A document from Tesla CEO Elon Musk vs OpenAI CEO Sam Altman (2026) trial, scheduled to start from April 27, is now available for all. This document consists an email from Musk to Nvidia chief executive Jensen Huang where the SpaceX boss is asking for an early units of the world’s first AI supercomputer – the Nvidia DGX-1 – for OpenAI.The exchange began at midnight on April 13, 2016, with Musk reaching out to Huang for a unit. Musk was careful to explain exactly what OpenAI was at the time: an independent non-profit. At the time, Nvidia was launching the DGX-1 as a revolutionary “AI in a box,” and demand was already skyrocketing. Huang replied six days later, noting that he has not forgotten about his request.

Read full email Elon Musk wrote to Jensen Huang here

From: Elon MuskSent: Wednesday, April 13, 2016 12:01 AMTo: Jensen HuangSubject: DGX-1Can OpenAl buy one of the early units? The team was asking me about this earlier today.OpenAl is unaffiliated with Tesla. It is a non-profit funded by me and a few others with the goal of developing safe AGI (and hopefully not paving the road to hell with good intentions).On Apr 19, 2016, at 10:11 AM,Jensen Huang wrote:I’ve not forgotten.This is the first supercomputer that’s selling itself off the web! Demand coming from all over.First shipment starts end next month. And already taking orders for delivery in September.I will make sure OpenAl gets one of the first ones.

Elon Musk praises Jensen Huang

In a reply to the post, Musk praised Huang and provided context on the partnership between the two tech leaders.“This email is a follow-up to an earlier conversation I had with Jensen asking about obtaining the first DGX. Jensen is great,” Musk replied. Later that year, Huang personally delivered the first DGX-1 to the OpenAI office in San Francisco. A photo of the moment – where Musk and Huang are seen – has famously been shared on social media, specifically on X (formerly Twitter).

Why the DGX-1 mattered for Elon Musk and OpenAI

The DGX-1 was seen as a massive leap forward in computing power, equivalent to 250 conventional servers. It provided the raw “muscle” required for OpenAI to train the early models that eventually evolved into the ChatGPT chatbot technology we see today.