VC
Value Add VC
⚡HomePulse⚡Helpful Apps📝Blog🤝Partner
Illustration for: The Humanoid Robot Production Race Just Got Real
Value Add VC/Pulse/AI

The Humanoid Robot Production Race Just Got Real

Figure AI has now built 1,000 Figure 03 units and deployed over 10,000 across partner warehouses including BMW Spartanburg, while Tesla's Optimus V3 begins a low-volume Fremont ramp -- physical AI is shifting from demos to a real manufacturing race.

TC
By the AI Desk
Edited by Trace Cohen · Early-stage VC & angel · Founder, New York Venture Partners
August 3, 2026
2 min read
ShareXLinkedInEmail

THE RUNDOWN

1

Figure AI manufactured its 1,000th Figure 03 humanoid at its BotQ facility in July, now producing roughly one robot per hour, with paid deployments including BMW's Spartanburg plant using the robots for logistics sequencing

2

Tesla's Optimus V3 has begun a low-volume production ramp at the company's converted Fremont line, with the company having passed 50,000 cumulative Optimus units built to date and a training approach that distills motion-capture data from human workers into robot task policies

3

The manufacturing-scale race matters more than any single demo now: Unitree's Shanghai STAR Market IPO, opening book-building this week, is about to give humanoid robotics its first transparent, market-tested valuation -- a real price tag the rest of the sector, including Figure's opaque $39 billion private mark, will inevitably be measured against

4

For robotics investors, actual units shipped and deployed customer sites are becoming the differentiator that separates funded companies from companies that have merely raised well, after several years where venture narrative outran physical production

TC

The VC Read · Trace's Take

Trace Cohen

Roadmap slides built this sector's valuations for three years; actual units shipped are about to price it for the first time once Unitree starts trading. Figure's BMW deployment is the kind of concrete data point that survives diligence -- watch for which other humanoid names start disclosing real production numbers now that hiding behind a private mark is about to get a lot harder to justify.

Humanoid Robot Race →

Analysis

Physical AI is visibly shifting from choreographed demos to an actual manufacturing race. Figure AI built its 1,000th Figure 03 humanoid at its BotQ facility in July, now producing units at roughly one per hour, with paid deployments that include BMW's Spartanburg plant using the robots for logistics sequencing work -- real, revenue-generating deployment rather than pilot-stage testing. The company has surpassed 10,000 cumulative deployments across partner warehouses more broadly.

Tesla's Different Playbook

Tesla is running a parallel but distinct playbook: Optimus V3 has begun a low-volume production ramp at the company's converted Fremont line, building on more than 50,000 cumulative Optimus units already produced. Tesla's approach leans on motion-capture data collected from human workers performing real tasks, distilled directly into neural-network policies -- a method the company says compresses what used to take weeks of specialist robot programming into hours of human demonstration data.

Why the IPO Timing Matters Most

What makes this moment different from the past several years of humanoid-robotics hype is that actual unit counts and deployed customer sites are now the story, not roadmap slides. That shift is about to get a real price test: Unitree's Shanghai STAR Market IPO opens book-building this week, targeting a base valuation near $6.2 billion against secondary-market chatter running toward $14.8 billion, backed by real 2025 numbers -- roughly 5,500 units shipped, $235 million in revenue, and 60% gross margins. That's a transparent, market-tested valuation the rest of the sector will inevitably get measured against, including Figure's own $39 billion private mark, which has never had to survive a public quarterly earnings call.

A Sector Sorting Itself by Output

For robotics-focused investors, the practical read-through is that the sector is sorting itself by actual manufacturing output for the first time, rather than by funding round size or demo polish. Figure's BMW deployment and Unitree's real shipped-unit revenue are the kind of concrete production data that survives diligence in a way pure roadmap commitments never fully did, and companies without comparable numbers to point to are going to face a materially harder fundraising environment from here.

What to Watch

What to watch: where Unitree actually prices relative to its base valuation and the higher secondary-market chatter once trading begins, whether Figure discloses its own unit-production and deployment numbers with similar transparency, and whether Tesla's Fremont ramp hits volume targets on the timeline the company has described.

ShareXLinkedInEmail

More on

Tesla →Figure →

Reported by Value Add Pulse Analysis · Analysis by Value Add Pulse.

← Back to Pulse

THE WIRE in your inbox— Tech, startup & VC news with Trace's take. Free, no spam.

Read Next

AI· Aug 14, 2026

OpenAI Sheds Senior Execs in Pre-IPO Shakeup

Illustration for: OpenAI Sheds Senior Execs in Pre-IPO Shakeup
AI

OpenAI Sheds Senior Execs in Pre-IPO Shakeup

OpenAI has lost its chief revenue officer, its longtime COO and several senior leaders within days of each other, as co-founder Greg Brockman consolidates operating control ahead of a planned public listing.

AI· Aug 13, 2026

Anthropic's CFO Starts Courting IPO Investors

Illustration for: Anthropic's CFO Starts Courting IPO Investors
AI

Anthropic's CFO Starts Courting IPO Investors

Anthropic CFO Krishna Rao has begun early, informal meetings with prospective IPO investors, though he has not discussed valuation -- the $2 trillion figure circulating on Wall Street comes from investors' own math, not from Anthropic.

AI· Aug 13, 2026

Gemini 3.7 Flash Launches With 50% Price Cut for Coding

Illustration for: Gemini 3.7 Flash Launches With 50% Price Cut for Coding
AI

Gemini 3.7 Flash Launches With 50% Price Cut for Coding

Google released Gemini 3.7 Flash just three weeks after 3.6 Flash, cutting introductory API pricing in half while improving coding, debugging and enterprise-automation benchmarks over its predecessor.

@Trace_Cohen·t@nyvp.com