart of a market that reacts violently in both directions. Microsoft captures the pattern: the world's largest software company rallied 29% over eight sessions after its July 29 earnings release, even though the stock had been stuck in reverse for much of the year, down about 20% year to date before the report. Options traders had positioned for a roughly $190 billion swing in Microsoft's market value around that earnings print — an unusually large move for a company of its size, and a signal that the market was pricing uncertainty, not a verdict.
The earnings data undercut the doom narrative. Based on results and estimates, the index was on track for second-quarter aggregated earnings growth of 24.4% versus expectations of 16.4% as of July 1, before reporting began, according to Tajinder Dhillon, head of earnings research. When a sector grows earnings nearly 25% and still trades more than 12% below its record, the explanation is not fundamentals alone.
The next test arrives quickly. Salesforce and CrowdStrike report quarterly results Wednesday afternoon, followed by Oracle in mid-September. Each will be read not only for what it says about AI competition but for how violently the shares move afterward — and that reaction function is where the structural change shows up.
The Mechanism: Momentum Trading and Leveraged Plumbing
Even for an industry facing genuine AI disruption, the intensity of the swings is a testament to the power of momentum trading — buying shares that are rising and selling those that are falling — and the rise of leveraged funds that can double or triple daily moves. These are not new strategies. What is new is their scale, and scale changes market structure.
The leveraged ETF segment has exploded in the last two years. The number of single-stock leveraged or inverse ETFs has surged to 486 from 28 at the end of 2023, according to ETF industry data. About 40 of those products target the software-infrastructure category and 23 are tied to software-application companies. In the last two months alone, Corgi Investments, a newcomer to the ETF industry, rolled out a suite of inverse ETFs tied to individual stocks and indexes, including 11 products based on individual software firms such as Microsoft, Oracle, AppLovin and ServiceNow.
The mechanical reason this amplifies volatility is straightforward. Leveraged ETF providers rebalance their exposure to the underlying stock or index at the start of each trading day to maintain a constant multiple of daily returns. When a stock rises, a 2x or 3x product must buy more into strength; when it falls, it must sell into weakness. That daily rebalancing flow arrives at the same time every day, in the same direction as the move, and it makes momentum self-reinforcing at exactly the moments when prices are already moving fastest.
"That's where you tend to see the most of that in action, in those momentum names. Those are folks that are most likely day traders that are fueling the momentum through leveraged products," said Joe Saluzzi, co-founder and head of equity market structure research and co-head of equity trading at Themis Trading.
"It tends to slow down when things are going the other way for a bit. And then once it starts heating up again, they get right back involved, and you can see the options market just explode with volume."
Tony Welch, chief investment officer at SignatureFD, drew the line between the fundamental spark and the momentum fuel. Rallies begin because market participants detect an impending fundamental improvement, he said, but then they build on themselves.
"These rallies, they start because there is an impending fundamental improvement that the market participants sniff out, but then they continue to build on themselves with momentum," Welch said, adding that "as you get later into a move it does tend to be more of a momentum driver."
The implication is uncomfortable for anyone trying to value software stocks on fundamentals. Early in a move, price action may still track earnings revisions and guidance. Later, the move is being driven by flows that have nothing to do with the business — and those flows reverse just as mechanically when momentum turns.
Cyclical Fear, Structural Amplification
The first question is whether this is a cyclical fluctuation that will mean-revert or a structural shift that will not. The answer requires separating two forces that are operating at the same time.
The AI-anxiety leg is cyclical. It was triggered by a specific event — Anthropic's January product release — and it is already being corrected by the earnings data. Three historical-cycle comparisons make the point. First, the 33% decline from the October record to April 10 was followed by a 33% rebound during first-quarter earnings, a near-symmetric mean reversion once actual results arrived. Second, the sector sold off again and then partly recovered during the second quarter, repeating the same pattern within a single year. Third, software has repeatedly absorbed technology shocks — the client-server transition, the rise of open source, the shift to cloud subscriptions — without the sector's earnings power being permanently impaired. Each time, the initial fear priced in obsolescence that never fully arrived.
The evidence for mean reversion is concrete: second-quarter aggregated earnings growth of 24.4% against a 16.4% pre-report expectation is not the profile of a sector whose products are being made obsolete. It is the profile of a sector where the demand shock was overstated and earnings are catching up to the panic.
The leveraged-trading leg, by contrast, is structural. It rests on three pieces of evidence that a regime has changed rather than a sentiment wave that will pass. First, the product count — 486 single-stock leveraged or inverse ETFs versus 28 at the end of 2023 — is a seventeen-fold increase in roughly two and a half years. Products do not disappear when sentiment cools; the shelf space is now built. Second, the daily rebalancing mechanism is a permanent feature of how these funds operate, not a discretionary positioning choice. Third, new entrants are still arriving: Corgi's suite of 11 single-name inverse software products launched within the last two months, meaning product innovation is accelerating into the volatility rather than retreating from it.
Blending the two forces produces the wrong verdict. An investor who concludes that software volatility is purely an AI-fear cycle will expect calm to return once earnings prove resilient — and will be surprised when the next good-earnings day still produces a double-digit whipsaw. An investor who concludes the whole move is structural will miss that the earnings rebound is real and that valuations have already been repriced. The correct read is layered: the direction of travel is being set by cyclical fundamentals that are improving; the magnitude of every daily move is being set by structural leverage that is not going away.
The Second-Order Risk the Market Is Not Pricing
The conventional worry is that software earnings will disappoint and the sector will fall. That risk is obvious, widely stated, and already reflected in an index trading 12% below its record. The second-order risk is different: even if earnings keep improving, the market's new leveraged plumbing means volatility itself becomes the baseline.
Trace the chain. The first-order effect is mechanical: leveraged ETF rebalancing amplifies daily moves in the names they track. The second-order effect crosses into expectations: when daily moves are amplified, risk managers and systematic strategies respond by cutting position sizes or raising hedging, which raises the cost of holding the stocks and compresses the multiple investors are willing to pay for the same dollar of earnings. The third-order effect is an expectation gap: the market has priced the possibility that software earnings disappoint, but it has not fully priced the possibility that software earns well and still trades like a volatile asset class.
That gap is visible in the Microsoft setup. The company delivered a fiscal fourth-quarter beat — adjusted earnings of $4.74 a share on sales of $90 billion for the quarter ended June 30 — and the stock still carried options-implied expectations of a $190 billion market-value swing around the print. A beat was not enough to produce a clean re-rating; it produced a battle between fundamental improvement and leveraged positioning, and the stock remained well below its prior highs even after the rally.
Marta Norton, chief investment strategist at Empower, framed the asymmetry directly. Current prices reflect what investors have already seen from earnings and an expectation that the earnings stream will continue — which means the upside from here requires earnings to exceed an expectation that is already elevated, while the downside opens if that stream breaks.
"Right now, the prices reflect what we've already seen from an earnings perspective, and they kind of reflect an expectation that the earnings stream will continue," Norton said. "But to the extent that that is not true, then there is real risk in these stocks."
The counter-thesis is that this is all just a positioning cycle that will normalize on its own. Leveraged products are not new — they existed through the 2020 volatility spike and the 2022 bear market, and volatility did eventually compress in both episodes. If retail interest cools and flows stabilize, the amplification fades without any permanent change to how software stocks are valued. Under this view, today's volatility is a tax on impatience, not a regime shift, and the sector's earnings recovery will reassert a calmer trend as it has before.
That argument is plausible but incomplete. It treats leveraged products as a sentiment variable when they are increasingly a structural one. The difference is in the product count and the rebalancing rule. Sentiment-driven positioning shrinks when traders lose interest; a 486-product ecosystem with daily mandatory rebalancing does not. The signal that would prove the structural call wrong is specific: if the number of single-stock leveraged and inverse ETFs falls back below 200 over the next four quarters — a sustained contraction of more than half — while software-sector daily realized volatility compresses to its pre-2024 range, then the amplification was a cycle after all. Until that happens, the burden of proof sits with the normalization view.
What Comes Next
The near-term path runs through the remaining earnings reports. Salesforce and CrowdStrike report Wednesday afternoon, and Oracle follows in mid-September. Strong results met with muted price reactions would signal that fundamentals are beginning to dominate the leveraged flows — the first sign that the cyclical leg is winning. Strong results met with violent whipsaws in both directions would confirm that the structural amplification remains in control.
Split by time horizon, the picture is mixed. In the short term, sentiment and liquidity dominate: the next leg is set by the earnings calendar, the options positioning around those prints, and whether leveraged ETF rebalancing flows add fuel. In the medium term, fundamentals should gain influence: earnings growth running near 24% against 16% expectations is a tailwind that typically supports multiples, and the sector's 33% rebound from the April lows shows the re-rating mechanism still works. In the long term, the structural question is whether AI is a substitute for software spending or a complement that raises it — and the earnings data so far points to complement, not substitute.
Three scenarios frame the path. The base case is continued elevated volatility with a slowly rising trend: earnings keep improving, the AI-obsolescence narrative fades, but daily moves remain outsized because the leveraged product base stays large. The upside case is a clean fundamental re-rating: several consecutive strong software reports arrive with calm price action, leveraged flows stabilize, and the index reclaims its October record. The downside case is an earnings break: if Salesforce, CrowdStrike, or Oracle guide lower on AI-driven substitution, the same leveraged plumbing that amplified the rebound would accelerate the decline, and the 12% drawdown from the record would look like the first leg rather than the last.
The signal to watch is not just the earnings numbers but the market's reaction function to them. If good news stops producing violent selloffs, the cyclical forces are winning. If good news keeps getting whipsawed, the structural amplifiers are still in charge.
The software sector's problem is no longer whether AI will replace its products — the earnings say it has not. The problem is that the market now trades those products through leveraged plumbing that turns every piece of news into a roller coaster, and that plumbing was built to stay.
Explore more exclusive insights at nextfin.ai.
