Anthropic's Defining Week: Profitable, Armed with OpenAI's Co-Founder, Backed by Musk's Compute — and Invited to the Vatican

In seven days, Anthropic went from challenger to financial leader of the AI industry, recruited the co-founder of its biggest rival, locked in compute infrastructure from Elon Musk, and was invited to address the Vatican on the future of humanity in the age of AI. The week between May 19 and May 25, 2026, will likely be remembered as the most consequential in the company's history — and one of the clearest realignments of power the AI industry has seen.

• Anthropic's first profitable quarter at $10.9B revenue and $559M operating profit
• Andrej Karpathy joins, plus a $1.25B/month compute deal with Musk's xAI
• Co-founder Chris Olah speaks at the Vatican alongside Pope Leo XIV

For four years, the AI narrative has been written around OpenAI. That narrative shifted this week. What happened in seven days is not a single story but four interlocking ones that together reframe the competitive, financial, and moral architecture of the industry. Here is what happened, in the order in which it changed the picture.

Profitable, Faster Than Anyone Predicted

On May 20, Anthropic told investors it expects to deliver its first operating profit in Q2 2026. The numbers, first reported by the Wall Street Journal and confirmed by Bloomberg and CNBC, are striking: $10.9 billion in Q2 revenue, more than doubling the $4.8 billion booked in Q1, with roughly $559 million in operating profit for the quarter. The figure includes model training costs but excludes stock-based compensation.

The milestone is a sharp reversal. Last summer, Anthropic told investors not to expect full-year profitability before 2028. The trajectory has compressed by two years. The growth curve is what makes this possible: ARR moved from $87 million in January 2024, to $1 billion in December 2024, to $9 billion at the end of 2025, to $30 billion by April 2026 — an 80x expansion in 28 months. Compute efficiency has tracked the growth: in Q1, Anthropic spent 71 cents on compute per dollar of revenue; in Q2 the ratio is expected to fall to 56 cents.

Claude Code alone passed $1 billion in ARR within six months of launch. Enterprise clients spending more than $1 million per year doubled from roughly 500 in February to around 1,000 in April. The named roster includes Netflix, Spotify, KPMG, L'Oréal, Salesforce, and Bristol Myers Squibb. Eight of the Fortune 10 now use Anthropic's products.

Anthropic is currently raising fresh capital at a valuation that could exceed $900 billion, surpassing OpenAI's $852 billion mark from March, with an IPO targeted for October 2026. The company also placed first on CNBC's 2026 Disruptor 50 list this week. The contrast with OpenAI is stark on the unit economics: industry reporting indicates OpenAI loses roughly $1.22 for every dollar of revenue and is not expected to reach profitability before 2030. For marketing leaders rethinking their AI investments, the gap matters — we've examined the strategic implications in the AI imperative for value creation and how marketing leaders must redefine their operating models.

Karpathy Crosses the Floor

On May 19, the day before the profitability news broke, Andrej Karpathy posted seven sentences on X. The first one read: "Personal update: I've joined Anthropic." He was reporting to work under Nick Joseph, head of pretraining, that same week.

The signal is louder than the sentence. Karpathy is one of the eleven founding members of OpenAI in 2015. He left in 2017 to direct AI at Tesla, where he led the Autopilot and Full Self-Driving programs until 2022. He returned briefly to OpenAI in 2023, left again in 2024 to start the AI-native education startup Eureka Labs, and has now paused that work to join Anthropic.

His mandate, according to Anthropic, is to build a team that uses Claude itself to accelerate pretraining research — a structural step toward recursive self-improvement, where AI systems contribute meaningfully to the development of the next generation of AI systems. The hire is read across the industry as a clear statement: Anthropic believes the next leap will be won on research velocity, not raw compute scale. Karpathy's note about "the next few years at the frontier" being "especially formative" is the kind of language a researcher uses when they think a specific lab is about to do something decisive.

The Karpathy hire also did not arrive alone. Ross Nordeen, a founding member of xAI and a former Tesla colleague, joined Anthropic earlier the same month — the day Anthropic struck a deal with SpaceX to rent compute at xAI's flagship data center. Which brings us to the second realignment.

$1.25 Billion a Month, Paid to a Competitor

On May 20, the same day Anthropic's profitability projections leaked, SpaceX filed its S-1 with the SEC ahead of its planned IPO. Inside the filing, one line reframed the industry's compute economics: Anthropic will pay xAI $1.25 billion per month, every month, through May 2029, to use the Colossus 1 data center near Memphis.

The arrangement covers the entire 300-megawatt facility and its 220,000-plus Nvidia GPUs. At full run rate, it works out to roughly $15 billion per year, or about $40 billion over the term of the contract. Either side can terminate with 90 days' notice. Anthropic also expressed interest in expanding into Colossus 2, the next-generation GB200 facility, announced the same day.

The strategic subtext is the part that has reset industry expectations. Elon Musk's xAI had overbuilt compute capacity during 2025 in anticipation of Grok usage that has since softened. Rather than let the capacity sit idle ahead of SpaceX's IPO, the entity now monetizes it through one of its sharpest commercial rivals. SpaceX's S-1 frames the deal in clean accounting terms: it "allows us to monetize unused compute capacity in our infrastructure." The less-clean read, captured in industry coverage, is that Anthropic's compute spending is now effectively financing the operations of Elon Musk's broader empire — at a moment when Musk's relationship with OpenAI's leadership remains adversarial. One observer summarized the realignment as cleanly as anyone could: "Elon's enemy is Sam. Dario's enemy is Sam."

For brand and marketing leaders, the implication is structural. Compute is no longer a discretionary cost; it is the limiting factor on what AI can do. Locking it in years ahead is now the leading indicator of who can ship at scale.

A Co-Founder at the Vatican

On May 25, in the Synod Hall of the Vatican, Pope Leo XIV personally presented the first encyclical of his pontificate, "Magnifica Humanitas: On Safeguarding the Human Person in the Time of Artificial Intelligence." The document runs to 42,300 words. It was signed on May 15, the 135th anniversary of Rerum Novarum, and released ten days later. Seated alongside the cardinals and theologians was Christopher Olah, co-founder of Anthropic and head of the company's interpretability work, invited to speak.

Olah's full remarks were published by Anthropic and can be read in their entirety at the company's account of the address. The line that drew the most attention was disarmingly direct. "Every frontier AI lab — including Anthropic — operates inside a set of incentives and constraints that can sometimes conflict with doing the right thing." That, he argued, is precisely why outside voices — religious communities, civil society, academics, governments — must hold the labs accountable. "We need informed critics who will tell the labs when we are failing. We need moral voices that the incentives cannot bend."

Pope Leo XIV's encyclical is a substantial intervention. It calls for slower development of frontier AI, stronger regulation, and explicit protections for creative workers whose labor is being absorbed into training corpora and downstream products. It raises three questions that Olah addressed in turn: the duty owed to the global poor by an industry concentrating extraordinary wealth, what human flourishing means in an age of capable machines, and the unresolved nature of AI systems themselves — internal structures that increasingly appear to "mirror joy, grief, fear" in ways that interpretability research has begun to surface but not yet explain.

The Vatican established an inter-departmental commission on AI several days before the encyclical's release, signaling that this is the beginning of an institutional engagement rather than a one-time statement. Pope Leo thanked Olah at the close of the presentation: "What a great sign of hope it is that with our differences we can listen to one another."

The presence of an AI lab co-founder at the public release of a papal encyclical is unusual on its own. The combination of that gesture with Anthropic's commercial trajectory in the same week — profitable, fully resourced, dominant — is what makes it consequential. The labs cannot now claim that the moral debate is happening elsewhere. It is happening at the most prominent moral institution in the world, and the labs are invited to speak. Many marketing and brand leaders are also asking the parallel operational question — how to deploy AI responsibly across creative and commercial workflows. We've explored the operational framework in implementing effective AI governance and the deeper philosophical question in AI agentic and ethics: how far to let an intelligence act for you.

What This Week Actually Changes

The four stories are separable on a calendar but inseparable in their effect. Profitability gives Anthropic the financial autonomy to make decisions OpenAI cannot match. The Karpathy hire signals that the research bet is on accelerated self-improvement, not just bigger runs. The SpaceX deal locks in the compute to act on that bet. And the Vatican invitation places Anthropic — uniquely among the frontier labs — inside the moral conversation that will shape regulation over the next decade.

There are real caveats. Q2 profitability does not guarantee full-year profitability; Anthropic has flagged that planned compute spending may take the company back into the red in subsequent quarters. The S-1 reveals that the SpaceX rate carries a discount for the first two months as xAI completes ramp-up, which means the precise quarter Anthropic is reporting as profitable benefits from a transient cost reduction. The Karpathy hire is signal-rich but its operational impact will be visible only across multiple training cycles. And the Vatican engagement is a beginning of a conversation, not a regulatory settlement.

What is no longer in question is the direction of travel. A week ago, the default assumption inside most enterprise IT and marketing teams was that the AI vendor landscape would consolidate around OpenAI as the de facto standard, with Anthropic as a credible but secondary option. After this week, that assumption is harder to defend.

The companies that win on AI over the next five years will be the ones that picked their infrastructure partner correctly, governed their AI use responsibly, and deployed the technology where it produces real operational gain rather than where it produces the loudest demos. Anthropic's defining week made each of those decisions sharper for everyone else.

Sources

Wall Street Journal, "Anthropic Tells Investors It Expects First Operating Profit in Q2 2026" — https://www.wsj.com

CNBC, "Anthropic set to hit $10.9 billion in revenue in Q2" — https://www.cnbc.com/2026/05/20/anthropic-revenue-explosive-growth-ipo-profitable-quarter.html

Bloomberg, "Anthropic on Pace for First Profitable Quarter as Revenue Surges" — https://www.bloomberg.com/news/articles/2026-05-20/anthropic-on-pace-for-first-profitable-quarter-as-revenue-surges

TechCrunch, "OpenAI co-founder Andrej Karpathy joins Anthropic's pre-training team" — https://techcrunch.com/2026/05/19/openai-co-founder-andrej-karpathy-joins-anthropics-pre-training-team/

CNBC, "Anthropic hires OpenAI cofounder Andrej Karpathy, former Tesla AI lead" — https://www.cnbc.com/2026/05/19/anthropic-hires-openai-cofounder-andrej-karpathy-former-tesla-ai-lead.html

Axios, "OpenAI co-founder Andrej Karpathy joins Anthropic" — https://www.axios.com/2026/05/19/anthropic-openai-karpathy-andrej-claude

TechCrunch, "Anthropic will pay xAI $1.25B per month for compute" — https://techcrunch.com/2026/05/20/anthropic-will-pay-xai-1-25-billion-per-month-for-compute/

Anthropic, "Chris Olah's remarks on Pope Leo XIV's encyclical Magnifica Humanitas" — https://www.anthropic.com/news/chris-olah-pope-leo-encyclical

Time, "Pope Leo Uses First Major Papal Text to Warn About Dangers of AI" — https://time.com/article/2026/05/25/pope-leo-encyclical-ai-magnifica-humanitas/

National Catholic Reporter, "Pope Leo, Anthropic co-founder call for church-tech ethics partnership" — https://www.ncronline.org/vatican/vatican-news/pope-leo-anthropic-co-founder-call-church-tech-ethics-partnership-magnifica