AI stocks are rebounding. One analyst says there’s no spending slowdown in sight.
After a stretch of volatility that rattled even the strongest names, AI-linked shares are regaining momentum. The catalysts look familiar: hyperscalers are dialing up capital spending, chipmakers continue to signal demand that exceeds supply, and enterprise buyers—once cautious—are beginning to scale real deployments beyond pilot projects. Against that backdrop, at least one prominent tech analyst argues there’s no spending slowdown in sight, framing the current cycle not as a one-off splurge but as a multi-year infrastructure build that is still in its early innings.
What’s driving the rebound
– Hyperscaler capex visibility. Microsoft, Amazon, and Alphabet have made AI the centerpiece of data center investment. Throughout 2023 and into 2024, management teams guided to sharply higher capital expenditures, with AI infrastructure called out as the key driver. While quarter-to-quarter timing can be lumpy, the direction of travel has been consistent: more racks, more power, more accelerators, and faster networks.
– Unrelenting demand for compute. Training state-of-the-art models and serving inference at scale are both resource-hungry. GPU clusters remain supply-constrained, networking bandwidth is a gating factor, and data pipelines require storage upgrades. Chip suppliers have reported triple-digit year-over-year growth in data center segments through 2023–2024, reinforcing the idea that AI demand is structural, not cyclical.
– Early monetization signals. Generative AI assistants and copilots are moving from proof-of-concept to paid features across productivity suites, developer tools, and customer service. Cloud providers are seeing rising AI attach rates to core compute and storage, while independent software vendors are introducing premium tiers enabled by model inference. Enterprises are not yet declaring victory on return on investment, but adoption is broadening.
The analyst view: “No slowdown” in sight
The most bullish thesis making the rounds is straightforward: we are in the midst of a once-in-a-generation buildout of digital infrastructure to support AI, and the spending curve extends multiple years. Proponents point to:
– A greenfield compute layer for both training and inference, with a roadmap that spans several hardware cycles.
– A networking and memory arms race to remove bottlenecks created by larger models and higher context windows.
– A second-order boom in power, cooling, and real estate as data centers densify and expand.
In this view, any pullbacks in AI stocks tend to reflect positioning and valuation digestion rather than a fundamental rollover in demand. The “no slowdown” camp argues that, even if individual quarters fluctuate, the multi-year capex commitments already telegraphed by hyperscalers set a high floor under the ecosystem.
Where the money is going
– Semiconductors and accelerators. GPU leaders remain central, but custom silicon from cloud providers and alternative accelerators are broadening the market. AI-specific networking chips, high-bandwidth memory, and optical interconnects are critical beneficiaries.
– Cloud platforms. Higher AI consumption lifts core cloud metrics. Platform differentiation increasingly hinges on model access, tooling, vector databases, and integrations, which can support premium pricing and stickier workloads.
– Networking and storage. Spine-leaf architectures are evolving to accommodate massive east-west traffic for distributed training and low-latency inference. Lossless Ethernet, InfiniBand, and faster optical links are all in focus.
– Data center power and cooling. The physical footprint and energy profile of AI clusters are reshaping site selection and vendor lists. Liquid cooling, upgraded switchgear, and grid interconnects are seeing accelerated orders. Data center landlords with power-ready capacity are advantaged.
– Software and services. MLOps, security, observability, and data governance are moving from niche to necessary. As models embed into workflows, spending on integration and compliance rises alongside compute budgets.
Why “slowing” hasn’t shown up yet
– Adoption is diffusing from a handful of frontier model trainers to a long tail of enterprises deploying inference at scale. That broad base can offset pauses by any single customer.
– Model progress keeps resetting the performance bar, justifying fresh hardware as each generation promises better tokens-per-watt and total cost of ownership.
– Competitive dynamics are unforgiving. Cloud providers and software platforms see AI as an existential capability; falling behind is more expensive than overspending.
Risks that could interrupt the narrative
– Power constraints and supply-chain friction. Even with strong demand, bottlenecks in power availability, high-bandwidth memory, or advanced packaging can push out deployments.
– ROI scrutiny and budget cycles. If AI features don’t translate into measurable productivity or revenue, CFOs could stretch timelines, especially outside the tech sector.
– Regulation and geopolitics. Export controls, data sovereignty, and changing privacy rules can reshape go-to-market plans and addressable markets.
– Pricing pressure. As model access becomes more commoditized, margins could compress for providers without strong differentiation in data, distribution, or vertical expertise.
– Valuation sensitivity. Elevated multiples leave little room for execution slips; even strong prints can be met with profit-taking.
What to watch next
– Hyperscaler capex commentary. The tone and magnitude of forward capex plans remain the single best barometer for AI infrastructure demand.
– Cost-per-token and utilization trends. Improvements in model efficiency, inference batching, and caching inform whether spend scales linearly with usage or benefits from operating leverage.
– Power and real estate signings. Leases and grid interconnect deals can foreshadow where capacity—and revenue—will land six to eighteen months out.
– Enterprise case studies. Evidence of durable productivity gains or revenue lift will influence budget allocations for 2025 and beyond.
Bottom line
AI stocks are rebounding as the core thesis—durable, multi-year investment in compute, networking, and data center infrastructure—continues to hold. The most optimistic analysts see little sign of a spending slowdown, arguing that competitive imperatives and a still-young platform cycle will keep capital flowing. Volatility and periodic digestion are likely, but the structural drivers behind AI capex remain intact. For investors and operators alike, the center of gravity is simple: follow the capex and the power.
This article is for informational purposes only and is not investment advice.
