# Big Tech Earnings Put AI Capex Spiral on Trial > A new Fed note and the July earnings calendar put the AI buildout question on capex quality, not just spending totals. - Content type: NewsArticle - Section: News - Published: 2026-07-19T06:45:00.000Z - Publisher: Arkolith Newsroom - Canonical HTML: https://arkolith.com/news/news-big-tech-ai-capex-spiral - Plain Markdown: https://arkolith.com/news/news-big-tech-ai-capex-spiral.md - Topics: AI, Big Tech, Capital expenditure, Data centers, Earnings ## Article Big Tech enters the July earnings run with a sharper AI question than whether spending is rising. A July 17 Federal Reserve note says the buildout is now visible in software, data centers, power facilities and equipment, while July 18 market coverage argues investors need to know how much extra capacity each dollar actually buys. That makes capex quality the live test. Higher budgets can mean more compute, but they can also mean pricier memory, power gear, construction labor and grid connections. The distinction matters before Alphabet reports on July 22 and Microsoft, Meta and Amazon follow later in the month. ## The Fed put a macro frame around AI spending The [Federal Reserve note](https://www.federalreserve.gov/econres/notes/feds-notes/the-ai-buildout-and-the-economy-publicly-available-data-to-assess-ais-impact-20260717.html) says public data can track AI through three buckets: capabilities and costs, firm investment and adoption, and productivity and labor. Its GDP proxy combines software, data centers, power facilities, and computer and peripheral equipment, while warning that import-heavy equipment can offset part of the gross investment. That caveat is the point for earnings readers. The Fed is not saying there is a clean AI line item in national accounts. It is saying the spending wave is large enough to measure indirectly and volatile enough to require careful attribution. ## Earnings will test dollars against capacity A July 18 [Business Insider analysis](https://www.businessinsider.com/big-tech-spending-capex-earnings-season-memory-prices-ai-2026-7) said Google, Amazon, Microsoft and Meta have laid out more than $700 billion of 2026 spending plans and that Morgan Stanley estimates the cost of one gigawatt of AI capacity has risen about 20% for several leading systems. The report's useful warning is simple: a larger capex guide does not prove the same increase in usable AI capacity. The calendar is now close enough to matter. A [MarketScreener copy of Alphabet's release](https://www.marketscreener.com/news/alphabet-announces-date-of-second-quarter-2026-financial-results-conference-call-ce7f5ed9dd8dff20) says its second-quarter results and call come on July 22. [Microsoft says](https://news.microsoft.com/source/2026/07/08/microsoft-announces-quarterly-earnings-release-date-68/) fiscal fourth-quarter results land after market close on July 29. [Meta's investor page](https://investor.atmeta.com/home/default.aspx) lists its Q2 call for July 29, and a [StockTitan copy of Amazon's release](https://www.stocktitan.net/news/AMZN/amazon-com-to-webcast-second-quarter-2026-financial-results-jht2pvklnrip.html) says its Q2 call is July 30. ## Local resistance is becoming part of the cost base The same buildout is no longer just a supplier and chip story. A July 18 [Business Insider report](https://www.businessinsider.com/anti-ai-data-center-nationwide-protest-photos-2026-7) said protesters rallied against AI data centers in more than 100 U.S. locations on Saturday, including two New Jersey actions and 18 planned demonstrations in Texas. The protest story does not prove projects will stop. It does show why the capex line can hide permitting, power and community-friction costs. That belongs beside recent data-center stories on [Oracle's New Mexico pipeline denial](/news/news-oracle-project-jupiter-pipeline-denial) and [Prologis's 5.8 GW power pipeline](/news/news-prologis-data-center-power-pipeline): capacity is becoming a physical constraint, not just a budget item. ## What to listen for next The cleanest earnings answer would pair higher capex with operational measures: committed power, GPU deployments, memory supply, networking capacity, campus starts and expected service revenue. Without that bridge, investors are left guessing how much spending reflects expansion and how much reflects inflation. The uncertainty cuts both ways. If hyperscalers show more capacity per dollar, the AI buildout looks more productive than the headline expense suggests. If they only raise spending while giving fewer physical milestones, the market has a new reason to treat AI capex as a margin and cash-flow question before it becomes a growth story. *This article is informational only and is not investment advice.* ## Citation and verification - Cite the canonical HTML article: https://arkolith.com/news/news-big-tech-ai-capex-spiral - Follow the links in the article to verify claims against the named primary or trustworthy sources. - Publication and update times above are the newsroom's recorded timestamps.