EBM Newsdesk Analysis
MOUNTAIN VIEW, April 30 — On 29 April 2026, Alphabet reported Q1 revenue of $109.9 billion, up 22% year-on-year, with Google Cloud revenue surging 63% to $20.03 billion and operating income tripling to $6.6 billion. Capital expenditure for the quarter alone hit $35.67 billion. CFO Anat Ashkenazi confirmed full-year 2026 capex will land at $180-190 billion, with 2027 expected to “significantly increase”. The same earnings cycle also delivered confirmations from Amazon, Meta and Microsoft. The four hyperscalers collectively now plan to spend $725 billion on AI infrastructure in 2026 — a 77% increase on last year’s record $410 billion. CEO Sundar Pichai’s stated reason: “We are compute constrained in the near term.” The Business DeskCNN
The deeper read sits in what most coverage will miss. The four hyperscalers’ combined 2026 AI capex commitment now exceeds the annual GDP of the Netherlands ($1.1 trillion adjusted), Switzerland ($885 billion), or the combined GDP of Greece, Portugal, and Hungary. Approximately none of this $725 billion is being deployed in Europe. The capital allocation question this raises is structural and uncomfortable: when a handful of US corporations commit infrastructure spending equivalent to a mid-sized European economy and locate almost all of it outside Europe, what does European AI sovereignty actually mean in practice?
What the $725 Billion Actually Buys
Capital expenditure of this magnitude reshapes global supply chains, energy markets, and labour pools simultaneously. Three structural consequences worth tracking:
Join The European Business Briefing
New subscribers this quarter are entered into a draw to win a Rolex Submariner. Join 40,000+ founders, investors and executives who read EBM every day.
SubscribeSemiconductor demand at unprecedented scale. TSMC, Samsung, and SK Hynix, which collectively delivered the MSCI Emerging Markets index’s record April rally, now have a guaranteed customer base committing nearly three-quarters of a trillion dollars to AI infrastructure. The earnings momentum traces almost entirely to this hyperscaler capex.
Energy infrastructure constraints. Hyperscaler data centres at this scale require power generation capacity equivalent to small countries. The US grid is already constrained; the next round of capex is partly chasing power availability rather than just compute. Texas, Virginia, and Arizona are absorbing the largest share.
Labour market displacement. Specialist AI infrastructure engineering talent is being concentrated in hyperscaler payrolls at compensation levels European firms cannot match. Mistral, Ineffable Intelligence, and AMI Labs are competing for talent against budgets denominated in tens of billions.
For European businesses dependent on AI infrastructure for productivity gains, the $725 billion commitment is good news (more capacity arriving) and bad news simultaneously (the strategic ownership of that capacity sits entirely with four US corporations).
The Backlog Number That Matters More Than The Quarter
Buried inside Alphabet’s results is the genuinely consequential figure: Google Cloud backlog almost doubled quarter on quarter to more than $460 billion. That’s the future contracted revenue Google has signed but not yet delivered. It tells you two things. NPR
1. Enterprise AI demand is not slowing. $460 billion of forward contracted commitments means customers are not waiting to see if AI capex pays off — they’re locking in infrastructure access years ahead.
2. The hyperscaler oligopoly is hardening. Customers signing multi-year cloud commitments to Google (or AWS or Azure) are structurally locked in. Switching costs at enterprise scale make the relationships durable in ways that increase hyperscaler pricing power over time.
For European enterprises, this is the dynamic Boltzbit CEO Yichuan Zhang flagged in his recent EBM commentary on AI vendor dependency. The infrastructure most European businesses are building their AI deployments on is now contracted years forward to American providers, with switching costs that compound over time.
The Iran War Backdrop
Worth noting: all four hyperscalers reported results on Wednesday, updating investors for the first time since the US began combat operations in Iran in late February. The earnings cycle landed against energy prices at $111 a barrel, European stagflation pressure, and 5%+ UK gilt yields. None of it dented hyperscaler results. Al Jazeera
That structural insulation — US tech infrastructure earnings continuing to compound while energy-dependent and consumer-facing European business absorbs the war’s costs — is the genuine asymmetric risk European institutional investors should be processing. AI infrastructure exposure has become its own asset class, structurally uncorrelated with the geopolitical and energy variables that hit traditional sectors.
What to Watch From Here
Three signals matter from now through year-end. First, whether the $725 billion 2026 commitment translates into European data centre allocation announcements (currently negligible) or whether the capital concentration in US territory deepens further. Second, whether European AI labs successfully attract significant equity investment to compete at infrastructure scale, or whether the funding gap widens beyond catch-up territory. Third, whether the hyperscaler backlog growth signals durable enterprise demand or speculative AI spend that compresses if economic conditions deteriorate.
For European business strategy, the $725 billion year is a structural recalibration. The infrastructure that will define the next decade of productivity is being built now, almost entirely outside Europe, by four corporations with capex budgets larger than national governments.
The compute is American. The capital is American. The question is whether the productivity gains will be European at all.
