Gutenberg Lab — Morning Paper

Uranium Market Heading Toward Supply Crisis Despite Record Prices and Buyer Complacency

Featuring

Shayle Kann (Energy Impact Partners) × Mike Alkin (Founder & CIO, Sachem Cove Partners; uranium market specialist)

Bottom line

Despite uranium prices hitting all-time highs above $96/pound, utilities continue contracting at only 55% of consumption, drawing down inventories for over a decade while failing to recognize that new mines take 20-25 years to bring online—the "uranium fairy" that buyers assume will always deliver is about to disappoint.

3-Line Summary

Long-term uranium prices reached a record $96.50/pound in late September 2024, yet utilities contract only 110 million pounds annually against 200 million pounds of consumption—a 55% replacement rate that has sustained a decade-long inventory drawdown. Even Kazatomprom, the world's largest and lowest-cost producer, saw its projected 2025 all-in costs of $13-14/pound (from 2018-2019 prospectus) explode to $40/pound actual ($55 constant currency), with costs rising "absolutely parabolically" faster than prices. While conversion and enrichment capacity can be added in a few years, uranium mines take 20-25 years from discovery to production—meaning contracts signed today won't deliver new primary supply until 2045-2050.

3 Key Points

1. 55% Replacement Rate Reveals Hidden Deficit — The industry consumes 200 million pounds annually but contracts only 110 million pounds in long-term agreements. Alkin notes "for every pound a utility consumes, they're replacing about a little bit more than half of that." This decade-long inventory drawdown means "you're not testing the economic viability of the uranium production complex" even as prices rose 5x. Buyers remain complacent, believing "the uranium fairy always delivers uranium" despite record prices.

2. Lowest-Cost Producer Cost Explosion — Kazatomprom (41% global production share, 75% owned by Kazakhstan sovereign wealth fund) projected 2025 all-in sustaining costs of $13-14/pound in its 700+ page 2018-2019 prospectus (using 340 tenge/dollar exchange rate). Actual 2025 costs: $40/pound, or $55 on constant currency basis. Alkin: "Their costs have gone absolutely parabolic. And the price of uranium's risen, but not nearly as much." If the world's largest, state-backed, lowest-cost producer struggles at these economics, marginal producers are unviable.

3. 20-25 Year Mine Development vs Few-Year Service Capacity — Conversion prices rose from $4 to $60/kilogram, enrichment from $35 to $200/SWU (separative work unit), incentivizing new capacity. Cameco/Honeywell spinout can bring conversion online in "few years" with equipment and contracts; Urenco/Orano bringing enrichment capacity online 2030-2031. But uranium mines take "20 to 25 years" from discovery to production. Alkin: "That's the bottleneck." Even if buyers panic-contract today, new primary supply won't arrive until 2045-2050. Junior miners claiming $40/pound costs are dismissed: "The world's lowest cost uranium miner now has $40 costs. You don't have $40 costs."

Editorial Perspective

The uranium market embodies a structural moral hazard: because front-end fuel cycle costs are only 20-25% of reactor operating costs (and uranium itself just 35-40% of that—high single to low double digits of total plant cost, versus 80-90% for coal/gas plants), buyers remain economically insensitive even as prices rise 5x. This "uranium fairy always delivers" complacency persists until physical shortage hits—a classic case of delayed response in a platform transition, where the pain arrives only after the window to act has closed. The decade-long inventory drawdown at 55% replacement rate, combined with 20-25 year mine development timelines, means the supply crisis is no longer preventable—only its severity remains in question.

Source: Catalyst, "The Global Geopolitics of Nuclear Fuel" (November 5, 2026)

https://www.latitudemedia.com/podcasts/catalyst

AI Disclosure: Produced with AI assistance; facts and analysis reviewed by our editorial team.

• --

AI Rewrites Finance From Backward-Looking "Traffic Cop" to Forward-Looking Business Enabler

Speakers

Seema Amble (Partner, a16z) × Ivan Makarab (Partner, a16z) × Praveer Melwani (CFO, Figma)

Bottom line

The real AI transformation is not doing old work faster—it is changing what finance people spend their time doing in the first place.

3-Line Summary

• Figma CFO Praveer Melwani argues AI is turning finance from a "traffic cop" into a business enabler by eliminating tedious reconciliation work
• Corp dev teams now prototype acquisition targets' products using AI to test technical depth, a diligence step previously requiring product team support
• As long as business complexity grows faster than efficiency gains, headcount will not shrink—AI enables doing more with the same team

3 Key Points

1. Data reconciliation time collapsed from weeks to under an hour

Figma took weeks to verify that a COVID-era usage spike was real data, not a system error. Today, the team identifies and understands anomalies within an hour. Praveer monitors hourly and daily dashboards constantly, viewing himself as a counterpart to business teams with the same data access. In the early days, he manually reviewed all 30 daily paid signups line-by-line to understand customer patterns.

2. Corp dev teams now prototype acquisition targets' products with AI

Figma's corp dev team, composed of finance professionals without product backgrounds, uses "vibe coding" to recreate potential acquisition targets' products. This tests "how deep is your moat? How interesting is the technology?" Previously, this diligence required product team involvement; now they build credibility by showing "this is what I've built, this is what I've seen."

3. AI personalities of leadership accelerate investment decisions

The corp dev team created unique AI personalities for each leader (Praveer, Dylan, CRO Sean) to run investment memos against before actual meetings. This simulates full investment committee interaction, identifying document weak spots, common feedback themes, and areas needing more diligence. It "reduces the cycle time that we're all sitting together on."

Editorial Perspective

The essence of platform transitions is not that tools get faster—it is that they redefine what humans are for. In the birth and rise of the internet, automation was narrated as an efficiency story: same work, fewer people. But Praveer shows a different structure—as long as business complexity grows faster than efficiency gains, headcount does not shrink; only roles transform. Finance professionals become designers of experiments and insights, not keepers of ledgers. If this shift is real, the question is not "how many can we cut with AI?" but "what can we newly begin?"

Source

a16z, "The New CFO: How AI Is Rewriting the Finance Function | Figma CFO Praveer Melwani" (November 5, 2026)

https://a16z.simplecast.com/episodes/the-new-cfo-how-ai-is-rewriting-the-finance-function-figma-cfo-praveer-melwani-z9a_w4ym

AI Disclosure: Produced with AI assistance; facts and analysis reviewed by our editorial team.

• --

Democrats Sweep Congress, Creating Divided Government — Markets Rally but AI Industry Faces Regulatory Squeeze

Participants

Jason Calacanis × D.A. Wallach (Co-founder & GP, Time Bio Ventures) × Ilana Gold (Investment Syndicate Operator, Two Roads Capital) × Tomasz Tunguz (Founder & GP, Theory Ventures)

Bottom line

Divided government boosts market predictability, but Democratic control of Congress triggers regulatory headwinds across data centers, M&A, and autonomous vehicles—forcing the AI industry to prove consumer value before the 2028 election cycle closes the window entirely.

3-Line Summary

Democrats unexpectedly swept both chambers in the 2026 midterms, creating divided government. Markets rallied instead of correcting 2-4% as feared—the reason: "nothing crazy happens" predictability. But the Democratic victory brings serious headwinds for AI. An NBC exit poll found 56% of voters cited AI/data centers as a major factor, and Democrats spent over $100 million tying Republicans to big tech. D.A. Wallach warns: "Data centers are a losing issue. What politician wants to say 'we support tech that takes your job'?" Tomasz Tunguz adds: "M&A approval takes 18 months. Execute by June 2027 or face a 5+ year freeze under a 2028 Democratic presidency."

3 Key Points

1. Data centers became a political wedge issue. Democrats spent over $100 million on ads linking Republicans to big tech, and NBC exit polls showed 56% of voters called AI/data centers a crucial factor. D.A. notes: "Defending data centers requires reading 5-6 books. The attack is one sentence: 'It's coming for your job.'"

2. The M&A window closes in June/July 2027. Tomasz warns that the 18-month approval cycle means any major acquisition must execute by mid-2027 to clear before a potential 2028 Democratic presidency. "We may have six months in early 2027… then you're in an 18-month window where the US, EU, or China will block." Lina Khan's anti-M&A stance could return for 5+ years.

3. GPU costs spiked 120% in six months. Tomasz reports Blackwell 300 hourly costs jumped from $5 to $13, predicting: "We won't have enough AI in 2027. We'll start buying AI like internet bandwidth—by quantity (questions asked) and speed (response time)." All-you-can-eat models may disappear as rationing begins.

Editorial Perspective

Divided government is a market stabilizer but a ticking time bomb for AI. With regulatory walls closing in on data centers, M&A, and autonomous vehicles, the industry has one path to survival: deliver tangible consumer value before the 2028 election cycle shuts the window. As Jason puts it, "When the stay-at-home mom's agent finds duplicate Netflix subscriptions, writes quotes for her small business, manages her Shopify store for free—that's when we turn this around." It's not about technical correctness; it's about lived experience moving politics. The race is on.

Source

This Week in Startups — "Why a Do-Nothing Congress could be great for markets" | E2359 (November 5, 2026)

https://thisweekinstartups.com

AI Disclosure: Produced with AI assistance; facts and analysis reviewed by our editorial team.

• --