Cloud Backlog Surges 150%, JPMorgan: Tech Stock Valuations Have Bottomed
Miles Bennett
Hyperscaler cloud backlogs jumped over 150% year-on-year to roughly $1.7 trillion, far outpacing capex growth; JPMorgan argues Big Tech's valuation compression is near its end, with potential upside of 56% if multiples revert to the long-run mean.
Backlogs are growing twice as fast as capex — what does that tell us?
Hyperscaler cloud backlogs — contracted revenue not yet delivered — rose over 150% YoY to about $1.7 trillion; capex grew roughly 80% over the same period.
This means → every dollar of AI infrastructure spend is locking in more future revenue than before — the payoff is outrunning the market's expectations.
In plain terms = the cloud giants are building data centers fast, but customers are lining up even faster. That gap is the clearest sign demand is real, not just a spending spree.
What are the CEOs actually saying about future demand?
Amazon CEO Andy Jassy made a rare upward revision on AWS's long-term scale during the Q2 earnings call: the prior view was "hundreds of billions" in revenue; he now sees at least double that, with a path to $1 trillion in annual revenue.
He added that the company already sees 2028 demand at a scale he called "staggering" — and enterprise adoption of AI inference is still in its early stages.
Microsoft, Alphabet, and Meta management echoed the message: AI commercialization is just getting started; enterprise demand is far from mature.
This means → it is not one CEO being optimistic — the entire hyperscaler cohort is signaling that current AI demand is only the tip of the iceberg.
How far have valuations actually fallen?
After the July correction, the S&P 500 IT sector's forward P/E dropped to about 20× — the 1st percentile of its ten-year historical range, below the decade average of roughly 23×.
Large-cap tech stocks (excluding semis) now trade at forward multiples more than 2 standard deviations below the mean since 2018.
In plain terms = plot the last decade of valuations on a line and today's price sits almost at the very bottom — earnings are rising while the price tag is shrinking.
If multiples recover, how much room is there?
JPMorgan's math: a reversion to 1 standard deviation below the historical mean implies about 30% upside; a return to the long-run mean implies roughly 56%.
Hyperscaler performance relative to the S&P 500 has fallen to the bottom of its three-year range — historically, similar readings have been followed by strong mean reversion.
This reflects a mismatch: fundamentals are accelerating while sentiment has pushed valuations to historic lows — and that gap is exactly why JPMorgan calls this a bottom.
Has institutional money caught up yet?
Deutsche Bank data show institutional positioning in large-cap tech is only modestly overweight — well below the allocation levels seen in past strong-earnings cycles.
This year's flows have concentrated in semiconductors; large-cap tech ex-semis remains under-owned.
This means → if the market narrative shifts from "is capex excessive?" to "returns are being delivered," there is significant room for institutions to add. The next leg up may come from rotation within tech rather than another semiconductor rally.
What is the technical picture saying?
The MAGS index has rebounded nearly 10% from its recent low, reclaiming the 200-day moving average and approaching the long-term uptrend line dating to April last year.
The 200-day moving average is flattening, indicating a prolonged consolidation phase.
In plain terms = consolidation is like a compressed spring — history shows the longer the sideways grind, the sharper the breakout once direction is chosen. Whether institutional positioning can keep pace with improving fundamentals is the key variable for this valuation recovery to materialize.
Content is for reference only, not financial advice.