Every new car sold in Europe now has to watch the driver's face. Two companies in the world can actually do it, Seeing Machines and Smart Eye, and they spent twenty years and hundreds of millions of dollars getting there. Hugo Navarro's argument is that the market has not repriced what happens next: a roughly 55 million dollar fixed cost base, automotive production going from 488,000 vehicles in Q4 2025 to 2.1 million in Q4 2026, and 20 to 40 million of free cash flow in fiscal 2027 against a 330 million market cap. If Japan and the US follow with their own mandates, close to every incremental dollar of revenue drops straight to free cash flow.

I push back hard in a few places. There is a 55 million dollar convertible due in October that this company has let get within two months of expiry, and my view is that no healthy business does that. Receivables are up 120% against 45% revenue growth. The fleet business, Guardian 3, is running trials that keep not converting, and "we are in a trucking recession" is the kind of management excuse I have learned to distrust. We also get into whether a new entrant can just build this now that the market is 16 million vehicles, why no tier one ever bought them, and whether full autonomy eventually kills the whole thesis.

Hugo's write-up on Seeing Machines: https://smallcaptreasures.substack.com/p/a-cheap-tech-duopoly-posts-333-growth?r=1od1d5

This episode is sponsored by Fiscal.ai: https://fiscal.ai/yav. Fiscal.ai is the modern financial data provider for global equities, and I am a customer who pays with my own money for the API. Two things I use it for constantly: a huge database of fund letters wired into the API, so the first thing I do prepping for a podcast is pull every recent letter on the company, and audit-linked financials where every line in the model clicks through to the source. Use fiscal.ai/yav for 15% off their AI connector.

Chapters:

(0:00) The setup: a duopoly Europe just made mandatory

(0:54) Sponsor: Fiscal.ai

(2:49) Why Hugo kept pitching this one

(3:56) What Seeing Machines does, and why DMS is harder than it looks

(5:13) The math: fixed opex, Europe now, Japan and the US later

(8:06) The seatbelt manufacturer analogy

(10:25) My pushback: what stops a new entrant or an in-house build?

(11:47) Naturalistic data, Mitsubishi Electric, and the accuracy gap

(14:54) The elephant in the room: a $55m convertible due in October

(17:46) Footnote 21 and the accelerated royalty payment

(20:06) Can the regulation slip or get watered down?

(22:00) Robotics: $20 of silicon versus $20,000 chips

(24:39) Smart Eye versus Seeing Machines: software only or full system

(27:38) Why no tier one ever bought them

(29:16) Fleet: Guardian 3 and trials that keep not converting

(35:04) The balance sheet: receivables up 120%

(37:37) How much operating leverage is left in Europe alone

(40:10) Does full autonomy kill the DMS story?

(42:39) Chinese OEMs selling into Europe

(44:14) Licensing the fleet software to telematics players

(46:34) CEO incentives and the overpromising track record

(48:30) My last pushback: at some point it is them, not you

(50:14) Why the stock reacts slowly, and where the risk really sits

Hugo Navarro / Undervalued and Undercovered: https://smallcaptreasures.substack.com/

Links:

Yet Another Value Blog - https://www.yetanothervalueblog.com

See our legal disclaimer here: https://www.yetanothervalueblog.com/p/legal-and-disclaimer

Rates just screamed to 20 year highs and stocks have barely blinked. That looks to me like the mirror image of the mid-2010s, when Treasuries yielded 2%, the math said stocks should trade for 25x, and they sat in the mid-teens instead because the equity risk premium quietly widened from 4% to 6%. If the premium can widen when rates fall, why would it not widen again when rates rise? That is the double whammy running in reverse: earnings that got a decade of help from the Trump tax cuts and the AI boom, multiplied by a multiple heading the wrong way.

The other thing I cannot stop chewing on is what higher rates do to the AI data center buildout. These are 15 year leases where the NPV of the payments roughly covers the build cost, which means the developer is really underwriting the terminal value 15 to 25 years out. Move rates from 4% to 5% and you have to jack the lease rate up 5% to 10% just to stand still, and you discount that terminal value harder, right as the tenant credit gets scarier. If the AI trade cracks, you get hit twice: your tenant may not be around, and the release in year 15 goes from a $100m NOI lease to whatever the next best bidder pays. I do not think we are there yet, but finance 101 says investment gets crowded out eventually.

Then two management questions. UWMC and Cogent both ran capital allocation that looked designed for the CEO's personal balance sheet rather than for shareholders, and I want a way to spot that before the blowup rather than after. And a friend's text about a CEO everyone was calling the next Mark Leonard got me wondering how you would ever know, because a real compounder and one great bet with hidden leverage look identical for the first ten years.

I wrote the rates piece up this morning: https://www.yetanothervalueblog.com/p/rates-are-screaming-and-stocks-arent

The UWMC post: https://www.yetanothervalueblog.com/p/uwmc-lost-600m-hedging-a-deal-theyd

The Cogent episode with Aaron Chan: https://www.yetanothervalueblog.com/p/recurve-capitals-aaron-chan-on-cogent

This episode is sponsored by Trata: https://trata.com. Trata is two buy siders talking to each other about a name they both follow closely. Trata records it, anonymizes it, and publishes it. It is the fastest way I know to get up to speed on something new.

Chapters:

(00:00) What is on my mind this month

(01:07) Sponsor: Trata

(01:41) Rates screamed higher and stocks did not listen

(04:39) Should the equity risk premium rise with rates?

(06:29) Rising rates meet the AI data center buildout

(09:33) What a 15 year data center lease is really betting on

(13:03) Does higher for longer start crowding out AI capex?

(14:11) UWMC, Cogent, and CEOs who run capital allocation for themselves

(18:45) How would you know if someone is the next Mark Leonard?

(23:01) One great bet, or actual genius?

(24:37) Wrapping up

Links:

Yet Another Value Blog - https://www.yetanothervalueblog.com

See our legal disclaimer here: https://www.yetanothervalueblog.com/p/legal-and-disclaimer

School is starting, which means a dozen college and MBA teams are about to email me asking how to win their stock pitch competition. So I made the answer. The core of it: a pitch is a game, and most people lose it before they open their mouth by picking an idea that does not fit the contest rules or the judges in the room.

From there it is three things. Design the pitch for the timeframe the contest actually asks for and for the people judging it, because what wins with a concentrated-book judge is not what wins with a pod shop. Lead with the one thing only you know, not a sell side price target or a multiple that has compressed. And make your bull case the base case instead of hedging yourself into a 15% price target that reads as average. Then the three traps I see every single year: burning five slides on a DCF nobody will ask about, drowning the room in risk factors, and death by background. I also walk through the La Quinta pitch that won me Ira Sohn in 2018, and why the CSL and DoorDash teams at the Pershing Square Challenge won on legwork rather than modeling.

Fair warning: I had AI build the slides, so do not hold the exact wording on any of them against me.

If you are pitching to get hired rather than to win a contest, the companion episode is here: https://www.yetanothervalueblog.com/p/how-to-get-a-job-in-investing-podcast

This episode is sponsored by Trata: https://trata.com. Trata is two buysiders swapping thoughts on a stock they are both involved in. If you are prepping a pitch, go on and say you are thinking about pitching company XYZ, and they will find you someone to talk to about it. It is a very good way to hear the other side of your idea before a judge hands it to you.

Chapters:

(00:00) Why I made this one alone, with a deck

(01:57) Disclaimer and a word from Trata

(03:02) Why this matters even if you never enter a contest

(05:20) Who am I to talk about stock pitches

(07:18) Why a pitch is a free lottery ticket

(08:51) Rule one: know the game you are playing

(10:12) Know your judges: concentrated books, event funds, pod shops

(12:19) Rule two: tell a story, and lead with something only you know

(14:14) The La Quinta pitch that won Ira Sohn

(16:07) Be bold: make your bull case the base case

(18:04) Do the legwork: hard hats, expert calls, customer checks

(21:43) What to avoid: excessive modeling

(23:17) What to avoid: drowning in risks

(25:08) What to avoid: death by background

(26:22) Formatting is table stakes

(27:48) Go win the thing

Links:

Yet Another Value Blog - https://www.yetanothervalueblog.com

See our legal disclaimer here: https://www.yetanothervalueblog.com/p/legal-and-disclaimer