This was the week agentic commerce stopped being a conference slide and started colliding with physics, politics, and payroll. We published four pieces that, taken together, tell a single story: the AI-powered future everyone is racing to build has dependencies that nobody wants to talk about.
The payments industry spent Q1 debating protocols. This week, the conversation got uncomfortable.
Here is what we covered.
Block Cut 4,000 Jobs and Blamed AI
Jack Dorsey halved Block’s workforce and told investors most companies would follow within a year. The stock surged 24 percent. The narrative was clean: AI makes companies leaner.
We looked past the narrative. Bloomberg called it “AI-washing.” Former employees called it organisational bloat in costume. An Oxford Economics report found that many AI-attributed layoffs are actually corrections from pandemic-era over-hiring. Forrester questioned whether the productivity gains are even real.
For payments and fintech companies that handle other people’s money, the question is not whether AI can replace roles. It is whether cutting compliance, risk, and operations teams in the name of transformation creates gaps that no model can fill.
The Agentic Commerce Standards Race
Mastercard launched Verifiable Intent. Google expanded the Universal Commerce Protocol. OpenAI shipped the Agentic Commerce Protocol. Cloudflare built Web Bot Auth. Stripe released Shared Payment Tokens. Five competing frameworks, all announced within weeks of each other.
We mapped the full competitive landscape and identified the real battle underneath the announcements: this is not about which protocol wins. It is about whether the trust infrastructure beneath AI-driven commerce gets built as an open standard or a proprietary moat.
The companies that set the rules will set the market. Everyone else will comply.
The AI Fraud Paradox
Visa found that 80 percent of global consumers were targeted by a scam attempt in the past year. Mastercard is investing in AI-powered identity verification. Cloudflare is building bot authentication. Meanwhile, deepfakes, synthetic identity fraud, and AI-powered social engineering are scaling at the same rate as the defences.
We explored the uncomfortable symmetry: AI is simultaneously the biggest fraud threat in payments history and the primary tool being deployed to fight it. The companies selling AI fraud detection are often the same ones whose platforms enabled the attack vectors. The arms race has no finish line.
The Strait That Powers the AI Economy
This was our deep dive for the week, and the one that connected everything.
The Iran conflict is not just a geopolitics story. It is a payments infrastructure story. The Strait of Hormuz handles a fifth of global oil and LNG. Traffic dropped 70 percent. South Korea imports 97 percent of its energy through those waters. Samsung and SK Hynix fabs cannot run without it. Global DRAM inventory sits at two to three weeks. No buffer.
Seventeen submarine cables run through the Red Sea carrying the majority of Europe-Asia-Africa data traffic. Qatar supplies helium essential for chip fabrication. LNG spot prices jumped 50 percent. Defence production is diverting rad-hardened chips from commercial use.
Every agentic commerce protocol being built right now assumes abundant, cheap compute. Every AI fraud model requires continuous training at scale. Every real-time payment system runs on cloud infrastructure that consumes chips and energy in growing quantities. The Iran conflict is stress-testing all of those assumptions simultaneously.
What Connects the Week
Strip the headlines away and a pattern emerges.
Block cut 4,000 jobs betting that AI can replace human judgment in a regulated industry. The standards race is a fight over who controls the trust layer beneath AI commerce. The fraud paradox shows that AI creates as many vulnerabilities as it solves. And the Hormuz piece revealed that the physical infrastructure underneath all of it is more fragile than anyone in the industry is acknowledging.
The common thread: the gap between ambition and infrastructure. Between what the models can do in a demo and what the systems can sustain at scale, under pressure, in the real world.
The AI economy is being built on assumptions that have not yet been tested. This week, some of them were.
Coming Next Week
Prediction markets are no longer just a niche crypto curiosity. Kalshi has partnered with Rotten Tomatoes to embed real-time betting data into Oscars coverage. Polymarket partnered with the Golden Globes. DraftKings is launching a prediction markets super app. Oscar-related contracts have surpassed $48 million in trading volume.
We are working on a piece that examines how prediction markets are systematically building distribution partnerships with media institutions, turning cultural events into betting infrastructure, and what that means for how we consume entertainment and news.
Watch for it this week.
The AI economy met the real economy this week. Which assumptions held up, and which ones cracked?