Where Deep-Tech Capital Actually Flows

$375M Series C at a $2B valuation, co-led by a16z and DFJ Growth; other backers include Alpine Space Ventures, Washington Harbour Partners, StepStone and In-Q-Tel, plus an undisclosed HSBC Innovation Banking credit facility; over $500M raised to date.

Northwood builds standardized, quickly deployable dishes (its Portal and Prism antennas) and lets satellite operators reserve capacity through an API — ground stations as a service.

Anchor customer is the US Space Force, under a three-year, $49.8M Satellite Control Network contract; the company runs a 185,000 sq ft factory in Torrance, California.

Launch and satellite costs have collapsed, but ground infrastructure hasn’t scaled with them — downlink is a genuine constraint on the whole sector.

A shared, API-driven network lets operators avoid building bespoke ground stations, echoing the cloud model.

The In-Q-Tel and Space Force involvement signals this as dual-use national infrastructure, not just commercial plumbing.

A rare private pure-play on space infrastructure rather than another constellation — it benefits whoever wins upstream.

Blue-chip syndicate (a16z, DFJ, In-Q-Tel) and a government anchor contract de-risk early revenue.

Risk caveat: scaling a global physical antenna footprint is capital-intensive and execution-heavy; a $2B valuation prices in rapid network expansion, and demand depends on the continued pace of satellite deployment.

$75M Series B led by a16z, with earlier leads Union Square Ventures and Neo also participating; total funding now $117M.

Tutor’s internet-connected robots (named Sonny and Cassie) are guided by remote human “tutors” and run the company’s own models; every completed task generates training data for its robot foundation model.

Customers pay by the hour or per pick with no purchase or long-term contract; deployments have nearly quadrupled in a year, with a target of 1,000+ US-built robots under contract by end-2027.

The robots-as-a-service model removes the capex and integration risk that stalls factory automation for most manufacturers.

A human-in-the-loop “tutor” design turns every deployment into a data-collection engine — a compounding moat if it holds.

It validates the “specialized beats general” argument: narrow tasks, real revenue, faster payback than humanoid moonshots.

Recurring, usage-based revenue with a built-in data flywheel and an a16z-led syndicate across three rounds.

MIT CSAIL founding team (CEO Josh Gruenstein) and quadrupling deployments point to genuine commercial pull.

Risk caveat: RaaS is balance-sheet-heavy to scale, margins depend on robot utilisation and the economics of remote human oversight, and the segment is crowding fast.

Listed as AGLT via a Churchill Capital Corp XI SPAC merger; roughly $465M raised with Foxconn as lead investor, at an approximately $2.5B valuation.

Agility’s bipedal robot Digit works in warehouses, logistics hubs and manufacturing; named customers include GXO, Amazon, Toyota, Schaeffler and AGCO.

The Oregon-based company reported only about $1.78M in revenue, and filings show steep losses — scrutiny that several outlets flagged on debut.

The first pure-play humanoid listing gives the public market a direct read on the category — and a valuation to argue over.

Foxconn as lead strategic investor signals manufacturing muscle behind the deployment story.

It sharpens the week’s core tension: humanoids command the headlines and the IPO bell while private capital concentrates on specialized physical AI (see Tutor above).

A liquid, transparent way to track humanoid economics as real quarterly numbers arrive.

Multi-year deployments at blue-chip logistics customers give it more commercial grounding than most humanoid peers.

Risk caveat: minimal revenue against steep losses and a ~$2.5B valuation means shares will trade almost entirely on future expectations — this is context for the sector, not a recommendation.

$200M round at a $1.2B valuation (per the FT), with investors including Sequoia, Lux Capital, Franklin Templeton and In-Q-Tel.

The company plans to clone and genetically enhance working dogs, plus biobanking, training and animal care, and to equip dogs and handlers with Anduril technology; Palmer Luckey helped conceive it but holds no equity or role, Ben Lamm sits on the board, and ex-Palantir/DoD figure David Spirk is CEO.

The stated goal is to cut US reliance on European breeders — who supply over 80% of military working dogs — and improve low service-training pass rates.

A serious syndicate (Sequoia, Lux, Franklin Templeton, In-Q-Tel) backing applied synthetic biology for defence marks a notable widening of the deep-tech frontier.

It extends the Anduril/Colossal playbook — defence hardware meets de-extinction biotech — into living systems.

It’s an early indicator of capital flowing toward biosecurity and defence-biology crossovers.

A unique position at the intersection of synthetic biology and defence procurement, with founders who have both built category-defining companies.

Government-adjacent demand (In-Q-Tel) and a clear import-substitution thesis for a real supply gap.

Risk caveat: this is an early, single-purpose seed-stage concept in an ethically and regulatorily charged area, with no product yet and significant public-perception risk — a signal to watch, not an investable round for most.