Hyperscaler CapEx and the AI Bull Market: Chip Stock Investor Explains

According to Chip Stock Investor (CSI), the latest Q2 2026 reports from the major hyperscalers show that the core driver of the current semiconductor-driven bull market is still intact: massive capital expenditures on AI data centers.

The host walks through fresh CapEx and cloud-revenue data and argues that recent sharp pullbacks in semiconductor stocks look more like a mid‑cycle shakeout than the end of the AI bull market. Despite some big drawdowns in recent weeks, CSI notes that many semiconductor names still show very strong year‑to‑date gains.

CSI's central message is that the scale and trajectory of hyperscaler CapEx, combined with long cloud-compute backlogs, continue to support a multi‑year build‑out of AI infrastructure. The channel frames hyperscalers as a kind of "CapEx hedge" in portfolios that are otherwise heavy in semiconductors.

CSI's core thesis is that hyperscaler and related CapEx for AI data centers remains the key engine behind the AI and semiconductor rally, and that this engine is still running hard as of early August 2026.

On a trailing 12‑month basis, the host cites roughly $580 billion of CapEx across Amazon, Microsoft, Google, Meta, Oracle, Tesla, and SpaceX, most of it tied to AI data-center build‑out. For full‑year 2026, CSI says expectations now sit around $890 billion, which the host rounds to "about $900 billion" in CapEx.

CSI believes this CapEx wave heavily benefits semiconductor makers, especially memory and other AI‑data‑center components, and contends that the recent market turbulence reflects leverage unwinding rather than a fundamental demand collapse. The channel projects that total hyperscaler CapEx could potentially reach around $1.5 trillion in 2027 before eventually moderating later in the decade.

To support the bullish take on AI infrastructure, CSI leans on several data points from hyperscaler reports and third‑party research.

First, the host highlights trailing 12‑month cloud revenue (excluding Meta and Tesla) for the big platforms plus Oracle at nearly $400 billion. Within that, Amazon Web Services is described as being above $148 billion in trailing 12‑month revenue and "fast approaching" a $200 billion annualized run‑rate. Microsoft Azure and the broader intelligent cloud business are pegged at $138 billion, with Google Cloud at nearly $78 billion, and Oracle's total cloud (infrastructure plus software) at $34 billion.

CSI notes that the hyperscalers report "significant backlogs" expected to take years to work through. The host points to commentary from Amazon indicating AWS alone believes it can at least double revenue over time and might eventually reach $1 trillion in annualized revenue. The channel also cites Gartner estimates: earlier in 2026, Gartner projected about $790 billion in data‑center systems spending (a 56% year‑over‑year increase), later revised to over $820 billion (about 63% growth) by late July.

Crucially, CSI notes Gartner now breaks out "infrastructure as a service" separately at $287 billion in expected 2026 end‑market spending, up 29% year over year, while the broader IT services line was reduced to under $1.6 trillion with only about 5.3% growth.

CSI repeatedly stresses that the current pace of hyperscaler CapEx is not sustainable and represents an "all‑out" growth cycle rather than a new normal. The channel points to free‑cash‑flow pressure as a near‑term risk: some hyperscalers have slipped into negative quarterly free cash flow, and others show restricted or declining free cash flow because of the heavy investment.

The host argues that CapEx‑to‑revenue ratios clearly illustrate the lack of long‑term sustainability, and expects these ratios to moderate "later this decade" as companies pull back on CapEx to restore stronger profitability and free‑cash‑flow growth. Oracle is singled out as an outlier, with trailing‑12‑month CapEx said to be consuming nearly all of its revenue, leading CSI to describe it as a kind of "neo cloud" player at this stage.

CSI also underscores that profitability metrics are "messy" during this build‑out. Free cash flow is distorted by current CapEx, while EBITDA excludes depreciation on existing infrastructure, and GAAP net income is pressured by very large depreciation charges spread over roughly five years or more. The channel says this leaves investors with genuine uncertainty about how profitable hyperscalers can be once this CapEx boom fully flows through the financials.

Looking ahead, CSI recommends watching several forward‑looking signals to gauge whether the AI‑driven bull market remains intact.

First, the host points to ongoing increases in CapEx guidance, such as Amazon's 2026 CapEx plan rising from $200 billion to $220 billion, partly attributed to higher memory prices. CSI interprets these revisions as evidence that hyperscalers still see strong demand and that more of their revenue will flow to semiconductor suppliers.

Second, the channel highlights Gartner's structural change in IT spending categories: "infrastructure as a service" is now broken out and growing rapidly, while traditional IT services slow. CSI argues this supports the view that legacy, on‑premises IT is being disrupted by AI‑centric cloud infrastructure, with many industries newly incentivized to move workloads to centralized, AI‑enabled data centers.

Finally, CSI suggests investors monitor how quickly AI opens new workloads across sectors and how hyperscaler CapEx‑to‑revenue ratios evolve. The host expects the trend toward AI infrastructure to last "at least a few more years," but warns that market action will likely remain bumpy, as seen in late July and expected into August 2026.