ISSUE 034 · JUNE 28, 2026
TIMPS PostCards
AI Intelligence Digest · Your Daily Signal from the Machine Frontier
Hyderabad, IndiaVol. 03 · No. 0346 Lead Stories
⬡ Built with the TIMPS Ecosystem
Humanoid IPOsChip ScalingBiosecurity Bet
01 · LEAD
Markets · Robotics
Wall Street Gets Its First Pure-Play Humanoid Robot Stock
Agility Robotics is taking its warehouse-tested Digit robot public through a $2.5 billion SPAC — and Amazon, Nvidia and SoftBank are already on the cap table.
$2.5BSPAC Valuation

Agility Robotics announced a merger with Churchill Capital Corp XI on June 24, valuing the Oregon-based humanoid robotics company at roughly $2.5 billion and generating more than $620 million in proceeds once the deal closes. Once listed, it will trade under the ticker AGLT — the first US-listed pure-play humanoid robotics company.

Unlike rivals still largely in demo mode, Agility's robot Digit has logged more than 65,000 commercial operating hours, holds the industry's first humanoid Robots-as-a-Service contract with GXO Logistics, and signed a production agreement with Toyota in February. The filing cites over $300 million in committed multi-year Digit v5 orders, though that figure depends on contractual milestones tied to a single undisclosed customer.

The new Digit v5 integrates Nvidia's Halos safety architecture — built on more than 18,600 engineering-years of autonomous-vehicle safety work — letting the robot operate near people without workcell barriers for the first time. CEO Peggy Johnson has turned down outside acquisition interest, betting that real deployment data, not just funding size, is what separates Agility from better-capitalized rivals like Figure AI and Boston Dynamics.

Full story · Tech Times
02 · M&A
Creative AI · Acquisitions
Adobe Buys Topaz Labs to Bring On-Device AI Into Creative Cloud
An Emmy-winning specialist in AI upscaling and restoration is joining Firefly — handing Adobe sharper tools just as Canva and Blackmagic push harder into AI editing.

Adobe signed a definitive agreement on June 25 to acquire Topaz Labs, the Texas-based company behind Topaz Photo, Video, Gigapixel, Astra and Bloom — AI models used by 20 of the world's 50 largest companies for upscaling, denoising, stabilization and footage restoration. The deal is expected to close in the second half of 2026; terms were not disclosed.

Topaz's edge is "Neurostream" technology that runs large AI models locally on consumer devices rather than in the cloud, a capability Adobe says will let it deliver faster, cheaper AI editing inside Photoshop, Lightroom and Premiere without per-app subscriptions. The two companies already partnered in October 2025 to fold Topaz's upscalers into Photoshop's AI credit system.

Topaz CEO Eric Yang, who will lead the team inside Adobe, called the move a way to "dramatically expand what's possible for filmmakers and creators everywhere." The acquisition follows Adobe's broader push toward agentic creative tooling, including deepening partnerships with Microsoft and Anthropic.

Full story · TechCrunch
03 · HARDWARE
Research · Semiconductors
IBM Squeezes 100 Billion Transistors Onto a Fingernail-Sized Chip
A new "NanoStack" 3D transistor architecture pushes past the 1-nanometer wall by building upward instead of shrinking sideways.
0.7nmNanoStack Node

IBM unveiled what it calls the world's first sub-1-nanometer chip technology on June 25 — a transistor architecture at the 0.7nm, or 7-angstrom, node that stacks transistors vertically rather than continuing to shrink them along a flat plane. The design packs nearly 100 billion transistors onto a chip roughly the size of a fingernail, almost double the density of IBM's 2nm generation.

IBM projects the architecture could deliver up to 50% higher performance or 70% better energy efficiency than its current 2nm chips, alongside a 40% reduction in SRAM area — memory circuitry already heavily used in AI accelerators from Nvidia and Cerebras. Research director Jay Gambetta framed it as reinventing how chips are built, not just making smaller parts.

IBM doesn't manufacture leading-edge chips itself and hasn't named a commercialization partner, with production expected within five years. Analysts caution that's roughly five GPU design cycles away — but the research keeps IBM relevant to a race increasingly defined by power efficiency, not just raw density, as AI data centers strain power grids.

Full story · IBM Newsroom
04 · BIOSECURITY
Research · Philanthropy
Anthropic, OpenAI and Stripe Bet $500M That the Common Cold Is Optional
A new nonprofit called Intercept wants to fund the science and air-cleaning tech needed to make respiratory infections a thing of the past.
$500MInitiative Size

Stripe launched Intercept on June 24, a $500 million philanthropic initiative backed by Anthropic, the OpenAI Foundation, Bill Gates, Flu Lab and several Jane Street traders, aimed at radically reducing — and eventually eliminating — respiratory infections like the common cold and flu.

The plan funds two tracks: broad-spectrum preventatives such as nasal sprays that catch viruses before infection, and large-scale air-cleaning technology including far-UVC light and filtration for offices, schools and airports, modeled on how municipal systems clean drinking water. Stripe's Nan Ransohoff, who leads the effort with venture capitalist Charlie Petty, says corporate partners including Warby Parker, Mastercard and JPMorgan have already signed on as pilot sites.

Respiratory viruses kill roughly 1 million people a year and cost the global economy an estimated $600 billion annually, according to Intercept's own figures. Critics on social media questioned the air-cleaning approach's safety for children, while public-health researchers noted Phase 1 and 2 trials alone typically cost $20–30 million per drug — a reminder of how far $500 million has to stretch.

Full story · MIT Technology Review
05 · PRODUCT
Design Tools · AI Agents
Figma Stops Calling AI a Feature and Starts Calling It Material
At Config 2026, Figma shipped motion, 3D, shaders and code layers directly onto the canvas — and a design agent built to use all of it.

Figma used its June 23–25 Config conference to reposition itself against the entire creative and coding stack at once. CEO Dylan Field argued that "AI has lowered the floor" on production speed "but it has not raised the ceiling" on quality — the rationale behind a wave of new canvas-native materials: Figma Motion for animation, WebGPU shaders, 3D transforms, and code layers that pull live GitHub repositories directly into design files.

The Figma design agent, now rolled out to everyone, can build custom "skills" from team workflows, connect to outside tools like Notion, Slack and GitHub, and generate shaders or motion from a text prompt. Twenty-plus generative image workflows from Figma's Weave acquisition — style transfer, 3D renders, consistent characters — are now built into the left panel as one-click tools.

The framing is deliberate: rather than compete with text-to-UI generators from OpenAI or coding agents that skip design entirely, Figma is betting the canvas stays the place where human judgment and AI generation meet. Several features remain in open beta through the summer.

Full story · Figma
06 · TRUST
Consumer Research · Agentic Commerce
Three in Four People Still Won't Let an AI Agent Pay For Anything
Even as Mastercard, Coinbase and Visa race to build the rails for autonomous spending, Forrester finds most consumers stop the agent right before checkout.

A Forrester survey of US, UK and Canadian adults, published June 15, found 75% of respondents uncomfortable letting an AI agent complete a purchase and pay autonomously — even with preset spending limits and rules in place. Analyst Lily Varon identified four recurring concerns: loss of control, errors and liability, distrust of agentic decision-making, and data-security worries.

People draw the line precisely at the payment step. Many are fine with an agent recommending products, comparing prices, or even filling a shopping cart; the holdouts who do tolerate autonomous payment restrict it to small, routine purchases under strict thresholds.

The gap between infrastructure and trust is widening fast. Mastercard launched "Agent Pay for Machines" this month for machine-speed microtransactions, Coinbase rolled out agent-linked crypto accounts, and Gartner projects AI agent software spend will hit $206.5 billion in 2026 — up 139% year over year. The rails are arriving well ahead of the willingness to use them.

Full story · Forrester via Let's Data Science
07 · SIGNALS
This Week's Quick Hits
5 Key Signals
1
JD.com's founder says robots will replace all 700,000 delivery workers
Richard Liu says the shift is coming eventually, and JD has launched a retraining program with 120 schools to move staff into robot-maintenance roles.
24/7 Wall St ↗
2
Mastercard launches Agent Pay for Machines
A new rail lets AI agents transact continuously at machine speed, including sub-cent microtransactions for compute and data.
Mastercard ↗
3
Gartner: AI agent software spend to hit $206.5B in 2026
That's up 139% from 2025, making agentic software the fastest-growing slice of enterprise spend this year.
AI Agent Store ↗
4
An open-source project claims WiFi routers can "see" through walls
RuView uses ESP32 hardware and machine learning on WiFi signal distortion for through-wall presence and pose detection — no camera required.
explainx.ai ↗
5
Google's Nano Banana Pro goes fully GA in the Gemini API
The high-fidelity image model now ships at $0.134 per 1K/2K image with best-in-class text rendering, alongside the faster Nano Banana 2.
Google AI ↗
08 · CLOSE
Top Themes This Week
Humanoid Robot IPOs
Creative AI Consolidation
Chip Architecture Race
Biosecurity Philanthropy
Agentic Commerce
Design Platform Wars
Consumer AI Trust Gap
Tool of the Week
Tool of the Week
Nano Banana Pro
Google's Gemini 3 Pro Image model — now GA, fast, and the best image model around for legible text, infographics and studio-quality edits.
Sources Cited