AI Got Important Enough To Bomb

Aug 6, 202610 min readNewsletter
GROW SMART INCOME
A weekly briefing on AI · Markets · Power
Week 32
Aug 3–9, 2026
AI Got Important Enough To Bomb
Data centers under fire, a startup that didn’t exist in January just landed $10 billion, and Europe’s old guard becomes AI’s quiet winner.
S&P 5007,723.55 −0.2%
Nasdaq26,363.44 −0.83%
Bitcoin$64,838 +0.84%
WTI$74.97 −1.06%
This week the market fears:
A war that’s turning AI’s own infrastructure into a target
Est. reading time: 6 min
The Week in 60
This week, AI’s infrastructure crossed a line: it became something worth bombing, not just backing. Iran’s strikes on Gulf data centers made physical war a line item for the industry — right as Anthropic committed $10 billion to a compute startup that didn’t exist a year ago. Europe’s old tech firms turned out to be AI’s quietest winners this earnings season.
Big Idea

AI infrastructure has become too economically important to leave undefended — which is exactly why it’s now getting hit. This year marked the first time data centers have been physically attacked in war, not just hacked.

Iran’s Revolutionary Guard published a target list naming 29 regional tech facilities. Gulf data centers have already absorbed multiple confirmed strikes, and the US hit one of Iran’s in return. Brent crude spiked as a $2 trillion regional tech buildout suddenly looked exposed to missiles, not just markets.

AI infrastructure got important enough to bomb before it got old enough to trust.

That line should worry anyone tracking how fast the capital behind AI is still moving, because it isn’t slowing down for any of this. Four days after the strikes made headlines, Anthropic signed a $10 billion, six-year compute deal with Volta — a company that didn’t exist eight months ago.

It’s one of at least two $10 billion bets Anthropic has in motion right now, alongside separate talks with Meta of the same size. Diversifying suppliers is sound risk management on its own. War zones remain the exception, not the rule.

But stack the two trends and a pattern appears. The industry’s appetite for compute is now outrunning its own appetite for scrutiny — and nobody buying capacity this fast has time to ask hard questions about who’s actually selling it.

The Model
THE COMPUTE GOLD RUSH LOOP
AI becomes economically indispensable
↓
Its infrastructure becomes a strategic target — Gulf data centers under fire
↓
Risk on physical AI infrastructure should rise
↓
Capital flows in anyway — into newer, less-vetted counterparties
↓
Dealmaking speed outpaces due diligence
The real vulnerability isn’t the missiles — it’s that no one in this industry can afford to slow down long enough to check who they’re actually buying compute from.
By the Numbers

ANTHROPIC’S COMPUTE SUPPLIER LIST

Reported deal / commitment size, $ billions

0 10 20 30 $25B Amazon $30B Google/Broadcom $5B AMD $10B Volta $10B* Meta*

Source: TheNextWeb, citing Bloomberg, Aug 4 2026 · *Meta deal reported in talks, not signed

Anthropic now spreads compute across at least five suppliers, and the newest one — Volta — didn’t exist when this year started. Diversification is smart insurance, but it also means due diligence on counterparties is happening faster than ever.

THE SKILLED-TRADES GAP BEHIND THE AI BUILDOUT

US workers, thousands

Electricians needed by 2030 130K Construction laborers needed 240K Construction supervisors needed 150K Current factory worker shortfall 600K Current construction worker shortfall 500K

Source: Fortune, citing McKinsey Global Institute and Ford CEO Jim Farley, Aug 1 2026

The projected need (orange) is smaller than the shortfall that already exists today (green). Nvidia’s Jensen Huang is right that the buildout needs tradespeople — the problem is the US was already short before AI added to the bill.
Signal vs Noise
Trump-linked robotics startup arms up for war
Foundation’s Eric Trump-advised humanoid robots are headed toward frontline military testing, blurring the line between venture capital and defense policy.
China drops two frontier models in one week
DeepSeek’s V4 Flash and Alibaba’s Qwen 3.8-Max now match Western models on benchmarks while costing roughly 40% less to run.
Europe’s old tech guard becomes AI’s quiet winner
SAP, Capgemini, and OVHcloud are profiting more from making AI usable inside legacy systems than the labs building the models.
Nvidia’s CEO says the next boom wears a tool belt
Huang says AI’s buildout needs plumbers and electricians as much as engineers — and the US doesn’t have enough of either.
AI leaders now claim we’re “in” the singularity
Altman and Hassabis say we’re near or past the threshold — but the people selling AI have every reason to say that.
What I’m Reading
Chip War: The Fight for the World’s Most Critical Technology
Chris Miller · 2022 · Scribner
I picked this up again this week because the Iran data-center strikes made the argument feel current rather than historical. What I agree with: Miller’s case that chips became the resource wars get fought over, the way oil did last century. What I find incomplete: the book was written before AI compute became the specific chokepoint — it explains the chip war, not yet the compute war we’re watching now. What I’m taking from it: infrastructure that’s “too important to fail” is also infrastructure that’s too important not to attack, and the AI industry hasn’t priced that in yet.
One Number
29
That’s the number of Gulf tech facilities Iran’s Revolutionary Guard has publicly named as potential targets this year. Five have already been hit — the first confirmed physical strikes on data centers in the AI era.
This Week I Noticed

Google added AI image generation to Google Earth last week, letting anyone reimagine any spot on the planet with a prompt. Within 48 hours, per The Register, it pulled the feature after people started generating and sharing manipulated satellite imagery that violated Google’s own policies. Google says people “uniquely trust” Earth for a reliable view of the world — which is exactly why building a fabrication tool into it lasted about as long as a bad idea should.

Quote
“Everybody should be able to make a great living. You don’t need to have a PhD in computer science to do so.”
— Jensen Huang, Nvidia CEO
The optimistic counterweight to a week that also included armed robots and a $10 billion bet on an eight-month-old company — a reminder the AI buildout creates jobs outside Silicon Valley too.
Three things to remember
AI infrastructure is now a physical war target, not just a cyber one — that’s new as of this year.
The industry’s compute appetite is outrunning its own risk controls, with billion-dollar bets going to companies that didn’t exist twelve months ago.
The AI boom’s most durable winners this earnings season weren’t the model-makers — they were Europe’s decades-old integrators.
One Thing To Do
If you hold positions in AI infrastructure plays, check exactly which counterparties they’re exposed to this quarter. A supplier you’ve never heard of, backed by a $10 billion contract, isn’t automatically a red flag — but it isn’t automatically fine either.
Support This Newsletter
Sponsor
Trading 212
No account yet? Use my link — you get a shot at free shares, and you help keep this newsletter going. No catch.
Open Trading 212 →
© 2026 Grow Smart Income · by Kaloian Parchev

Enjoyed this article?

Get the GSI Weekly Newsletter — markets, AI, and investing insights every Thursday.

Subscribe for free →

No spam. Unsubscribe anytime.

Leave a Reply