the Now & the Next

A bi-weekly speculative fiction suggesting the shape of things to come.
(sourced from trustworthy trade pubs, think tanks + frontier science news)

300+ Signals Tracked
7 Collisions Detected
14 Industries
50+ Publications
2 weeks Signal Window

Three hundred signals from the first half of March 2026 — and seven collisions that tell you what none of them could alone. From the agentic commerce wars to the physical AI factory floor, here is the Now and the Next.

01
The Agentic Commerce Meltdown

When the checkout lane becomes a battlefield between bots

+

Amazon secured a court order blocking Perplexity's AI shopping bot from making purchases on its platform — the first legal shot fired in a war over who controls the buy button. Days later, OpenAI quietly killed Instant Checkout inside ChatGPT, retreating from the very feature it had launched with fanfare. Stripe responded by open-sourcing the Agentic Commerce Protocol, a universal checkout spec designed for transactions between buyers, AI agents, and sellers. Google published documentation for its Universal Commerce Protocol, enabling native checkout directly on Google properties. Estée Lauder's Georgina Cunningham warned that AI shopping agents are fundamentally reshaping how consumers discover products. And Criteo became the first ad tech company integrated into ChatGPT's advertising pilot. The buy button used to be a settled question. Now every major platform is fighting over who owns the transaction layer in a world where the shopper might not be human.

⚡ The Now

The commerce layer of the internet is being renegotiated in real time. Amazon is suing to keep bots off its shelves. OpenAI is retreating from direct checkout. Stripe and Google are racing to define the protocol layer. The old assumption — that a human clicks 'buy' on a retailer's website — is breaking down faster than anyone expected.

→ What's Next

Commerce becomes API-first. Within 18 months, every major retailer will need an agent-facing checkout API alongside its human-facing storefront. The winners will be the platforms that define the protocol standard — Stripe's ACP and Google's UCP are the early candidates. Brands that ignore agent-readiness will find themselves invisible to the fastest-growing category of shoppers: the ones that never open a browser.

Retail Brew
Amazon secured a court order blocking Perplexity's Comet browser agent from making purchases and accessing customer data on its platform.
The Drum
OpenAI discontinued in-app checkout within ChatGPT, reflecting challenges in integrating fragmented retailer systems.
Stripe Blog
Stripe launched the Agentic Commerce Protocol, an open checkout spec enabling transactions between buyers, AI agents, and sellers via APIs.
MarTech
Google documented its Universal Commerce Protocol enabling native checkout flows directly on Google properties.
The Drum
Estée Lauder highlights how AI shopping agents and the creator economy are fundamentally transforming product discovery.
AdExchanger
Criteo became the first ad tech partner integrated into OpenAI's advertising pilot for ChatGPT.
02
The Health AI Land Grab

When Big Tech, pharma, and startups all reach for the stethoscope at once

+

Microsoft launched Copilot Health, a dedicated AI chatbot that lets consumers upload medical records and get personalized health insights. The same week, Amazon expanded its health AI assistant beyond One Medical to all U.S. consumers — and separately launched Amazon Connect Health, a suite of agentic tools to automate patient triage and scheduling. CVS Health announced Health100, a consumer engagement platform built with Google Cloud that aggregates data across pharmacy, insurance, and clinical touchpoints. Abridge shipped a feature letting clinicians edit medical notes with natural language prompts. And a study published in Nature Medicine warned that OpenAI's ChatGPT Health service often fails to recommend emergency care when urgently needed. Five companies, five different entry points, one patient, and so far, nobody coordinating.

⚡ The Now

Every major tech platform now has a consumer-facing health AI product. Microsoft owns the personal health chatbot. Amazon owns the front door to scheduling. CVS is building the cross-system aggregation layer. The race is no longer whether AI enters healthcare — it's who controls the patient relationship. Meanwhile, safety research is struggling to keep pace with deployment.

→ What's Next

Health AI converges on the patient record. Within two years, whoever integrates the most data sources — EHRs, pharmacy, wearables, insurance — becomes the default health interface for consumers. The risk: fragmented AI health products that each see a slice of the patient but none see the whole person. Interoperability becomes the trillion-dollar question.

Healthcare Dive
Copilot Health lets users upload medical records and health histories for personalized AI-driven health insights.
Healthcare Dive
Amazon's health AI assistant, initially for One Medical members, is now available to all U.S. consumers.
Healthcare Dive
Amazon Connect Health deploys agentic AI for patient identity verification, appointment scheduling, and clinical triage.
Healthcare Dive
Health100 aggregates patient data from pharmacy, insurance, and clinical touchpoints into a single AI-powered platform.
Abridge
Clinicians can now edit medical notes using natural language prompts, shifting documentation from manual revision to guided AI assistance.
Computerworld
Nature Medicine study finds ChatGPT Health often fails to recommend emergency care when urgently needed.
03
The Optical Backbone Bet

When the AI buildout hits the speed of light — literally

+

Nvidia invested $2 billion each in optical component makers Lumentum and Coherent, placing a $4 billion wager that photonics — not copper — will connect the next generation of AI clusters. Cisco introduced the Open Transport 3000 Series, a multi-rail open line system designed to link distributed AI workloads across multiple data centers. HPE completed its $14 billion acquisition of Juniper Networks, tripling its networking revenue and positioning itself against Cisco and Huawei in the AI infrastructure race. NTT Data secured 115 MW of new data center capacity for hyperscale AI workloads. Nvidia and Thinking Machines Lab signed a multiyear partnership to deploy at least one gigawatt of AI compute. And the biggest tech companies — Google, Amazon, OpenAI — signed a White House pledge to cover the cost of new energy infrastructure their data centers require. The bottleneck in AI is moving from compute to connectivity and power, and the checks being written to solve it are measured in billions.

⚡ The Now

The AI infrastructure race has shifted from GPUs to the physical layer beneath them. Optical networking, power capacity, and data center real estate are the new constraints. Nvidia is vertically integrating into photonics. HPE is buying its way into networking scale. NTT is building out megawatt-class capacity. The supply chain for intelligence now looks like an energy and telecom buildout.

→ What's Next

AI infrastructure becomes a sovereign asset class. Countries and companies that control optical interconnect, power capacity, and data center real estate will set the terms of AI access. Within 24 months, expect optical networking companies to be valued like semiconductor firms, and energy contracts to be the hidden variable in every AI roadmap.

Light Reading
Nvidia invested $2 billion each in Lumentum and Coherent, signaling that optical components are critical to next-gen AI cluster connectivity.
Network World
Cisco's Open Transport 3000 adds multi-rail optical networking for distributed AI workloads across data centers.
Light Reading
HPE completed the $14 billion Juniper Networks acquisition, tripling networking revenue and challenging Cisco and Huawei.
Data Center Knowledge
NTT Data secured 115 MW of new data center capacity from hyperscale and enterprise customers across three U.S. campuses.
NVIDIA Blog
Multiyear partnership to deploy at least one gigawatt of Nvidia Vera Rubin AI systems for advanced model training.
E&T
Google, Amazon, and OpenAI pledged to cover all costs of new energy infrastructure required for their expanding data centers.
04
The Agentic Enterprise Arrives

When every productivity platform ships a coworker you never interviewed

+

Microsoft launched Copilot Cowork — built with Anthropic — a cloud-powered AI agent that works autonomously across M365 apps in the background while you do other things. Zoom expanded its agentic AI platform to automate workflows across Workplace, Phone, and CX with no-code agent builders. Salesforce launched Agentforce Contact Center, a native CCaaS solution integrating voice, digital channels, CRM, and AI agents in a unified system. Microsoft also announced a strategy of letting enterprises 'hire' AI agents via subscription — agents that act as continuous team members, not passive tools. And Gartner reported that agentic AI's first enterprise impact will be augmenting middle managers by automating the administrative burden that eats half their workweek. The enterprise software stack just gained a new default employee, and it came preinstalled in the tools companies already pay for.

⚡ The Now

Every major enterprise platform — Microsoft, Zoom, Salesforce — now ships autonomous agents as a built-in feature. These aren't chatbots answering questions. They're background workers completing tasks across applications without human prompting. The shift from 'copilot' to 'coworker' happened in a single product cycle.

→ What's Next

Agent density becomes a competitive metric. Within 18 months, enterprises will measure agent-to-employee ratios the way they once measured laptop-to-employee ratios. The companies that figure out agent orchestration — which agents talk to which, who has authority over what — will outperform those still debating whether to buy the license. The org chart just got a species it never planned for.

VentureBeat
Cloud-powered AI agent built with Anthropic that works autonomously across M365 apps in the background.
Computerworld
Zoom introduces no-code agent builders and autonomous workflow automation across its collaboration suite.
No Jitter
Native CCaaS solution integrating voice, digital channels, CRM data, and AI agents in a unified system.
Computerworld
Microsoft shifts strategy from passive Copilot tools to autonomous AI agents that enterprises 'hire' as continuous team members.
HR Executive
Gartner highlights agentic AI's first enterprise impact: automating the routine administrative tasks that consume half of middle managers' time.
05
The Workforce Inversion

When the machines get hired and the humans get the severance package

+

Atlassian cut 1,600 jobs — 10% of its workforce — explicitly to fund AI development and enterprise sales expansion. Across the broader tech sector, layoffs surpassed 45,000 in early 2026, with Amazon, Meta, and Block all citing AI as a factor. A federal judge allowed key claims to proceed in a discrimination lawsuit against Workday, ruling that AI-driven hiring tools can be held liable under age discrimination law. Gartner surveyed supply chain leaders: 55% expect agentic AI to reduce entry-level hiring, and 51% expect overall workforce reductions. And Brookings called for a national AI adoption platform to maintain the U.S. edge in AI without leaving workers behind. The same companies deploying AI agents as 'coworkers' are simultaneously reducing their human headcount to pay for them.

⚡ The Now

The relationship between AI investment and workforce reduction has become explicit. Companies are no longer pretending these are unrelated decisions. Atlassian cut humans to fund AI. Amazon and Meta cited AI directly in layoff announcements. The Workday lawsuit means AI hiring tools now carry legal liability for bias. The social contract around AI and employment is being rewritten in real time.

→ What's Next

Entry-level roles face the steepest displacement curve. The Gartner data is blunt: the jobs most vulnerable to agentic AI are the ones that used to be the on-ramp to a career. Within two years, companies will need to redesign career ladders entirely — because the first three rungs are being automated. Expect regulatory pressure, retraining mandates, and at least one major class-action suit over AI-driven hiring bias to reshape the landscape.

Computerworld
Atlassian cut 10% of its workforce, explicitly reallocating capital toward AI development and enterprise sales.
Network World
Over 45,000 tech industry layoffs in early 2026, driven by strategic shifts toward AI and automation investments.
SHRM
Federal judge ruled AI-driven hiring tools can be held liable under age discrimination law, rejecting Workday's defense.
Consumer Goods Technology
Gartner survey: 55% of supply chain leaders expect agentic AI to reduce entry-level hiring, 51% expect overall workforce cuts.
Brookings
Brookings advocates for a national AI adoption platform with increased funding for worker retraining and regional innovation.
06
Physical AI Hits the Factory Floor

When the robots stop simulating and start shipping

+

ABB Robotics partnered with Nvidia to integrate Omniverse simulation libraries into RobotStudio, creating industrial-grade digital twins that let manufacturers train robots in simulation before deploying them on real production lines. NEURA Robotics and Qualcomm entered a long-term collaboration to build cognitive robotics platforms combining edge AI with human-like dexterity. Samsung Electronics announced the adoption of AI-driven factories across its manufacturing operations. Figure AI released Helix 02, a humanoid robot that autonomously tidied a living room using a single neural system controlling full-body movement from raw sensor input. And Nvidia deployed open models on its Jetson Thor platform for real-time edge AI inference in industrial and robotic systems. The common thread: simulation-to-production timelines are collapsing, and the gap between research demo and factory deployment is now measured in months, not years.

⚡ The Now

Industrial robotics is crossing the simulation-to-production threshold. ABB and Nvidia are making digital twin training production-ready. Samsung is retrofitting entire factories around AI. Figure AI's humanoid handles unstructured environments with a single neural controller. The lab demo era for physical AI is over.

→ What's Next

The factory of 2028 runs on physical AI by default. As simulation-to-deployment pipelines mature, the cost of deploying a new robotic task drops from a multi-month integration project to a training run. Qualcomm's edge AI chips will put cognitive capabilities on robots that cost as much as a car, not a house. The labor economics of manufacturing are about to get rewritten alongside the white-collar workforce.

The Manufacturer
ABB integrates Nvidia Omniverse into RobotStudio, enabling industrial-grade digital twin training for production robotics.
Qualcomm
Long-term partnership combining Qualcomm's edge AI silicon with NEURA's cognitive robotics for next-gen physical AI platforms.
Assembly Magazine
Samsung commits to AI-driven manufacturing across its factories for process automation, quality control, and predictive maintenance.
Figure AI
Humanoid robot autonomously tidies a living room using a single neural system controlling full-body dexterity from raw sensor input.
NVIDIA Blog
Nvidia deploys open AI models on Jetson Thor for real-time, low-latency inference directly on industrial and robotic systems.
07
The Personalization Flywheel

When every shelf, screen, and sip is tuned to a market of one

+

Colgate-Palmolive deployed AI-powered data clean rooms, digital twins, and personalized promotions to globalize product success across its brands. PepsiCo scaled generative AI to create personalized Gatorade bottle designs, driving a measurable surge in cross-sells and subscriptions. Costco reported that digital personalization — product recommendation carousels and modernized display pages — drove $470 million in e-commerce sales growth in fiscal 2025. Walmart began a chainwide rollout of digital shelf labels, enabling centralized, dynamic pricing updates across every U.S. store. And Spotify launched Taste Profile, letting Premium users see and shape how the platform understands their preferences. From toothpaste to sports drinks to streaming, the personalization stack is going live at scale across industries that have historically treated customers as segments, not individuals.

⚡ The Now

Mass personalization is crossing from pilot programs to operating-at-scale. Costco can quantify it: $470 million. PepsiCo is generating custom product designs with AI. Walmart is digitizing every shelf tag in the country. The personalization thesis has been talked about for a decade, but the infrastructure — data clean rooms, generative AI, dynamic pricing — is finally in place to deliver it.

→ What's Next

The default customer experience becomes personalized. Within two years, a generic shelf display or static product recommendation will feel as outdated as a print catalog. The companies that build closed-loop personalization — where the AI learns from purchase data and feeds it back into product design, pricing, and placement — will compound their advantage every cycle. The rest will be selling to averages in a world that stopped being average.

Consumer Goods Technology
Colgate-Palmolive leverages AI data clean rooms, digital twins, and personalized promotions to scale product success globally.
Consumer Goods Technology
PepsiCo uses generative AI to create personalized Gatorade bottle designs, boosting consumer engagement and cross-selling.
Grocery Dive
Costco's personalized product recommendations and modernized display pages drove over $470 million in e-commerce sales growth.
Chain Store Age
Walmart rolling out digital shelf labels across all U.S. stores, enabling centralized dynamic pricing and eliminating paper tags.
Spotify Newsroom
Spotify's Taste Profile lets Premium users see and directly shape how the platform understands their music and podcast preferences.

Frontier Science Feeding the Machine

The research signals underneath the enterprise headlines — breakthroughs in compute, biology, security, and the physical sciences that will shape the next wave of commercial AI.

AI Security
Claude Opus 4.6 independently discovered 22 security vulnerabilities in Firefox over two weeks, including one critical flaw missed by existing audits.
Encrypted Computing
Intel's Heracles chip accelerates fully homomorphic encryption 5,000x faster than current server CPUs, enabling computation on encrypted data without decryption.
Medical Imaging
AI method analyzes standard mammograms to quantify arterial calcification, predicting cardiovascular disease risk from routine breast cancer screenings.
Clinical AI
Prospective clinical study of Google's AMIE conversational AI system for clinical history-taking before physician consultations.
Drug Discovery
Machine learning tool predicts chirality outcomes of chemical reactions, making drug synthesis faster and more efficient.
Agent Memory
Microsoft Research introduces a memory system that restructures raw interaction histories into compact, reusable knowledge units for AI agents.
Mathematics
Agentic AI systems are generating rigorous mathematical proofs from high-level prompts, accelerating research that previously took months.
Labor Economics
Anthropic researchers developed 'observed exposure' metric combining theoretical LLM capabilities with real-world AI usage data to assess job displacement risk.
Custom Silicon
Meta deployed four generations of its homegrown MTIA AI inference chips in two years, optimized for ranking and recommendation at scale.
Neuroscience
Carnegie Mellon researchers developed transcranial electro-acoustic stimulation, a noninvasive method combining ultrasound with electrical stimulation for brain modulation.