NVIDIA’s power play paves the way for firms to reshore into the machine

NVIDIA’s GPU Technology Conference 2025 (GTC25) marked a turning point in which compute power becomes the foundation of a new wave of digital reshoring for enterprises. It’s time to reshore not just supply chains, but intelligence.

Enterprises under pressure from skills shortages, geopolitical instability, and the rising capabilities of AI are being driven to find ways of shifting critical capabilities back in-house, but now more as software and less as people. This ‘reshoring into the machine’ enables them to reduce their risks and control their supply chains.

Exhibit 1: NVIDIA CEO Jensen Huang makes the case for new power demands reasoning AI will place on every enterprise—making compute a P&L lever

Source: HFS Research, San Jose, 2025

Compute power is no longer an IT decision—it’s a P&L lever

NVIDIA CEO Jensen Huang’s declaration of “power equals future revenues” is a call to action for enterprise leaders to rethink their infrastructure strategy. Compute power is no longer an IT decision; it’s a P&L lever. Especially as agentic AI begins to scale, with tasks such as planning and reasoning requiring 150x the compute of traditional AI, Huang forecasts a staggering ten billion-fold increase in compute demand. This isn’t a niche projection—it’s a macro shift. Imagine every knowledge worker supported by ten digital agents. That’s the scale enterprise leaders may need to plan for.

But the market isn’t blindly following. NVIDIA’s post-event share dip reflected a cautious investor outlook: enterprises aren’t yet convinced the AI returns will match the hype. And they’re right to question. The pivot from AI aspiration to enterprise transformation still lacks consistent, measurable outcomes. You should demand evidence of the delta between outcomes using agentic AI and less inference-intense forms of AI (machine learning for example) before committing to the kind of infrastructure investments Huang has in mind.

Where reshoring gets real: AI meets operations

There were signs of tangible enterprise value at GTC25. Lowe’s showcased the MyLowe app—a GenAI solution built on NVIDIA tech that enhances both customer and employee experience by embedding domain-specific retail knowledge. It’s a clear example of reshoring expertise into digital workflows, not offshoring it to third parties.

This signals a broader pattern HFS is tracking: enterprise buyers are no longer content with abstract AI proofs of concept. They want AI deployed at the core of their value chains—reskilling frontline teams, optimizing decision-making, and restoring control over data.

Pressure grows on GSIs to pivot from labor to tech arbitrage

Citigroup has announced plans to cut its IT contractors by 50%, down to 20% of the total staff. While they are recruiting people, it would be naive to believe they aren’t also scaling up their agentic capabilities.

This new pressure on service providers to pivot toward technology arbitrage comes as global system integrator (GSI) revenues are impacted by the new US administration. Accenture, for example, warned that the Trump administration’s plan to reduce spending on federal contractors is creating ongoing uncertainty and its share price fell 7%.

Service providers begin to reshape their value proposition—as software

Cognizant, Capgemini, Deloitte, and EY all announced deepened NVIDIA partnerships. However, what matters to enterprise leaders is how these translate into outcomes, not alliances.

Cognizant’s GTC examples of digital twins to reshape the factories of tomorrow, through to an opportunity to take out $60 billion in costs in US healthcare, are directionally aligned with our ‘Services-as-Software’ vision—where service providers become enablers of reshored, intelligent operations (see Exhibit 2).

Exhibit 2: Service providers are rushing to agentic—helping enterprises reshore their capabilities into the machine as the next step on the journey to Services-as-Software

Source: HFS Research, 2025

Capgemini. Deloitte, and EY announced new deeper and agentic-focused partnerships with NVIDIA at GTC25—each also aligned with our prediction regarding the shift toward Services-as-Software. And these are just three examples. New service provider-NVIDIA deals are getting inked by the day. For enterprise leaders, the real differentiator won’t be their level of access to NVIDIA tech—look instead at their ability to orchestrate complex AI ecosystems across proprietary data, governance constraints, and real-time enterprise workflows.

Three ways to ramp up your reshoring

1. Redesign digital infrastructure strategies around AI-driven value creation, not just cost optimization.
2. Push service partners to deliver software-like, outcome-based offerings, not just tech integrations.
3. Reclaim operational and data sovereignty by investing in localized, high-performance AI ecosystems.

The Bottom Line: Reshore your core capabilities into AI-native sovereign infrastructure.

Seize this moment to lead a strategic shift: reshoring core capabilities into AI-native infrastructure. HFS sees this as a defining inflection point: enterprises that embed digital intelligence through sovereign, scalable compute will outpace those stuck with outsourced models designed for a pre-AI world.

The machines aren’t coming—they’re already here. The only question is whether your enterprise is ready to bring them home.