Self-learning factories
1. Key Themes
Physical AI Is Entering a Hypergrowth Investment Phase
VC funding for physical AI / warehouse robotics has exploded, with H1 2026 already nearly matching all of 2025 in deal count and more than doubling in dollar terms.
"Overall VC deal count for the first half of 2026 was 21, nearly as many as the 23 transactions in all of 2025... Total VC funding was over $8 billion, more than twice the $3.8 billion deal value for all of 2025."
Mega-rounds are becoming normalized, with tier-1 investors and strategics piling in:
"Deals exceeding $1 billion are becoming more common. During that period, buzzy startups such as Physical Intelligence, SkildAI and Mytra raised capital from leading investors, including Sequoia, Eclipse and Khosla Ventures. NVIDIA and other strategic investors also placed their bets."
Labor Economics Are the Core Driver of Physical AI Adoption
The business case for autonomous warehouses is structural, not speculative — rising wages, extreme turnover, and costly errors are compounding pressure on operators.
"Wages are rising—up 12% over three years as of May 2026, per the US Bureau of Labor, and worker turnover exceeds 100% annually, according to the National Employment Law Project. Meanwhile, warehouse errors result in expensive returns and customer service calls, eroding already-narrow ecommerce margins."
Amazon's public commitment crystallizes the demand signal for the entire industry:
"With Amazon recently announcing a goal of doubling products sold by 2033 without increasing its US workforce, physical AI is the only answer."
Gaming VC Is Pivoting from Content to Infrastructure
Capital is rotating away from game studios toward AI tooling and infrastructure companies serving the gaming industry — a signal that investors see more durable value in the picks-and-shovels layer.
"Video gaming VC is increasingly flowing to the companies building AI, not the games themselves, as our Q2 2026 Gaming VC First Look makes plain."
AI Infrastructure IPOs Are Hitting a Wall of Skepticism
Despite the broader AI investment boom, the public markets are showing clear fatigue with legacy-infrastructure plays rebranded as AI beneficiaries — particularly those with debt-heavy balance sheets.
"The IPO priced at $21 per share and raised $1.05 billion... This is below its target range of $23 to $27 per share."
"Proceeds of the IPO are devoted almost entirely to debt repayment, which may have discouraged some investors. The data center operator also paid a $785 million distribution to its owner last year, equivalent to more than four times the business's $172 million of operating cash flow in 2025."
AI Sovereignty Anxiety Is Reshaping Global Data Center Investment
Governments outside the US and China are rethinking data center strategy as a matter of national security, creating a new geopolitical driver for infrastructure investment.
"AI sovereignty anxiety is reshaping the data center investment calculus for every government outside Washington and Beijing."
2. Contrarian Perspectives
The "AI data center boom" narrative obscures serious asset-quality problems in legacy infrastructure. The Csquare IPO underscores that not all data center assets are equal — older facilities built for cloud, not AI, may require costly retrofits and carry structural financial risk that the market is beginning to price in.
"Much of Csquare's portfolio wasn't originally designed for the largest high-density AI workloads... 'They're going to have to make investments in their data centers,' said Tim Caulfield, the founding chief executive of Evoque Data Center Solutions."
"Csquare's financing costs have continued to outpace its growth, and expanding the business will require more capital still."
Physical AI's real disruption is labor replacement at scale — and the labor market is beginning to feel it. While most coverage frames physical AI as a productivity story, the article opens with a labor-conflict angle that hints at the coming social and political friction.
"Hyundai workers in South Korea went on a partial strike this week, concerned about exactly what the physical AI sector aims to achieve: fully automated factories that pick, pack and push packages through with little need for lights or labor."
Traditional warehouse automation is already obsolete — not just inefficient. The shift to probabilistic, self-learning systems is a paradigm replacement, not an incremental upgrade, which means incumbents in the legacy automation space face existential risk, not just competitive pressure.
"Traditional automation relies on explicit, preprogrammed rules that break down when conditions change, while physical AI systems learn from data and extrapolate to novel situations. This shift moves warehouses from static, engineered systems toward adaptive, probabilistic ones that are expected to take over a much larger share of warehouse work."
3. Companies Identified
- Description: Physical AI startup
- Why mentioned: Named as one of the "buzzy startups" raising mega-rounds from top-tier VCs in H1 2026
- Quote: "Buzzy startups such as Physical Intelligence, SkildAI and Mytra raised capital from leading investors, including Sequoia, Eclipse and Khosla Ventures."
SkildAI
- Description: Physical AI startup
- Why mentioned: Same cohort of large-round physical AI raises alongside Physical Intelligence and Mytra
- Quote: Same as above
Mytra
- Description: Physical AI / warehouse robotics startup
- Why mentioned: Part of the H1 2026 physical AI funding wave
- Quote: Same as above
Csquare
- Description: Dallas-based data center operator (born from Cyxtera + AT&T assets) owned by Brookfield Asset Management
- Why mentioned: Featured as a case study of a troubled AI infrastructure IPO that priced below its range
- Quote: "The IPO priced at $21 per share and raised $1.05 billion... This is below its target range of $23 to $27 per share."
- Description: Boutique AI cloud provider
- Why mentioned: Executed an atypical direct listing, flagged as a novel market structure event
- Quote: "A boutique AI cloud provider is bringing the direct listing back, but QumulusAI's public listing yesterday is nothing like the ones we've seen before."
Amazon
- Description: Global e-commerce and logistics giant
- Why mentioned: Its stated goal of doubling product volume without growing its US workforce is presented as the clearest demand signal for physical AI
- Quote: "With Amazon recently announcing a goal of doubling products sold by 2033 without increasing its US workforce, physical AI is the only answer."
- Description: Enterprise AI model training and serving platform
- Why mentioned: Raised a $1.5B Series D at a $17.5B valuation — one of the largest VC rounds in the issue
- Quote: "Fireworks, which helps enterprises train and serve custom AI models, raised a $1.5 billion Series D at a $17.5 billion valuation led by Atreides Management, Index Ventures and TCV."
- Description: Maker of the Atlas humanoid robot
- Why mentioned: Hyundai acquired SoftBank's remaining ~10% stake, completing full ownership consolidation
- Quote: "Hyundai agreed to acquire SoftBank's remaining roughly 10% stake in Boston Dynamics for about $325 million, taking full ownership of the maker of the Atlas humanoid robot."
Wonder
- Description: New York-based food technology company
- Why mentioned: Raised $650M Series D at a $9B valuation from Accel, GV, and NEA
- Quote: "New York-based Wonder, a food technology company, raised a $650 million Series D at a $9 billion pre-money valuation."
Crypto.com
- Description: Singapore-based cryptocurrency exchange
- Why mentioned: Raised $400M from Citadel Securities at a $20B valuation
- Quote: "Crypto.com, a Singapore-based cryptocurrency exchange, raised $400 million from Citadel Securities at a $20 billion valuation."
Astro Mechanica
- Description: Developer of turboelectric supersonic jet engines for data centers
- Why mentioned: Raising a $250M Series B; notable for the unusual application of aerospace tech to data center power
- Quote: "Astro Mechanica, a developer of turboelectric supersonic jet engines it plans to supply to data centers, is raising a $250 million Series B."
Harvey
- Description: Legal AI company
- Why mentioned: Made its third acquisition of 2026 (Benchmark), signaling aggressive M&A-driven growth
- Quote: "Harvey, the legal AI company, acquired Y Combinator-backed Benchmark... its third acquisition of 2026."
Bunkerhill Health
- Description: San Francisco-based developer of agentic AI tools for hospitals and health systems
- Why mentioned: Raised a Series B led by Khosla Ventures
- Quote: "San Francisco-based Bunkerhill Health, which builds agentic AI tools for hospitals and health systems, raised a Series B led by Khosla Ventures."
Inspectorio
- Description: Minneapolis-based AI-powered supply chain quality and compliance software company
- Why mentioned: Received a growth investment from Apax Digital
- Quote: "Apax Digital made a growth investment in Inspectorio, a Minneapolis-based AI-powered supply chain quality and compliance software company."
4. People Identified
Jonathan Geurkink
- Description: Senior Research Analyst, Mobility Tech and Supply Chain Tech at PitchBook
- Why mentioned: Author of the physical AI / self-learning factory article
- Quote: Byline credit for the piece
Tim Caulfield
- Description: Founding CEO of Evoque Data Center Solutions (the AT&T data center business that became part of Csquare)
- Why mentioned: Provided expert commentary on Csquare's infrastructure limitations and capital needs
- Quote: "'They're going to have to make investments in their data centers,' he said."
Ryan Beiermeister
- Description: Newly appointed partner at Founders Fund; former frontier safety systems lead at OpenAI
- Why mentioned: Notable talent movement from AI lab to VC firm
- Quote: "Founders Fund appointed Ryan Beiermeister as a partner on its investment team, joining from OpenAI, where she led frontier safety systems."
Kashmir Hill
- Description: Reporter at The New York Times
- Why mentioned: Investigated AI-generated biographies of journalists appearing on Amazon, flagging an AI content integrity issue
- Quote: "New York Times reporter Kashmir Hill found AI-generated biographies of herself and other journalists all over Amazon."
Michael Klein
- Description: SPAC veteran
- Why mentioned: Filed Churchill Capital XIII for a $300M IPO, signaling continued SPAC market activity
- Quote: "Churchill Capital XIII, the latest blank-check company from SPAC veteran Michael Klein, filed for a $300 million IPO."
5. Operating Insights
For warehouse operators: Reframe automation ROI around labor volatility, not just cost. The article's data — 12% wage growth over three years and >100% annual turnover — gives operators a compelling internal business case for physical AI investment that goes beyond simple headcount math. The hidden costs of errors, returns, and customer service calls are the sleeper argument.
"Warehouse errors result in expensive returns and customer service calls, eroding already-narrow ecommerce margins."
For data center operators: AI workload density is a critical asset-class distinction that investors are now pricing in. Legacy data centers repositioned as "AI infrastructure" without the power density to support high-density GPU clusters face both a capex gap and a credibility gap in the public markets. Operators should be transparent about retrofit requirements or risk valuation compression at exit.
"Much of Csquare's portfolio wasn't originally designed for the largest high-density AI workloads... 'They're going to have to make investments in their data centers.'"
For gaming companies: Compete for AI tooling investment, not just game funding. VC dollars are flowing to infrastructure and AI companies serving the gaming ecosystem, not to studios. Founders in the gaming space should consider whether their product is closer to a "picks-and-shovels" AI tool than a content play — and position accordingly.
"Video gaming VC is increasingly flowing to the companies building AI, not the games themselves."
6. Overlooked Insights
Astro Mechanica's turboelectric supersonic jet engines for data centers is a signal of how acute the power problem has become. That a company developing aerospace propulsion technology is raising a $250M Series B to supply power to data centers — and attracting VC attention — suggests the data center power constraint is severe enough to make exotic energy solutions commercially viable. This is a leading indicator of where energy infrastructure investment may flow next.
"Astro Mechanica, a developer of turboelectric supersonic jet engines it plans to supply to data centers, is raising a $250 million Series B."
Verdane's acquisition of BP's minority tech stakes in energy digitalization and transport electrification companies is a quiet but potentially significant secondaries signal. Large corporates offloading minority tech positions to PE firms often precedes broader portfolio rationalization. It may also indicate that energy majors are pulling back from venture-style bets — creating a pipeline opportunity for specialist funds in the energy-tech secondaries market.
"Verdane acquired a portfolio of minority stakes in more than 10 technology businesses from BP, with the companies focused on energy digitalization and transport electrification."