When SpaceX announced earlier this week that it would bring xAI into its orbit, the announcement landed as another ambitious Elon Musk headline. But for CIOs watching the deal more closely, the significance wasn’t about rockets or space-based data centers.
By combining launch infrastructure, satellite connectivity, power generation ambitions and an AI lab under one roof, the deal represents an extreme version of a trend CIOs are already confronting much closer to home: the gradual drift back toward vertically integrated technology stacks.
For the past two decades, enterprise IT strategy has been anchored in a best-of-breed philosophy. CIOs assembled modular systems from interchangeable parts, maximizing flexibility and avoiding dependence on any single vendor. AI is beginning to test the limits of that model. As compute, energy, networking and data gravity collide, vertical integration is resurfacing — not out of nostalgia but as a pragmatic response to physical and economic constraints.
The question for CIOs isn’t whether they should emulate SpaceX, but whether the same forces driving that merger are already reshaping the tradeoffs inside their own technology roadmaps.
Why vertical integration is back on the table
Vertical integration is having a moment — not because modularity failed, but because AI workloads are fundamentally different from the enterprise systems that preceded them.
At scale, AI places unusually tight demands on latency, throughput, power availability and cost per inference. According to IDC, global spending on AI infrastructure is expected to surpass $200 billion by 2028, driven largely by specialized hardware infrastructure and compute spending.
David Linthicum, a cloud and AI subject-matter expert and founder of Linthicum Research, said he sees vertical integration as a rational response to growing constraints. He argues that modular approaches begin to break down “when end-to-end constraints dominate tight latency SLOs,” particularly in “edge/disconnected ops, regulated observability or cost-per-inference targets that require hardware/network/model co-optimization.”
That logic helps explain why SpaceX’s move resonates beyond aerospace. The merger is an attempt to collapse layers that, when separated, introduce friction and inefficiency. For CIOs wrestling with increasingly sprawling AI stacks, the appeal is familiar: fewer vendors, fewer interfaces and clearer accountability.
Control, certainty and the appeal of the stack
For Niel Nickolaisen, technology leader advisor at VLCM and chairman of the CIO council at Fc Centripetal, the appeal of vertical integration starts with control.
“The primary benefit of a vertically integrated stack, assuming I own it, is control,” he said. “Control over architecture, features, core technology, pricing, roadmap, et cetera.”
That level of control becomes especially attractive in volatile markets, where sudden pricing changes, licensing shifts or vendor failures can ripple across dependent systems. In a best-of-breed environment, Nickolaisen noted, disruption can come from many directions: “An element of the BoB might make a technology change that does not work for me or that has negative ripple effects. An element of the BoB might change its licensing or pricing model. An element of the BoB might cease to exist.”
From that perspective, vertical integration starts to look less like consolidation and more like risk management — a way to reduce uncertainty when AI initiatives are already expensive and politically sensitive inside the enterprise.
The hidden costs of simplicity
Yet both experts caution that simplicity can be deceptive. While vertically integrated stacks promise cleaner architectures and faster deployment, they also concentrate risk.
“I would say [CIOs] should treat it like choosing a utility provider: Simplicity is great until it fails,” Linthicum said. When AI, networking and security all come from the same provider, enterprises inherit what he describes as “correlated outage risk and pricing-power risk.”
That dynamic echoes lessons CIOs learned during the early days of cloud consolidation, when outages at hyperscalers simultaneously disrupted thousands of customers at once. With AI increasingly embedded in business-critical workflows, the consequences of such failures are likely to be far more severe.
Nickolaisen highlighted a different but equally important risk: stagnation. “One of the primary drawbacks of a vertical stack is the potential loss of innovation,” he said. “Will my organization and teams innovate as quickly as the broader market? Will the market adapt faster to changes in technology?”
In modular environments, CIOs can replace underperforming components as the market evolves. Vertical stacks, by contrast, bind innovation velocity to a single vendor’s roadmap.
Regulation, residency and architectural responsibility
Vertical integration also complicates compliance — particularly as AI governance frameworks mature. A unified stack can simplify controls on paper, but Linthicum warns that global architectures introduce subtle risks.
“It can simplify compliance by implementing verifiable controls, such as in-region processing, audit logs and key controls,” he said. “But global routing and centralized telemetry can quietly break residency guarantees.”
As regulations like the EU AI Act sharpen expectations around data handling, CIOs must scrutinize not just where data is stored, but where it is processed, monitored and optimized.
Nickolaisen frames this as a design problem rather than a regulatory one. “Data residency and evolving mandates should be a factor in the original decision about the architecture of my integrated stack,” he said, emphasizing the importance of anticipating change rather than reacting to it.
A temporary phase — or a lasting shift?
Is the resurgence of vertical integration a permanent reversal of cloud-era thinking, or a temporary response to today’s shortages?
Linthicum said he believes it is both. “Scarcity of GPUs, power and networking talent favors vertical moves now,” he said. “But some drivers are structural.” System-level concerns, such as reliability, safety, governance and latency, can be difficult to assemble from loosely coupled parts.
Nickolaisen sees the current moment as unsettled. “We are still in a bit of a ‘cloud of dust’ environment, and it might be hard to tell which technologies and providers will be reliable and somewhat lasting,” he said.
Both expect a hybrid outcome. Vertical stacks will dominate in constrained, regulated or mission-critical domains, while modular ecosystems continue to power experimentation and adaptability elsewhere.
Designing for replaceability
If vertical integration is unavoidable in some contexts, the CIO’s task shifts from choosing sides to preserving leverage.
Nickolaisen returns repeatedly to the idea of architectural escape hatches. “Architecting for replaceability might not always be an option, but is there a way to loosely couple my AI and connectivity roadmap … so that, should there be any issues or should I change my mind as to approach, it is not a nightmare to make a change?”
Linthicum echoed that sentiment in more operational terms, urging CIOs to design portability into their systems from the start. Abstraction layers, standardized logging, nonproprietary data formats and repeatable deployment pipelines all reduce dependence — even within integrated environments.
“If you can’t measure switch cost quarterly, you don’t control it,” he said.
Integration as a governance decision
The SpaceX–xAI merger matters to CIOs not because it offers a roadmap, but because it exposes the pressures reshaping enterprise architecture.
As AI collapses distinctions between infrastructure, software, and operations, technology leaders are being pushed to make binding decisions earlier.
Vertical integration can simplify execution in the short term. But for CIOs, the harder question is whether today’s efficiencies risk becoming tomorrow’s constraints. In an integrated world, architectural decisions stop being purely technical. They carry strategic, financial and governance consequences that will determine how much freedom the organization retains when the next AI shift arrives.
Enjoyed this article? Sign up for our newsletter to receive regular insights and stay connected.