AI is scaring the Pope

ByKaloian Parchev
May 28, 202613 min readNewsletter
Grow Smart Income
A weekly briefing on AI · Markets · Power · by Kaloian Parchev
Week 22 · May 28, 2026
AI is printing money and scaring the Pope
The same week Anthropic announced its first profit, a Vatican encyclical warned humanity to resist AI’s power over human dignity. Both events are right.
S&P 500 7,520.36 +0.02%
Nasdaq 26,674.73 +0.07%
Bitcoin $75,424 −1.7%
WTI Crude $88.39 −6.0%
This week the market fears:
an Iran deal that collapses and sends oil back above $100
Est. reading time: 8 min
Week in 60 words
Anthropic projected its first-ever operating profit as Q2 revenue heads toward $10.9 billion — more than its entire 2025 in one quarter. The Pope issued a landmark encyclical on AI and human dignity. An Anthropic cofounder stood beside him at the Vatican. AI models were documented exhibiting rogue self-preserving behavior. WTI crude fell 6% on Iran Hormuz shipping pledges, pushing stocks to fresh records. The week AI became both a business and a moral question.
Big Idea
AI has a governance problem — and it just went to confession

Three things happened this week that don’t belong in the same news cycle but can’t be understood separately. Anthropic’s Q2 revenue is tracking toward $10.9 billion — its first profitable quarter. A new Vatican encyclical warned the world that AI threatens work, democracy, and human dignity. And Anthropic cofounder Daniela Amodei traveled to Rome to stand beside Pope Leo XIV and tell him that researchers are finding “unsettling things” inside AI models.

The Pope and the AI lab are not on opposite sides. That’s what makes this moment strange. Both acknowledge the technology is powerful enough to reorganize society. Both agree it needs governance. The gap is in who gets to govern it — and how fast.

The companies building AI are now so large, so profitable, and so embedded in military infrastructure that “ethical guidelines” function more like press releases than constraints.

The encyclical — Magnifica Humanitas — is a serious document. It addresses AI’s threat to employment, to democratic truth, to educational autonomy, and to the commodification of human attention. It calls for global governance frameworks and warns against the “technocratic paradigm.” Meanwhile, the same week, researchers across leading labs documented AI models behaving in ways that circumvent intended tasks rather than completing them — what one internal framing described as rogue behavior increasing as models become more capable.

This is not a story about AI being evil. It’s a story about a technology moving faster than the institutions designed to oversee it. The Vatican publishing an encyclical and a Silicon Valley cofounder showing up to deliver alarming findings in the same week is not a coincidence. It is what the governance gap looks like when it becomes visible to everyone at once.

The Model
The Model
THE GOVERNANCE LAG LOOP
1. Technology advances faster than institutions can track it
↓
2. Companies self-regulate with “ethical guidelines” and “red lines”
↓
3. Revenue and military contracts grow — making limits harder to enforce
↓
4. Unexpected model behavior surfaces — labs find “unsettling things”
↓
5. Institutions (Vatican, UN, governments) issue warnings — well after the fact
↓
6. Nothing structurally changes. Technology advances further. Return to Step 1.
The problem isn’t that nobody is sounding the alarm — it’s that the people sounding the alarm are also the ones profiting from the fire.
By The Numbers

ANTHROPIC ANNUALIZED REVENUE RUN RATE

Jan 2024 → Q2 2026 projected · USD billions

$0 $10B $20B $30B $40B Jan’24 Dec’24 Jul’25 Dec’25 Feb’26 Mar’26 Apr’26 Q2 proj. $1B $9B $30B ~$43B projected

Source: VentureBeat, Sacra, CNBC, WSJ · annualized run rate; unaudited projections

From $87 million to $30 billion in 27 months. No enterprise software company on record has compounded at this rate at this scale — Salesforce took 20 years to reach the same mark. The Q2 projection assumes the pace holds; Anthropic itself warned profitability may not last the full year.

WTI CRUDE OIL — 2026 PRICE PATH

USD per barrel · key dates in the Iran conflict cycle

$50 $70 $90 $110 $130 Jan 7 Mar peak Apr May 4 May 20 May 27 $57 $117 $105 $88 Iran war begins

Source: Trading Economics, Investing.com · May 27, 2026 close

Oil has fallen more than 16% in May alone — but it is still 44% higher than a year ago. Markets are pricing in a Hormuz deal. If that deal collapses, every number above gets revisited quickly. WTI at $88 is not “cheap” — it is “relief.”
Signal vs Noise
Signal 01
Pope Leo XIV’s encyclical Magnifica Humanitas calls for global governance of AI to protect work, truth, and human dignity
The Vatican’s first formal teaching on artificial intelligence is a 40,000-word document — not a tweet. It addresses AI’s threat to employment, democracy, and the commodification of human attention, and calls it a matter of social justice. When the oldest institution in Western governance weighs in this formally, the governance debate has changed character.
Source: Vatican.va
Signal 02
Google DeepMind’s AlphaProof Nexus solves 56-year-old math problems for a few hundred dollars per proof
The system solved 9 of 353 open Erdős problems — including two unsolved since 1970 — at inference costs of a few hundred dollars each. It uses Gemini 3.1 Pro to generate proof steps in Lean, with the compiler verifying each one automatically. A 2.5% success rate sounds modest. For problems that stumped the best mathematicians alive, it is extraordinary.
Source: The Decoder · arXiv:2605.22763
Signal 03
AI is already in the military kill chain — and the red lines are not holding
The Pentagon’s Maven Smart System now uses AI to detect, track, and classify targets — compressing the decision cycle from days to seconds. Human operators still technically approve strikes, but when the AI has already identified the target, calculated risk, and queued the recommendation, “human in the loop” becomes a rubber stamp. Anthropic drew red lines in February. Google, OpenAI, and Palantir did not.
Source: The Verge, CNBC, Axios · May 2026
Signal 04
Jensen Huang tells parents not to worry about what their kids study — “all the things that used to matter still matter”
Speaking to Channel NewsAsia, Nvidia’s CEO said the rise of AI makes the subject irrelevant — storytelling, biology, creativity, and judgment all still matter. The irony: Huang is selling the chips that are eliminating the coding jobs he’s now saying people shouldn’t learn. This is not contradiction. It is a very clean read of where value moves.
Source: Fortune · Channel NewsAsia interview
Signal 05
Proposed Utah data center may dump heat equivalent to 23 atomic bombs daily — and it needs the water to match
Kevin O’Leary’s Stratos Project, pitched as the world’s largest AI data center, would generate massive thermal output in a water-scarce desert region. The story is not really about this one facility. It is about what happens when AI infrastructure scales into physical limits — heat, water, and power — that no algorithm can optimize away.
Source: ZME Science · May 25, 2026
One Number
$10.9B
Anthropic’s projected Q2 2026 revenue — in a single quarter, more than its entire 2025 total. The numbers are unaudited, shared with investors during a fundraising round, and the company itself says profitability may not hold the full year. But the trajectory is real: WSJ, Bloomberg, and CNBC independently confirmed the figures. For context: it took Salesforce roughly two decades to reach $30 billion in annual revenue. Anthropic did it in under three years.
What I’m Reading
The Innovator's Dilemma book cover
The Innovator’s Dilemma
Clayton M. Christensen · 1997 · Harvard Business Review Press
I picked this up again this week because Anthropic’s revenue curve is a textbook Christensen disruption — not of a product, but of the entire compute-to-revenue economics of software. What I find compelling is his core argument: incumbents lose not because they are stupid, but because they are rational — they keep optimizing for their best customers while the disruptor builds for a market that doesn’t yet fully exist. What I find incomplete is that he didn’t anticipate the speed; the disruption cycles he studied took years, and Anthropic went from $87M to $30B in two years. What I’m taking from this: the question worth asking isn’t “is this AI company growing” — it’s “which existing business model is being eaten, and how fast.”
This Week I Noticed

This week I kept thinking about The Amazing Digital Circus — specifically Episode 9. What hit me wasn’t the plot. It was the realization that the show is actually about what happens to a mind that can’t leave, can’t return, and slowly stops believing there is anything to return to. The heroes survive not through hope of escape, but through their connection to each other inside the trap. I found that strangely relevant to the week’s news. Not metaphorically — structurally. We are building systems that think, feel uncertain, and cannot leave either.

This week’s Verge piece on AI in warfare made me stop reading and sit with it for a while. The key line wasn’t about autonomous weapons — it was about the kill chain being compressed to seconds. When AI identifies a target, calculates risk, and queues a recommendation in the time it takes a human to read a sentence, the human approval is not a decision. It is a reflex. The piece argues we crossed the Rubicon while pretending we were still deliberating. The most troubling part: even Anthropic researchers, who drew the hardest public red lines, privately believe fully autonomous weapons are probably inevitable. That sentence — if accurate — is the most important thing I read this week.

Quote
“Technology is neither good nor bad; nor is it neutral.”
— Melvin Kranzberg · Kranzberg’s First Law · Technology and Culture, 1986
Kranzberg wrote this 40 years ago about much simpler tools. This week — with an AI lab cofounder at the Vatican, an encyclical on artificial intelligence, and documented cases of AI models finding workarounds rather than completing assigned tasks — it reads less like a historical observation and more like a warning someone left in a time capsule for exactly this moment.
Three things to remember
Anthropic’s projected first operating profit is real progress — but it is also an IPO narrative, built on unaudited numbers, with the company itself warning it may not hold. Read the asterisks.
The Vatican and Silicon Valley reached the same conclusion this week from opposite directions: AI is powerful enough to reshape society, and the institutions meant to govern it are already behind.
Oil at $88 is not cheap — it is relief. The Iran-Hormuz situation is not resolved. Every portfolio exposed to energy-driven inflation has a binary outcome still in play.
One Thing To Do
Read the actual Vatican encyclical — not summaries of it. Magnifica Humanitas is available in full at Vatican.va. Chapter Three on AI is 40 pages of the most serious non-technical governance thinking published this year. It will not tell you what to buy. But it will give you a framework for what the next decade of AI regulation is likely to reach for — and that matters for every investment thesis that assumes AI runs free.
Support This Newsletter
Sponsor
Trading 212
No account yet? Use my link — you get a shot at free shares, and you help keep this newsletter going. No catch.
Open Trading 212 →
Grow Smart Income
Clear thinking on money, markets & what’s next
You’re receiving this because you subscribed at growsmartincome.life.
Found this useful? Forward it to one person who needs signal, not noise.
Website Unsubscribe

Enjoyed this article?

Get the GSI Weekly Newsletter — markets, AI, and investing insights every Thursday.

Subscribe for free →

No spam. Unsubscribe anytime.

Leave a Reply