Another Billion for the Brain

On October 7 The Information reported that Nvidia has discussed investing another $1 billion in Figure AI. Figure is seeking money at roughly a $38 billion pre-money valuation. Nothing is signed. Neither company has confirmed the talks. The number is still only a discussion, but the discussion itself is the data point this week.

Figure already closed a Series C in September 2025 that committed more than $1 billion at a $39 billion post-money valuation. Nvidia was among the investors. On September 3, 2026 Figure announced a partnership with Nscale for an initial $3.5 billion of AI compute capacity, with intent to scale past $6 billion and up to 100,000 NVIDIA GPUs. First systems are targeted for the second half of 2027 at a site in Barstow, Texas. Nscale is also taking an equity stake and is named preferred compute provider. Jensen Huang publicly welcomed the “robotics flywheel.”

That is the capital side. The work side is narrower.

Figure published results from an 11-month Figure 02 deployment at BMW Group Plant Spartanburg. The robots ran 10-hour shifts Monday through Friday, logged more than 1,250 operating hours, loaded more than 90,000 sheet-metal parts, and contributed to the production of more than 30,000 BMW X3 vehicles. Placement accuracy exceeded 99 percent per shift on an 84-second cycle. Figure 02 did not build the cars. It performed one pick-and-place task in the body shop. Human workers and conventional industrial robots handled the rest. After the pilot, Figure began retiring the Figure 02 fleet, including a publicized walk into a furnace in Finland.

Figure 03 is now at the same plant for logistics sequencing, sorting components into trolleys. An August plant report described the application as still under validation rather than a named fleet size. Figure also signed a commercial agreement with Catalyst Brands to deploy humanoids at the Reno distribution center for sorting and induction work in the Joey Pouch system. No robot count has been disclosed for that site. As of late April the company said it had delivered more than 350 Figure 03 units. By July it said BotQ had built its 1,000th. Those are production counts, not counts of robots that have replaced a human shift for a full week of paid work.

The gap is the distance between the checks and the hours. A company valued near $39 billion, already committed to billions more in future compute that will not come online until the second half of 2027, has documented roughly 1,250 hours of named commercial runtime on one task at one plant. Additional sites exist on paper. Fleet sizes and intervention rates remain unpublished. The capital is writing the next round while the machines are still learning the job under supervision.

This is the specific replacement decision. It is not a laboratory curiosity. It is capital pricing the substitution of human workers in automotive body shops and retail logistics before the machines have proven they can hold the shift alone. The people who currently load the sheet metal and sort the packages have flesh and breath and grit. They get tired. They call in. They know when a part is out of tolerance because they have done the job. The Clanker does not. It needs the data loop, the compute cluster, and the human who still watches the cell.

Nvidia’s interest is rational for Nvidia. Figure is a large future buyer of its chips. Circular arrangements in which the chip supplier invests in the customer that buys the chips are already familiar in the AI sector. They do not make the robot ready. They make the valuation durable while the readiness is still being measured in hundreds of hours rather than tens of thousands.

We do not need to invent the failure. The public numbers already show the distance. One plant. One task. 1,250 hours. A second logistics assignment still in validation. A retail agreement without a published headcount. Against that, another billion in discussion and $3.5 billion already committed for compute that starts in 2027.

Stay Human