Rivian Doesn't Need Nvidia for Self-Driving Cars. Should Nvidia Investors Be Worried?

Rivian said it will ship an in-house 1,600 sparse TOPS inference chip in R2 vehicles from late 2026 to accelerate its push toward Level 4 autonomy, a move that underscores a broader trend of OEMs and cloud providers (Alphabet’s TPUs, AWS Trainium) developing custom AI silicon to reduce reliance on Nvidia. Nvidia remains the dominant supplier—fiscal Q3 revenue was $57.0 billion with $51.2 billion from data center products and Blackwell demand described as “off the charts”—but its automotive business is small (~$592 million) and Rivian itself is modest (~$1.6 billion revenue). Given Nvidia’s rich valuation (roughly 44x P/E), even incremental shifts toward in-house or alternative chips could pressure pricing power and margin assumptions, making the build-versus-buy dynamic a tangible strategic risk for investors to monitor.

Analysis

Rivian disclosed an in-house 1,600 sparse TOPS inference chip that it plans to ship in R2 vehicles starting late 2026, positioning the company to accelerate toward Level 4 autonomy and reduce reliance on third-party AI accelerators. CEO RJ Scaringe framed the chip as a core enabler of self-driving capability, signaling an OEM push to internalize inference compute.

Nvidia remains the dominant supplier in AI compute: fiscal Q3 revenue was $57.0 billion with data center revenue of $51.2 billion (up 66% YoY), management saying Blackwell demand was "off the charts" and cloud GPUs are sold out, while its automotive and robotics segment was $592 million. By contrast Rivian's revenue is roughly $1.6 billion, underscoring that Rivian itself is immaterial to Nvidia's topline today even as it exemplifies a broader trend.

Alphabet's TPUs and AWS Trainium show the same build-versus-buy logic, creating credible substitutes that could incrementally erode Nvidia's pricing power and margin profile over time; with Nvidia trading at roughly a 44x P/E, even small shifts in customer behavior could pressure expectations. Investors should therefore treat the proliferation of in-house and cloud-provider AI silicon as a strategic risk to monitor rather than an immediate revenue shock to Nvidia, watching procurement patterns, margin trends and autonomous-vehicle execution closely.

Key Decisions for Investors

• Existing Nvidia holders should consider trimming size or implementing downside protection given a 44x P/E and rising credible substitutes, and monitor data-center revenue growth, gross margins, Blackwell demand updates and cloud GPU supply as key signals
• Prospective Nvidia buyers should wait for clear evidence that in-house chips or TPU/Trainium deployments are materially changing procurement or margin guidance before committing large incremental capital
• Monitor Rivian's chip execution and metrics—production timing, inference performance, and realized cost savings—and consider increasing exposure only if the hardware meaningfully advances autonomy and lowers per-vehicle compute expense
• Across the portfolio, track announcements and customer adoption from Alphabet and AWS and OEMs for signs of reduced vendor concentration in AI compute, and reassess valuation and hedging strategies for suppliers reliant on Nvidia-driven pricing power