Foresight

NVDA

Technology
Semiconductors
Illustrative prices

NVIDIA Corporation

Dominant GPU and accelerated-computing platform for AI training/inference, gaming, and auto.

$115.08

-0.19 (-0.16%) session

Demo quote — not a live market feed

Market cap

$2.9T

P/E

48.5

Div yield

0.03%

Volume

7.7M

Price history

+8.49% over 1Y

Illustrative demo series

At a glance

Next earningsAug 27, 2026
Fiscal periodQ2 FY2027
R&D spend~$10B+
R&D trendRising
Horizon score82

Primary R&D focus

Next-gen GPU architectures
Full-stack systems
CUDA
82

Horizon score

Strong forward setup

Unmatched platform position, but valuation and concentration risk demand perfect execution.

Forward brief

What matters next — not a recap of last quarter

NVIDIA’s forward question is not ‘did they beat last quarter’ — they usually do — it is whether Data Center demand compounds through the next architecture cycle while software/systems deepen the moat. R&D is almost entirely core platform defense. Next earnings: sequential Data Center, gross margin, and any China/export language. Those lines decide if the future is still accelerating or entering a digestion phase.

AI infra
GPU
Data Center
CUDA

Next earnings

Q2 FY2027 · Aug 27, 2026

8x recent beats

Expected EPS

$0.92

Expected revenue

$44.8B

Report date

Aug 27, 2026

Market prices perfection; any data-center sequential miss or China export update will reprice the multiple.

What to watch on the call

  • Data Center revenue sequential growth and backlog commentary
  • Blackwell / next architecture ramp vs supply constraints
  • Gross margin — is networking mix holding margins elevated?
  • China export compliance and alternative SKU contribution

Questions that actually matter

  • Is customer capex still accelerating or are hyperscalers digesting prior buys?
  • How durable is CUDA moat if custom ASICs (Google, Amazon, Microsoft) take share of inference?
  • When does networking (NVLink/Spectrum) become a first-class earnings story?

R&D deep dive

Not just “they invest in R&D” — in what, for when, and is it core?

Annual R&D / tech spend

~$10B+

Spend trend

Rising

% of revenue

~10%

R&D is laser-focused on staying the default AI compute platform — silicon, systems, and CUDA software. Automotive/Omniverse is adjacent diversification. Almost none of the spend is ‘science for science’s sake’; it is defense of an extraordinary franchise under competitive and export pressure.

Next-gen GPU architectures (Blackwell successors)

Core business
Near-term (≤18 mo)

Higher FLOPS/watt, larger HBM packages, improved transformer engines for training and inference.

Strategic bet: Stay 12–18 months ahead of AMD and custom silicon on training economics.

Full-stack systems (DGX, GB200 NVL racks, networking)

Core business
Near-term (≤18 mo)

Selling entire AI factories — GPUs + NVLink + Spectrum-X + software — not discrete chips only.

Strategic bet: Raise switching costs; make ‘NVIDIA rack’ the unit of purchase.

CUDA / software & inference frameworks

Core business
Mid-term (1–4 yr)

CUDA ecosystem, TensorRT, NIM microservices, enterprise AI software layers.

Strategic bet: Software stickiness is the real moat if silicon competitors close the gap.

Automotive / robotics / Omniverse simulation

Adjacent
Long-term (4+ yr)

Drive platforms, Isaac robotics, industrial digital twins.

Strategic bet: Diversify beyond hyperscaler GPU cycles into physical AI.

Strategic bets

Where management is placing multi-year chips

AI factory as product

Medium risk

Shift from chip vendor to systems vendor so customers buy multi-year platform roadmaps.

Horizon · 1–3 years

Software monetization of CUDA install base

Medium risk

Increase software attach so margins and stickiness survive silicon competition.

Horizon · 2–4 years

Upcoming catalysts

Aug 27, 2026

Q2 earnings

Data Center print + next-arch supply color

Next GPU architecture keynotes / sample timelines

Multiples re-rate on roadmap confidence

Forward risks

Customer concentration & capex digestion

High

A handful of hyperscalers drive the majority of Data Center; pauses show up immediately.

Export controls

High

China SKU restrictions can erase a material revenue stream with little warning.

Custom ASIC competition

Medium

Inference workload migration to in-house chips could slow GPU TAM growth longer-term.