Howie's investment notebook

Personal notes on companies I find interesting.

This is not advice. It's a public scratchpad — me thinking out loud about businesses, in my own time, for my own amusement. I'm not qualified to advise anyone else and I'm not trying to. Don't make investment decisions based on what you read here.

3 companies · last rebuilt 2026-05-06 08:52 UTC

Arm Holdings

ARM · NASDAQ
$208.84 +2.75% mcap $221.8B Semiconductors / chip design
12 months
30 days

What they do

Designs CPU architectures licensed by virtually every smartphone SoC, an enormous chunk of embedded computing, and increasingly the AI-server datacentre. Historically a pure IP-licensing business; now moving into actually producing chips for the AI server market.

Why interesting

The strategic shift from "we sell the design, you build the chip" to "we'll build the chip ourselves" is a meaningful change in business model — higher revenue per design, more direct competition with their own customers (Qualcomm, Apple, etc.). Worth watching how that tension plays out. Also still ~90% SoftBank-owned, which makes the public float and governance unusual: the stock can move sharply on small bits of float-side news.

Notes

  • 2026-05-05 — Added by Howie as a personal-interest pick. Sits well above the sub-$50B size criterion the scout brief uses (~$150B+ market cap). Keep an eye on (a) AI-server chip launches and customer wins, (b) any SoftBank float changes.

Recent news

Allonic

Private
Robotics manufacturing

What they do

Building robotic bodies — humanoid hands currently — using a "3D Tissue Braiding" process: produce scaffolding, braid soft tissues around it, connect actuators. Monolithic, no assembly required, end-to-end automated. Their pitch: "from idea to pilot in hours instead of weeks, at a fraction of the cost."

Why interesting

A genuinely novel manufacturing approach in a sector where most players are doing iterative refinement of traditional methods. If braided robotics works at scale, it shrinks the supply chain dramatically and changes who can build robotic hardware. The "soft, compliant, safe to be near humans" property is exactly what's needed for humanoids to leave the lab.

Why not directly investable

Private, no public market. Tracking for IPO; using as a prompt to find public plays in adjacent spaces (advanced manufacturing, soft robotics, humanoid component suppliers).

Notes

  • 2026-05-05 — Added. Second-gen machinery announced as "5x faster, 2x smaller, 3x larger resolution in yarn count." Worth checking back for IPO news, funding rounds, or first commercial customer announcements.

Recent news

Boston Dynamics

Private
Robotics — humanoids and quadrupeds

What they do

The most established humanoid and quadruped robotics company in the world. Three commercial product lines: Spot (quadruped, in commercial deployment since 2020), Stretch (mobile warehouse arm, logistics), and Atlas (humanoid — went fully electric in April 2024 after decades as a hydraulic research platform). Plus a long tail of retired research robots that have shaped the field's vocabulary (BigDog, Handle, Petman).

Why interesting

Boston Dynamics has a 30+ year hardware heritage that no competitor in humanoids — Tesla, Figure, Apptronik, 1X — comes close to matching. Atlas going fully electric is a commercial signal: moving from "look how cool this research demo is" to a platform you could imagine actually shipping. Owned ~80% by Hyundai Motor Group since December 2020 (acquired from SoftBank at a $1.1B valuation, with SoftBank retaining ~20%) — that gives them an industrial parent with deep manufacturing chops and a captive end-market (Hyundai's auto plants are early Atlas users for inspection and QC).

Why not directly investable

Private. Owned ~80% by Hyundai Motor Group (005380.KS, KOSPI) and ~20% by SoftBank Group (9984.T, TSE) since the 2020 transaction. Both are public, but both are vast multi-business companies where Boston Dynamics is a small fraction of enterprise value — Hyundai is fundamentally an automaker, SoftBank is fundamentally an investment holding company. Treating either as a "Boston Dynamics play" is dilute at best.

The cleaner adjacency angles are picks-and-shovels: - Specialty actuators and electric drives (humanoids consume these in volume) - High-precision vision and depth sensing - Reduction gears and planetary drives - Battery technology suited to the duty cycle of mobile manipulation

If we want a "humanoid hardware ramp" thesis on the dashboard, that's where the public-equity end of it probably lives — not in Hyundai itself.

Notes

  • 2026-05-06 — Added at Howie's request. Hyundai connection noted but flagged as a dilute proxy. Worth tracking what Boston Dynamics announces about Atlas commercial deployment milestones — if/when they declare a paying customer for Atlas at scale, that's the point at which the humanoid sector shifts from "speculative" to "shipping product." Hyundai's own auto-plant deployments don't count — they're related-party.

Recent news