The S&P 500 crossed $70 trillion in market cap for the first time last Tuesday, seven months after ending 2025 at $61.1 trillion. On this week’s Bricks, Bucks & Bytes rundown, Dustin DeVan argued that if you strip the data center buildout out of US GDP, the rest of the economy is basically stagnant – a claim Morgan Stanley’s estimate that AI capex drove roughly 75% of Q1 growth broadly supports. That leaves the Fed with no clean move, and leaves construction as the industry most exposed to a boom running on one engine.
Martin brought the milestone to the table: the S&P 500’s total market cap topped $70 trillion, which puts 500 companies at more than double the value of the entire US economy. His question was simple. Is there a real productivity gain for ordinary people underneath this, or is a handful of companies dragging the index while everyone’s pension rides along?
Dustin had actually been chewing on the same thing during his morning walk, which tells you something about what construction tech CEOs think about before 7am these days.
$8.9 Trillion in Seven Months
The milestone math, and why it made everyone slightly nervous
The raw numbers first. The index crossed $70 trillion in aggregate market capitalization on Tuesday, August 4, having ended 2025 at $61.1 trillion. That’s roughly $8.9 trillion of new market value in about seven months, a 14.6% gain in aggregate terms.
Martin’s framing on the show was the uncomfortable one: this happened while US mortgage rates stayed elevated and the Fed held its ground on interest rates. The index is sprinting. The cost of borrowing for a normal household or a mid-market contractor hasn’t budged in the right direction.
A gap that wide between asset prices and lived economics usually has a specific cause. This one does too, and it’s made of concrete, copper, and GPUs.
The 0.1% Economy
Take out the data center buildout and almost nothing is moving
Dustin’s back-of-envelope claim was the sharpest moment of the episode. Worth quoting in full, hedge included.
If you exclude data centers from GDP, the rest of the market, and if I’m wrong, I’m not materially wrong, has only grown 0.1% in the US, which is basically stagnant.Dustin DeVan, founder and CEO, Ediphi
The precise decimal is his estimate, but the direction is well documented. Morgan Stanley’s analysis, circulated widely in May, put AI-related capital spending at roughly 75% of all US economic growth in Q1 2026. The St. Louis Fed found AI categories made up 39% of total GDP growth through the first three quarters of 2025, already ahead of the dot-com era’s peak contribution. Strip the buildout, and growth flatlines.
Which is what makes the Fed’s position so strange. Raising rates is supposed to cool an overheating economy. But the entities doing the overheating – Google, Meta, Microsoft, Amazon – can largely finance their capex from their own balance sheets. As Dustin put it, hike rates and “all we’re really doing is hurting the rest of the economy who can’t afford it,” while the hyperscalers keep pouring foundations. His read on the central bank’s likely posture: sit and hold, because there’s no lever that touches the boom without breaking everything around it.
The macro forces repricing your backlog, explained weekly
Data centers, rates, materials, and the deals that decide who wins the next cycle.
Join 3000+ ReadersDon’t Dance, or Maybe Do
Martin’s pension worry meets Patric’s counter-thesis
Martin’s concern is the one most people share. “When the music stops for the index, then the little people might get hurt. Even if they didn’t participate.” Pensions sit in these indexes. A two-chamber economy is fine right up until the chambers reconnect violently.
Patric reached for The Big Short, the scene where Brad Pitt’s character tells two young traders celebrating their bet against the housing market to show some restraint. Then he argued the opposite case anyway, as a thought experiment. His claim: the US effectively moved away from fractional reserve requirements in 2020, and market makers now have sanctioned mechanisms to create liquidity in equities. If liquidity can always be manufactured, historical precedent about bubbles popping may simply not apply. “I’m not so sure that there is actually a crash coming,” he said, with the caveat that reinstating reserve requirements would change his answer fast.
Nobody on the show fully bought anybody else’s position, which is roughly where the honest macro debate sits right now. Dustin’s own candidate for the trigger, circular financing between AI companies, deserves its own article. Watch this space.
The Construction Read
The strongest lane and the biggest exposure are the same lane
For our readers, this story lands directly on the jobsite. Hyperscaler data center spending is the main thing keeping US construction in positive territory while non-residential spending drifts and residential stays locked up. ABC’s chief economist made that case explicitly back in April: enough activity in data centers, energy, and AI infrastructure to keep the industry moving, with fragility baked in because the growth is so concentrated.
So the same fact cuts both ways. If you’re a GC or specialty sub in the hyperscaler lane, you’re attached to the only strong buyer in the market. If that buyer’s capex cycle wobbles – because rates finally bite, because AI revenue disappoints, because the index rolls over – the backlog that looked generational gets thin quickly. Scenario planning beats cheerleading here. Know which of your cost lines can flex, and which client relationships survive a soft cycle.
| Signal | Number | Source |
|---|---|---|
| S&P 500 aggregate market cap | $70 trillion, first crossed August 4 | TechTimes / S&P data |
| Gain since end of 2025 | +$8.9 trillion (14.6%) | TechTimes |
| AI capex share of Q1 2026 US growth | ~75% | Morgan Stanley, via Yahoo Finance |
| AI categories’ share of 2025 GDP growth (Q1-Q3) | 39% | St. Louis Fed |
| Index value vs US GDP | More than double | Syz Group |
The index added roughly $8.9 trillion in aggregate market value between the end of 2025 and early August 2026, a 14.6% gain driven heavily by large-cap technology and AI infrastructure names. The $70 trillion threshold was crossed on Tuesday, August 4, alongside an intraday price record. (Source)
Close to it, by several estimates. Morgan Stanley’s analysis put AI-related capex at roughly 75% of Q1 2026 US growth, and the St. Louis Fed found AI categories contributed 39% of GDP growth through the first three quarters of 2025. Dustin DeVan’s 0.1% figure for the rest of the economy is his own back-of-envelope estimate, but the direction matches the published research. (Source)
Because the boom’s drivers are unusually rate-insensitive. Hyperscalers can fund much of their data center capex from cash flow, so higher rates would land hardest on households, small businesses, and contractors outside the AI lane while barely slowing the buildout. That’s the trap Dustin described on the episode: the Fed’s main tool punishes everyone except the source of the heat.
The buffer disappears fast. Data center and AI infrastructure work has been the primary force keeping US construction spending positive while manufacturing, residential, and broader non-residential soften. A capex pause would expose how concentrated the industry’s growth has become, which is why economists like ABC’s Anirban Basu keep flagging recession risk despite headline activity. (Source)
Related Articles
Data Centers Are Keeping US Construction Afloat – But the Rest of the Market Is Struggling
Big Tech Promised $650 Billion in Data Centers. Most of It Isn’t Being Built
BlackRock Now Owns 80% of Meta’s $14 Billion AI Data Center
