NVIDIA reported $2.38 in earnings per share on $104 billion in revenue, and simultaneously disclosed its backing of inferencing-chip startup d-Matrix — figures that land in the middle of what industry trackers put at roughly $1.5 trillion in cumulative AI infrastructure funding commitments across the sector.

The headline numbers frame the scale. At $104 billion in revenue, NVIDIA's top line now rivals the annual sales of the largest diversified semiconductor suppliers combined. The $2.38 EPS figure, reported by TradingView in its key-facts digest of the company's results, reflects a business still generating exceptional per-share profitability even as capital spending across the AI supply chain accelerates.

What does the d-Matrix investment signal?

NVIDIA's backing of d-Matrix matters because of where the startup sits in the stack. d-Matrix builds in-memory compute silicon aimed at AI inferencing — the phase of AI workloads that runs trained models, rather than training them. That is the segment most analysts expect to grow fastest as AI deployments shift from experimentation to production.

When the dominant supplier of AI accelerators invests in a company that attacks inferencing from a different architectural angle, two readings are possible. Either NVIDIA is hedging against a compute paradigm shift away from the GPU-centric model, or it is consolidating influence over the startup ecosystem before competitors — AMD, Intel, or the growing roster of custom-silicon programs at hyperscalers — can capture it. Both readings point in the same commercial direction: inferencing silicon is now contested ground.

How big is the $1.5 trillion funding figure?

The $1.5 trillion figure cited alongside NVIDIA's results represents the aggregate scale of AI-related funding commitments now circulating through the industry. For context, that amount exceeds the combined historical capex of the entire leading-edge foundry sector.

NVIDIA sits at the center of that flow. Its data-center products are the primary compute engines for the largest share of AI training deployments, and every dollar committed to AI infrastructure ultimately pulls through accelerator, networking, and memory demand. The $104 billion revenue figure is the clearest confirmation that this pull-through has already materialized in reported results, not just in announcements.

What holds the numbers together?

Three elements connect in this story:

• Reported results: $2.38 EPS and $104 billion in revenue, per the TradingView key-facts summary.
• Strategic capital: NVIDIA's backing of d-Matrix, a startup targeting in-memory compute for AI inferencing.
• Sector context: approximately $1.5 trillion in AI-related funding commitments tracked across the industry.

The inferencing angle deserves particular attention. Training capacity — the huge GPU clusters built over the past three years — increasingly looks like the first wave. Inferencing at scale, with lower power budgets and cost-per-token economics driving purchasing decisions, is where the next competitive round will be fought. d-Matrix's architecture is one of several bets that inference workloads will favor different silicon than training did.

What comes next?

NVIDIA's challenge is now execution at scale: sustaining margins on $104 billion of revenue while funding an ecosystem of startups whose technologies could, in time, erode parts of its own franchise. The d-Matrix investment suggests the company prefers to own that option rather than watch it. Whether inferencing silicon diversifies the AI compute market or consolidates further under NVIDIA's umbrella will be one of the defining competitive questions of the next capital-spending cycle.