The Physical AI Opportunity Nobody’s Talking About | First ForeignerDom DotzauerMarch 202612 min read

The Physical AI Opportunity Nobody’s Talking About — And the PE Dealmaker Building It from Shenzhen

Investors are moving from bits to atoms. From SaaS to steel. From software margins to physical moats. And one person building the infrastructure for this shift is a Norwegian PE veteran who taught himself to code.

There’s a massive shift happening in venture capital right now, and it has nothing to do with chatbots.

The “SaaSpocalypse” has wiped over $1 trillion in market cap from software companies since January 2026, while energy, hardware, and infrastructure stocks are surging. The trade has a name: Atoms over Bits.

But here’s what most people watching from the outside miss: the best place to build the next generation of physical AI companies isn’t San Francisco. It isn’t Boston. It isn’t even Shenzhen — as most people understand Shenzhen.

It’s Shenzhen as a starting point. Not as a factory. As a company builder.

And one person who understands this distinction better than almost anyone is Alexander Adamov, a Norwegian-born PE dealmaker who taught himself to code, moved between Oslo, London, and Germany — and is now launching B&A (Bits and Atoms), a physical AI accelerator based in Shenzhen, with its first cohort starting July 1.

From Billion-Dollar Deals to Breadboards

Alex’s background is not what you’d expect from someone launching a hardware accelerator in southern China.

He spent twelve years in private equity — first at Terra Firma Capital Partners in London, where he worked on deals including Infinis (UK renewables), Everpower (US wind generation), Odeon cinemas, and Tank & Rast, the German motorway service station concession. Then at Platinum Equity, where he was part of the London deal team executing transactions for Worldwide Flight Services (the world’s largest airline cargo handler), CBS Outdoor (later Exterion), and Consolidated Aviation Services.

After PE, he spent time at McKinsey in Oslo, then Yara International (the world’s largest fertilizer producer), and Kistefos AS — the prestigious Norwegian family office of billionaire Christen Sveaas — where Alex served as Investment Manager for B2B technology.

Then he did something unusual: he taught himself to code. Built real applications. PDF analysis tools, research-paper-to-audio converters. Full-stack, using Vercel, AWS, Google Cloud, and eventually AI coding tools like Claude Code.

And then he went to Shenzhen.

The Gap That Nobody Filled

Alex arrived with a specific ambition: build something in robotics. He had the software skills, the analytical framework from a decade of investment, and the conviction that physical AI was the next great opportunity.

What he didn’t have was the hardware infrastructure: factory relationships, mechanical engineering support, industrial design, and the kind of structured company-building help that software founders take for granted at YC or Techstars.

He went to TroubleMaker — a well-known prototyping lab in Huaqiangbei run by Henk, a veteran of Shenzhen’s maker ecosystem. TroubleMaker is excellent for what it does: shared workshops, 3D printing, electronics prototyping, supplier intros. It serves individual makers and hardware tinkerers.

But Alex needed something different. Not a workbench. A company-building platform. Someone to help sharpen the product thesis, validate the market, think about brand and design alongside engineering, and connect him to the right part of Shenzhen’s manufacturing ecosystem — not just any factory.

The software accelerators don’t understand BOMs and tooling. The hardware labs serve individual makers, not company builders. HAX had shifted its center of gravity away from Shenzhen. And the Chinese ecosystem — astonishingly capable — remained largely inaccessible to international founders.

That infrastructure didn’t exist. So Alex built the thing he wished existed.

What B&A Actually Is

B&A (Bits and Atoms) is a Shenzhen-based accelerator for founders building physical AI and smart hardware companies. First cohort: 10 slots, starting July 1, 2026.

What makes it different from the dozen other accelerators that have tried to crack Shenzhen:

It’s not a makerspace. B&A works on company thesis, go-to-market, brand, design, and fundraising — not just prototyping.

It’s opinionated. B&A has published six detailed Request-for-Proposal briefs — carefully framed company concepts that founders can apply to build. These aren’t vague themes. They’re investment-memo-grade analyses of specific market opportunities.

It’s built by an investor, not a maker. Alex brings 12 years of evaluating what makes companies work — and fail. When he writes “we pressure-test go-to-market,” that’s not marketing copy. That’s his profession.

It’s international-first. English-speaking, globally-networked, designed as a bridge between international founders and Shenzhen’s hardware ecosystem.

The Three RFPs Worth Watching

B&A has published six RFPs. Three of them are particularly interesting because they sit at the intersection of massive capital flows, proven market demand, and Shenzhen’s unique advantages:

1. Warehouse “Hard-Task” Robotics (RFP D)

The warehouse robotics market is exploding — 102,900 logistics robots sold in 2024, with adoption doubling in three years. But the real opportunity isn’t in AMRs on flat floors. It’s in the hard tasks: unloading mixed cartons from trailers, sorting damaged returns, depalletizing mixed SKUs. The messy stuff that requires unstructured manipulation and real-world intelligence. $7B+ Global warehouse automation market for mobile robots by 2025

The investment signal is strong: Tutor Intelligence (MIT CSAIL spinout) raised $34M in a Series A in December 2025. NEOintralogistics in Germany closed a €3M seed in January 2026. Lux Capital, Sequoia, and NVIDIA NVentures are all actively deploying.

Why Alex picked this: at Platinum Equity, he worked on Worldwide Flight Services — the world’s largest airline cargo handler. He knows first-hand how physical logistics businesses scale and where labor bottlenecks destroy margins.

2. Edge Inference Compute Modules (RFP F)

This is the infrastructure layer. Every other B&A company — the robots, the grid devices, the microfactories — needs edge inference hardware to run AI models locally. The market is massive: AI chip startups raised $5.1B in the first half of 2025 alone. $50B+ Projected inference-optimized chip revenues in 2026

The players to watch: Hailo (Israel, 26 TOPS at 2.5W), Axelera AI (Netherlands, €61.6M from EuroHPC), Groq ($6.9B valuation), Tenstorrent ($3.2B, backed by Samsung and Hyundai).

Why Alex picked this: at Kistefos, he was the B2B technology Investment Manager. He evaluated platform businesses with compounding advantages. Edge inference is exactly that — a horizontal enabler where each deployment makes the next one better.

3. Microfactory Cells (RFP A)

Software-defined, modular, copy-exactly manufacturing cells that can be rapidly deployed and reconfigured. Think of it as the “Shopify of factories” — standardized units with software configuration, targeting high-mix, low-volume manufacturers who can’t justify traditional automation. $435M+ Raised by Bright Machines for software-defined manufacturing

The validation: Bright Machines has raised $435M+. Formic offers RaaS for factory automation with zero upfront costs. Vention lets manufacturers design and deploy workcells in days.

Why Alex picked this: at Terra Firma, he worked on Tank & Rast — a business built on replicable physical units with standardized operations. The microfactory concept is the manufacturing equivalent.

All Six RFPs at a Glance

RFPWhat it is (simply put)Why it matters | A | Mini-Factories — modular, software-controlled production units | $658B smart manufacturing market by 2030 | B | Smart Grid Devices — sensors and controllers for power grids | Renewable energy can’t connect without grid upgrades | C | Industrial Heat Pumps — electric replacement for fossil furnaces | Biggest decarbonization opportunity in industry | D | Warehouse Robots — for the messy, hard tasks | 80% of warehouses still operate manually | E | Green Steel Tech — sensors and controls for cleaner steelmaking | Steel = 7% of global CO2 emissions | F | AI Chips for Devices — affordable edge inference hardware | $50B+ inference chip market in 2026

Why This Matters for China’s Next Chapter

Shenzhen’s government has published a 2025–2026 action plan projecting over 3,000 AI companies by 2026, including 10+ unicorns, with 10+ innovation incubators for AI and embodied intelligent robots. The city’s AI industry is targeting 20%+ annual growth.

But there’s a gap in that ecosystem: international founders. Shenzhen has the supply chain, the engineering talent, and the iteration speed. What it’s historically lacked is a structured on-ramp for builders from outside China.

B&A is designed to be that on-ramp. And Alex’s profile — Norwegian education, London PE, McKinsey strategy, Kistefos investment, self-taught coding, Shenzhen hardware ambition — makes him one of the very few people who could credibly build it.

The PE guy went to Shenzhen. Shenzhen didn’t have what he needed. So he built it.

Applications for Cohort 01 are open

10 founder slots. Starting July 1, 2026. Shenzhen. Visit bitsandatoms.tech → Dom Dotzauer is a physician, coder, and cross-border operator between Shenzhen and Berlin. He writes about China tech, biotech, and the intersection of physical and digital at firstforeigner.com.

Alexander Adamov can be reached on LinkedIn.

Related:

• B&A Accelerator
• B&A on LinkedIn
• TroubleMaker Shenzhen