Here's a sentence you wouldn't have expected to read in 2026: Warren Buffett is buying Alphabet, and the stock keeps falling anyway.
Google's parent is down 15.5% from its February peak of $344, sitting at $290. Extension and momentum percentiles have both dropped into the low teens — territory the stock has visited only a handful of times in the past five years. The signal scanner flagged it as an "extreme reading" yesterday.
The reflexive take is straightforward. Alphabet just posted its first-ever $400 billion revenue year. Google Cloud grew 48% year-over-year. Search revenue grew 17% — at a time when the narrative says search is dying. Gemini has 750 million monthly active users. Berkshire Hathaway just took a new position. This is obviously a dip to buy.
Maybe. But the data tells a more complicated story.
What's actually happening to the stock
Three forces are converging on Alphabet simultaneously, and they're worth separating because the market is conflating them.
The first is macro. The US-Iran conflict has sent oil past $100, which directly threatens Alphabet's cost structure. Running the world's largest AI infrastructure requires enormous amounts of energy. When Brent crude spikes 40%, Google's data centre costs follow. Ruth Porat warned at CERAWeek that the US isn't scaling power supply fast enough for AI demand. Rising energy costs are not an abstract concern for this company — they're a line item.
The second is the AI spending question. Alphabet guided $175 to $185 billion in capital expenditure for 2026 — nearly double the $91 billion spent in 2025, and far above the $120 billion Wall Street expected. Analysts warn that free cash flow across the hyperscalers could drop by up to 90% this year as capex outpaces revenue. The question investors are wrestling with has shifted from "Can Google do AI?" to "Can Google do AI profitably?"
The third — and this is the one the technical data actually captures — is the broader software selloff. The narrative that "AI is eating software" triggered a rotation out of the entire sector. Alphabet got caught in the downdraft despite being an AI beneficiary, not a victim. It's the difference between being hit by a recession and being hit by guilt-by-association. The stock is trading at a 36% discount to its peer median P/E ratio, partly because of genuine regulatory and competitive concerns, but partly because the market is painting all tech with the same brush.
Separating these forces matters because they resolve differently. The macro pressure is temporary (wars end, oil normalises). The capex question is structural (it either pays off or it doesn't, and we won't know for quarters). The sector rotation is sentiment-driven (it reverses when the next earnings cycle proves the "AI kills software" thesis wrong for quality companies).
What the percentiles show
Here's Alphabet's positioning as of March 24, 2026:
| Dimension | Percentile | Reading |
|---|---|---|
| Extension | 13th | Price is compressed relative to its moving averages — lower than 87% of the past year |
| Momentum | 14th | Rate of change indicators are near their floor |
| Flow | 33rd | Selling pressure is elevated but money hasn't completely fled |
| Volatility | 30th | The environment is relatively calm — this isn't a panic selloff |
The composite scores tell you "weak" — trend at 14, flow at 30. The same flat label they'd give to any stock that's been falling.
The percentiles tell you something more specific. Extension and momentum are in the danger zone (both below the 15th). But flow is holding above the 30th — selling pressure is elevated but not extreme. And volatility is calm at the 30th — this is an orderly decline, not a capitulation event.
That specific combination — oversold on extension and momentum, flow holding, volatility calm — has only occurred 6 times in Alphabet's history since November 2021.
Six observations, two very different outcomes
Here's where it gets interesting.
We tested every day where Alphabet hit this profile: extension below 20, momentum below 20, volatility below 40. Six observations. They cluster into two distinct episodes.
August 2022. Alphabet at $108. Extension at the 20th, momentum at the 14th, flow at the 43rd, volatility at the 36th. This was mid-decline — the stock was falling from $140 and hadn't finished. Flow was technically "holding" but CMF was deeply negative at -0.86. The stock fell another 10.6% over 21 days and 12.6% over 63 days. The calm environment wasn't a base being built — it was a slow bleed that still had further to go.
December 2022. Alphabet at $88. Extension at the 14th, momentum at the 16th, flow at the 34th, volatility at the 32nd. On the surface, almost identical to August. But this time the stock was near the end of a prolonged decline, not the middle. What happened? Up 10.8% at 21 days. Up 16.9% at 63 days. This was the bottom.
March 2024. Alphabet at $136. Extension at the 8th, momentum at the 10th, flow at the 43rd, volatility at the 38th. The tightest compression of the three. Result: up 12.7% at 21 days and an extraordinary 25.8% at 63 days.
The composite scores for all three episodes ranged from 10 to 17 on trend and 13 to 25 on flow. Indistinguishable. "Weak." But the forward returns ranged from -12.6% to +25.8%.
The discriminator
For Google, the volatile oversold episodes (volatility above the 60th percentile, which comprises the bulk of the 121 observations with low extension and momentum) averaged basically nothing — a median of -0.66% at 21 days and +0.66% at 63 days across 121 observations. The hit rate was barely above a coin flip at 47% and 52%.
The calm oversold episodes (volatility below 40, just 6 observations) produced a median of +10.8% at 21 days and +16.9% at 63 days, with an 80% hit rate at both horizons.
This is the opposite pattern to Tesla, where quiet breakdowns were catastrophic. For Alphabet, quiet oversold conditions have historically been the setup for significant recoveries. The logic is intuitive when you think about the companies differently. Tesla is a momentum stock — when it goes quiet during a decline, it means the decline is just warming up. Alphabet is a value stock dressed in growth clothing — when it goes quiet during a decline, it means the selling pressure is exhausting itself and the underlying business quality is providing a floor.
The one exception was August 2022, which caught the stock mid-decline. The key difference there? It was the first time in the cycle that extension and momentum hit these levels. By December 2022, the stock had been compressed for months — the oversold condition had persisted and deepened, and the eventual recovery was powerful.
The AI paradox
There's a fundamental tension in Alphabet's story right now that the percentile framework captures but can't resolve.
Google's search revenue grew 17% last year. That number alone demolishes the "search is dead" narrative. AI Overviews are generating more engagement, not less. Gemini has 750 million users. The Gemini 3 launch was a genuine milestone. Cloud grew 48%. By any operational measure, the AI transition is working.
And yet the market is selling the stock. Why?
Because the market has shifted from pricing AI as a revenue opportunity to pricing it as a cost problem. At $175 to $185 billion in capex — double last year — the question is no longer "Does Google have good AI?" It's "Will Google earn an adequate return on $185 billion?" That's a question that takes years to answer, and in the meantime, the investment compresses free cash flow and limits the multiple the market is willing to pay.
The 18 European industry groups urging action on Google's Digital Markets Act compliance add another layer of uncertainty. The DOJ antitrust case resolved more favourably than feared (no forced breakup of Chrome or Android), but ongoing restrictions on exclusive search deals could gradually erode distribution advantages. Ironically, the rise of AI competitors from companies like OpenAI and Anthropic actually helped Google's antitrust defence — the court acknowledged that alternative information pathways now exist.
This creates a situation where the business is performing well but the stock is being re-rated downward on cost and regulatory concerns. That's exactly the kind of environment where percentile extremes in extension and momentum tend to be meaningful — the selling is driven by narrative repricing, not fundamental deterioration.
What makes today's reading unusual
Today's profile matches December 2022 and March 2024 more closely than August 2022, but with one striking anomaly.
The risk profile score — which measures risk-adjusted return quality through Sortino ratio, drawdown characteristics, and upside/downside capture — sits at 91.2 out of 100. That's not just high. It's the highest reading of any Alphabet oversold episode by a wide margin. The December 2022 and March 2024 episodes both had risk scores around 42 to 55.
What this means: despite the 15.5% drawdown from the peak, the stock's risk-adjusted return profile over the past year is still excellent. The drawdown hasn't been deep enough or long enough to damage the Sortino ratio or capture metrics. The underlying trend — Alphabet up 65% through 2025 — is so strong that this pullback registers as a correction within a structural uptrend, not a trend reversal.
That's a meaningful distinction. It suggests the market is re-rating the multiple, not abandoning the thesis. Investors are paying less per unit of expected earnings, but the earnings themselves aren't in question.
The CMF reading adds another nuance. In August 2022 (the bad episode), CMF was at -0.86 — deeply negative selling pressure despite the flow percentile looking moderate. In December 2022 and March 2024, CMF was similarly negative (-0.82 and -0.74). Today? CMF is at +0.14. Positive. Money is actually still flowing in on a net basis, even as the stock declines. The selling is coming from price-based momentum, not from institutional distribution.
The software narrative versus the data
The "AI is eating software" trade has hammered the entire sector. The fear is that AI agents will replace the need for traditional SaaS products — why pay for Salesforce when an AI can manage your CRM? Why pay for ServiceNow when an agent handles your IT tickets?
Google's CEO Sundar Pichai pushed back on this directly on the Q4 earnings call, arguing that AI is an enabling tool and that the best software companies will use it to become better cloud customers. The 48% growth in Google Cloud revenue suggests he might be right — companies aren't replacing their cloud infrastructure because of AI, they're buying more of it.
But here's what the percentile data adds to this debate. The current selloff has pushed Alphabet's extension and momentum percentiles to levels last seen at major bottoms — December 2022 and March 2024. Both of those were followed by significant multi-month rallies. The market was pricing in fears that turned out to be overstated.
In December 2022, the fear was that ChatGPT would kill Google Search. Search revenue then grew 17%. In early 2024, the fear was that Google was falling behind in AI. Google Cloud then grew 48% and Gemini launched to 750 million users.
The pattern isn't "buy every dip." The August 2022 episode proves that mid-decline calm compression can lead to further pain. The pattern is that when Alphabet reaches these percentile extremes, the market is pricing in a narrative that tends to overstate the threat to the actual business. Whether today's narrative — AI capex destroying free cash flow — will follow the same trajectory is the question.
What to watch
Three things would increase conviction that this is a December 2022 / March 2024 setup rather than an August 2022 mid-decline:
Flow percentile holding above 20. Currently at 33. If it deteriorates below 20 — suggesting genuine institutional selling rather than sentiment-driven drift — the character of the selloff changes. The prior genuine bottoms had flow between 33 and 43. If flow holds at these levels or recovers, it supports the base-building thesis.
Volatility staying calm. At 30, the environment is orderly. If volatility spikes above 60, it would suggest the selloff is escalating from an orderly repricing into something more disruptive. For Google specifically (unlike Tesla), calm oversold conditions have been the better setup. You want the stock to stay boring while the percentiles are low.
Q1 2026 earnings. The April report will be the first test of whether $185 billion in capex is delivering proportional revenue growth. If Cloud growth accelerates above 50% and search maintains its 15%+ trajectory, the capex narrative starts to crack. If growth decelerates, the repricing deepens.
The data says Google is at a percentile extreme that has historically preceded significant recoveries — when the conditions are calm, flow is holding, and the business is intact. All three of those are true right now. The risk profile score at 91 suggests the market still views the underlying business favourably even as it re-rates the multiple downward.
This isn't a guarantee. August 2022 is a reminder that not every calm compression resolves upward. But with 6 observations, 5 of them positive at 63 days, and the one negative episode distinguishable by its position mid-decline rather than at a potential bottom, the weight of the data leans constructive.
The market is asking whether $185 billion in AI spending will pay off. Google's price is at the 13th percentile of its range while its search revenue grows 17% and Cloud grows 48%. Something has to give. The percentiles suggest it might be the price that adjusts upward.
Data as of March 24, 2026. Analysis based on GOOGL daily indicator data from November 2021 to present. Percentile calculations use a rolling 365-day window. This is not investment advice — it is a description of how the current conditions compare to historical episodes, based on our indicator framework. Past patterns do not guarantee future outcomes.