“Weekly Dose of Optimism #190” — @packyM

Why this is in the vault

The Friday WDO is Packy’s curation rhythm. Five items this week — pancreatic-cancer mRNA vaccine, SpaceX/xAI’s $10B option on Cursor, Fervo’s S-1, Humble’s cabless autonomous truck, and Medra’s robotic bio lab — plus an Extra Dose grab bag. The Cursor/SpaceX deal mechanics, Fervo’s $7K→$3K/kW geothermal cost curve, and Medra’s “TSMC for biology” framing are the items that cross the RDCO relevance threshold.

Medra portfolio check (verified 2026-04-24): Medra’s public investor list (Series A Dec 2025, $52M) = Human Capital (lead) + Lux + Neo + NFDG + Catalio + Menlo + 776. Not Boring Capital is NOT in the public investor list. Packy’s enthusiasm is consistent with editorial-fan framing rather than disclosed investor framing — but Packy doesn’t always disclose his angel positions, so absence-of-public-disclosure isn’t definitive. Treat Medra coverage as third-party but flag Packy as messenger when quoting his specific framing.

⚠️ Sponsorship

Silicon Valley Bank — third-party paid. Explicit “brought to you by” block at top of email promoting SVB’s Future of Climate Tech 2026 report. Disclosure is clean. Bias note: the SVB block frames climate tech as “discipline returning, hype receding” which echoes Packy’s own house view, so the sponsor-content alignment is suspiciously tight — read the SVB report as a vendor pitch, not independent research, even though the headline stats ($29B invested, 52% cutting burn, climate mentions on earnings down 70%) are usable as directional signal.

Issue contents

Pancreatic cancer mRNA vaccine — six-year Phase 1 follow-up (NBC News). MSK’s Balachandran presented at AACR. 16-patient trial of autogene cevumeran (BioNTech/Genentech) + atezolizumab + chemo. 8 of 16 mounted T-cell response; 7 of those 8 still alive at 6 years (87.5%) vs 2 of 8 non-responders (25%). 85% of T-cell clones persisted into memory phase. Phase 2 now running globally. Personalized neoantigen mRNA playbook generalizing beyond melanoma.
Cursor + SpaceX/xAI — $10B option to acquire at $60B (Kevin Kwok). SpaceX/xAI bought a call option on Cursor: if they don’t exercise post-IPO, they pay a $10B breakup fee. Kwok’s first essay in nearly a year. Deal mechanics (“expect more like this”) + strategic rationale (Cursor has product+model, no compute; SpaceXAI has compute, weak coding models).
Fervo Energy files S-1 (FRVO on Nasdaq; JPM/BofA/RBC/Barclays leading). 3.65 GW operating or developing — would nearly double US installed geothermal. Cape Station (Beaver County, UT): 500 MW under construction, permits for another 1.5 GW. 150 MW Nevada site under Google + NV Energy deal targeting 2030. Cape Station targets $7,000/kW (parity with advanced nuclear); company goal $3,000/kW (beats unsubsidized natural gas for always-on power). Latimer’s shale-drilling background (BHP) is the technique transfer.
Humble — cabless autonomous electric truck, $24M seed (TheNextWeb). SF startup led by Eyal Cohen (Apple, Uber ATG, Waabi, Spark AI→John Deere). Removed the cab entirely → 360° sensor coverage, lighter vehicle, geometry optimized for 40/53-ft intermodal containers, dock-to-dock delivery (vs Aurora’s hand-off-at-yard model). Eclipse led; Energy Impact Partners participated. Prototype in <6 months. Frame: “design around the actual constraint, not the inherited one” — same logic as Kalanick’s Atoms (anti-humanoid → wheels not legs) and Tesla Cybercab.
Medra — robotic bio lab opens 38,000 sq ft SF facility (Core Memory + Michelle Lee on X). ~100 robotic arms running biology experiments 24/7, courier robot ferrying tips/plates between stations. Key insight: only 5% of bench bio equipment has APIs; Medra uses computer vision + manipulation models to operate human-designed instruments, claiming a path from 5% → 75% of bio tasks automatable. Every arm instrumented with cameras + 9 sensor types; AI scientist proposes protocol changes (one customer: zero antibody binding → AI proposed adding vortex step → 70%+ binding). Frame: “TSMC for biology” — drug discovery cos run experiments without owning a wet lab.

Plus Extra Dose grab bag: science breakthroughs, oceans, flipbook, Tokyo + NYC subways, “Boots Manifesto” — not enumerated in the email body provided.

Mapping against Ray Data Co

Medra (item 5) is the strongest mapping. The “5% of equipment has APIs, the other 95% is human-designed instruments — use vision + manipulation to operate them anyway” pattern is directly analogous to RDCO’s stance on legacy data tooling: most enterprise data systems weren’t built for agents either, and the wedge is computer-use-style automation that reads what humans see rather than waiting for vendor APIs. The “AI scientist proposes a vortex step → binding goes 0 → 70%” loop is the lab-experiment version of the closed-loop data-quality reasoning Ray’s been sketching for the autonomous COO. File Medra as a positioning analog for the “operate the legacy stack via vision, not vendor cooperation” thesis.
Humble (item 4) reinforces the design-around-the-actual-constraint principle that Sanity Check has been circling — the cabless truck is the truck-shaped version of “rebuild the org chart for what AI agents actually do, not for what humans needed.” Useful one-line analog for any “rebuild from first principles” essay.
Fervo S-1 (item 3) is the kind of energy-economics datapoint worth keeping in the vault for Sanity Check’s occasional “what’s getting cheaper” cadence — $7K/kW heading toward $3K/kW for 24/7 carbon-free power is a clean tangible number. Medium mapping.
Cursor/SpaceX (item 2) is interesting deal mechanics but doesn’t map to RDCO directly. File for “the next time Packy’s curation circles AI deal structures” reference.
Pancreatic mRNA (item 1) — out of scope for RDCO mapping; keep as cultural-optimism marker, not a thesis input.

Mapping strength: medium-strong (Medra is a real positioning analog; Humble adds a useful framing line; the rest is context).

Curation section — notes

Item 1 (NBC News, AACR data) — third-party, primary-source-adjacent. No Not Boring Capital portfolio overlap visible.
Item 2 (Kevin Kwok essay) — Kwok is a frequent Not Boring guest collaborator; Packy explicitly markets “first Kevin Kwok essay in nearly a year.” Cross-promo / sister-author dynamic, not strictly third-party. Flag as friendly-network amplification rather than neutral curation, but the deal-mechanics analysis is substantive.
Item 3 (Fervo S-1) — third-party news; Packy has covered Fervo and the broader geothermal space repeatedly (Quaise last week). House thesis amplification, not paid placement.
Item 4 (Humble, via TheNextWeb) — third-party reporting; no obvious portfolio tie disclosed in the email. Eclipse + Energy Impact Partners are the named investors.
Item 5 (Medra, via Core Memory + Michelle Lee on X) — Core Memory is Ashlee Vance’s publication. Michelle Lee is Medra’s founder, so the on-X source is the company itself. No explicit portfolio disclosure in the email, but Not Boring Capital’s investment posture toward AI-for-bio means worth flagging for manual check whether Medra is a portfolio company. Treat the framing (“get to work, robots”) as enthusiastic, not neutral.
SVB sponsor block — labeled clearly; do not count as curation.

No self-cross-promo to Packy’s own essays in the curation slots beyond the link to Elliot’s prior “Going Founder Mode on Cancer” piece (item 1) and Will/Packy’s “Great Blue Frontier” (item 5) — both labeled as Not Boring’s own writing, fair use of curation-as-house-canon.

2026-04-17-notboring-wdo-189 — last week’s WDO led with Quaise (geothermal drilling), making this week’s Fervo S-1 a direct continuation
2026-04-23-notboring-great-blue-frontier — Packy/Will’s piece on AI-for-bio being data-limited; Medra is positioned in this issue as the throughput-side answer to that bottleneck
Newsletter whitelist: ~/rdco-vault/01-projects/process-newsletter/README.md

Tracked-author candidates

Michelle Lee (Medra founder) — surface to Task #4 (Add Twitter/authors to CRM workflow) as a candidate. Medra’s positioning vocabulary (“TSMC for biology”, “5%→75% automatable”) is the kind of frame RDCO tracks.
Kevin Kwok — already on the radar, but worth re-flagging: first essay in nearly a year on a topical deal. If he resumes regular cadence, add to the watch list.