By Armando Pereira | Founder, PVentures Consulting | Senior Member IEEE | Co-founder, OpenFog Consortium (IEEE 1934) | President, Autonomous Vehicle Computing Consortium | Former VP/GM Optical BU, Centillium Communications (CMOS PON SoC, NTT-qualified)

👋 Welcome to The Pulse™

The Pulse™ is your weekly operating brief for execs, founders, and investors. What changed across Industrial IoT, Telecommunications, Edge Computing, Autonomous Systems, and Artificial Intelligence, why it matters, and what to do next, all in one place.

Subscribe now

🩺 This Week’s Pulse™

Inference is splitting into two reinforcing stacks, and the middle is hollowing out. Anthropic’s $200 billion, five-year, 5 GW Google Cloud commitment now sits alongside OpenAI inside more than half of the $2 trillion in cloud backlogs. The same week, five European carriers federated their telco-edge into a sovereign AI fabric, EnCharge shipped an analog in-memory edge accelerator, and Waymo extended its driverless footprint past 1,400 square miles. Hyperscaler concentration and sovereign distribution are both winning. Single-vendor cloud-only AI looks like the stranded architecture in between.

🏭 INDUSTRIAL IOT: The construction boom moved to data centers

IoT Analytics’ May Industrial Macro Pulse confirmed what the construction permits already showed: factory reshoring did not match the narrative, but data center construction did. Annualized US industrial-construction spend on data centers grew from roughly $9.5 billion in January 2020 to $47 billion in January 2026, a fivefold jump on the same five-year clock.

• IoT Analytics: data centers are now the dominant industrial buildout, larger than factory reshoring

• Schneider Electric, via Microsoft Industrial Copilot on Azure AI, reports up to 50% engineering-time cuts for early adopters

• Hannover Messe 2026: Siemens introduced the Eigen Engineering Agent, moving copilot AI from assistance to autonomous execution

Running Wi-Next in Northern Italy a decade ago, the operating question was how to instrument what already existed; in 2026, the new instrument is the building going up next door, and it is full of GPUs, not motors.

📡 TELECOMMUNICATIONS: Five European carriers federate the edge

On May 13 to 15, Vodafone, Deutsche Telekom, Orange, Telefónica, and TIM advanced a federated European telco-edge cloud aimed at sovereign IoT and AI workloads. The same week, T-Mobile and Ericsson published trial results for an AI-native scheduler on a live 5G Advanced network: spectral efficiency up nearly 10%, download speeds up 15%. Nokia handed its mobile unit to a Siemens executive with an explicit mandate to be AI-native-by-design.

• Federated European edge: transport logistics, industrial automation, emergency services on EU-based network slices

• T-Mobile / Ericsson: commercial deployment of the AI-native scheduler planned for Q3 2026

• Nokia: AI-native 5G Advanced and 6G framed as the new operating discipline

The carrier playbook through the Lantiq spin-out and the Marvell NTT Labs qualification cycle was access-and-backhaul; the 2026 playbook is sovereign-compute-and-RAN-AI, and carriers who do not own a federated edge position by year-end will rent it from those who do.

⚡ EDGE COMPUTING: Analog in-memory inference reaches product

EnCharge AI moved its EN100 analog in-memory accelerator from research to shipping form factors for laptops, workstations, and edge devices. The architectural significance is larger than the product: analog in-memory compute targets the watts-per-inference ceiling that has limited local AI on power-constrained hardware. AMD outlined a deeper commitment to embedded microprocessors for industrial edge environments the same week.

• EnCharge EN100: analog in-memory inference, two form factors, runs locally without cloud dependency

• AMD: embedded edge processor roadmap framed as core, not peripheral, to the AI strategy

• The federated EU telco-edge gives EN100-class accelerators a sovereign home that is not a hyperscaler

The fog and edge reference architecture published as IEEE 1934 specified exactly this convergence point: compute and connectivity validated as one stack at the machine edge. Analog in-memory accelerators are the power envelope that the 2017 specification could not yet name.

🚗 AUTONOMOUS SYSTEMS: Waymo expands and recalls in the same week

On May 13, Waymo enlarged its driverless service footprint to over 1,400 square miles across 11 US cities, an area larger than Rhode Island, with Miami first and Austin, Atlanta, Houston, and the San Francisco Bay Area widening behind. The company runs roughly 400,000 paid rides per week against a stated one-million-per-week year-end target. The same week, it voluntarily recalled 3,800 vehicles due to a software issue that could allow cars to drive into standing water.

• Coverage: 1,400+ square miles, 11 cities, six are FIFA World Cup hosts (May 13)

• Recall: 3,800 robotaxis, software-only fix, no injuries reported (May 12)

• Uber publicly committed over $10 billion to Rivian, Lucid, and Nuro alternatives while still carrying Waymo in Austin and Atlanta

Consortia like AVCC spent eight years building the governance table that put OEMs, tier-1 suppliers, and compute-stack vendors in the same room. Waymo’s recall-and-expansion in the same week is the operating reality that the table was built to hold: scaled service and safety transparency cannot run on separate clocks.

🤖 ARTIFICIAL INTELLIGENCE: Two firms now hold half the cloud backlog

The week confirmed Anthropic’s $200 billion, five-year, 5 GW Google Cloud commitment and the accounting reality around it: Anthropic and OpenAI contracts now account for more than half of the roughly $2 trillion revenue backlog disclosed by the major cloud providers. On May 11, OpenAI began rolling out its GPT-5.5-Cyber variant to vetted EU institutions, including the EU AI Office. On May 12, Vodafone’s earnings call candidly stated the constraint: AI is of little use without performant fixed-fiber and mobile 5G networks (networks that deliver high throughput, low latency, low jitter, and low packet loss under real load, with headroom).

• Anthropic / Google: 5 GW TPU-based capacity, starting to come online in 2027

• OpenAI GPT-5.5-Cyber: EU rollout begins; Anthropic Mythos access still outstanding for EU institutions

• Backlog concentration: AI’s two largest buyers now exceed 50% of the $2T hyperscaler revenue book

AI deployment economics is the application end of a market I have been close to since the AVCC AI work; read the commercial structure above as a forward observation, not deep architectural commentary.

🔥 3 Non-Obvious Takeaways

1. The middle of the inference stack is the most stranded position in 2026

Analog in-memory accelerators expand the edge; Anthropic and OpenAI lock in the hyperscaler core. Single-vendor cloud-only inference, which was deployed in 2024 as the enterprise default, is the most exposed configuration heading into 2027. Renegotiate before the dependency hardens.

2. Sovereign compute is now a telecom product, not a policy paper

Five European carriers federated their edge clouds for sovereign IoT and AI workloads the same week Vodafone framed AI as useless without performant fixed and mobile networks. The European carrier industry is positioning itself as the buyer’s alternative to US hyperscaler concentration, with regulated-network SLAs as the differentiator. The next 30 days of contract language will decide whether enterprises take the offer.

3. The Waymo dual signal is the field-deployable AI template

Expanding to 1,400 square miles and recalling 3,800 vehicles in the same week is not a contradiction; it is the operating cadence safety-certified autonomy requires. Any board evaluating an AI deployment that touches physical infrastructure should read Waymo’s week as the working model: scale and recall in the same cycle, transparently, without losing the service rhythm.

🧭 Where to start

If splitting your inference strategy between a hyperscaler core and a sovereign or carrier-edge layer is a live decision for your leadership team in 2026, the Strategic Review Session is the right place to test the question before procurement locks.

Book a Strategy Review Session

❓ Question for you

Which of your four operational verticals, IIoT, Telecom, Edge Computing, or Autonomous Systems, has the weakest instrumentation for detecting the inference-cost gap between hyperscaler cloud and sovereign or on-prem edge?

🗺️ The arc so far

• The Sextant™ (May 6): "The Hidden Bottleneck in AI Compute". Board-level frame; first named the capacity constraint that made the bifurcation visible. The strategic-question anchor for Weeks 19 through 21.

• The Pulse™ (May 5, Week 19): "AI Capex Sets the Clock". Quantified the hyperscaler commitment surface that now anchors the Anthropic / OpenAI half of the fork.

• The Vector™ (May 7): "The AI-RAN Architectural Fork". Deep dive on the carrier-grade compute choice that the federated EU telco-edge now operationalizes.

• The Pulse™ (May 12, Week 20): "Physical AI goes operational". Deployment-wave frame; established that the stack had crossed into production, teeing up the architectural question answered this week.

• The Vector™ (May 14): "The Inference ASIC Fork". Silicon-level preview of the architectural bifurcation this week names at the full-stack level.

📅 Board Advisory Session

I work with a small number of deep-tech executives on a fractional basis: as a CxO who has sat on standards boards (IEEE, ATIS, TIA, EIA), run cross-functional GTM in Germany and Japan, and led the technical and commercial integration that produces IEEE standards from industry consortia.

If your board or leadership team is navigating an AI, edge, or autonomous systems decision in the next two quarters, a focused advisory session may be the fastest way to pressure-test the options.

Book a 30-min Intro Session

By Armando Pereira | Founder, PVentures Consulting | Senior Member IEEE | Co-founder, OpenFog Consortium (IEEE 1934) | President, Autonomous Vehicle Computing Consortium | Former VP/GM Optical BU, Centillium Communications (CMOS PON SoC, NTT-qualified)

👋 Welcome to The Pulse™

The Pulse™ is your weekly operating brief for execs, founders, and investors. What changed across Industrial IoT, Telecommunications, Edge Computing, Autonomous Systems, and Artificial Intelligence, why it matters, and what to do next, all in one place.

🩺 This Week’s Pulse™

Five verticals, one signal this week: physical AI crossed from pilot into production-grade deployment. A Berkshire Hathaway trucking subsidiary ran driverless Class 8 routes without a safety driver. A Tier 1 European carrier switched on a multi-agentic network management platform. NVIDIA’s most powerful industrial edge platform reached general availability. Two AI labs launched enterprise joint ventures on the same day. And a $3.2 billion fiber deal rewired the capacity math underneath all of it.

That is not a coincidence of timing. It is the week the physical stack matured.

🏢 Industrial IoT: ARBOR’s ARES-2100 targets the machine edge

ARBOR Technology announced the ARES-2100 on May 7: a fanless 1U slim edge AI system built on Intel Core Series 3 processors, rated at 40 TOPS, and certified to MIL-STD-810H for shock, vibration, and temperature extremes. The design targets industrial automation and smart factory inference at the machine level, where racks and active cooling are not options.

Three things make this more than a product announcement.

• 40 TOPS at the machine edge means local inference on vision, anomaly detection, and predictive maintenance without a round trip to a server.

• MIL-SPEC certification expands the addressable market beyond factory floors into defense logistics, transportation infrastructure, and field operations where environmental stress disqualifies most commercial hardware.

• The 1U slim form factor fits directly into existing industrial rack footprints, which shortens integration cycles for plant operators who are not retrofitting infrastructure from scratch.

The gap between lab-grade AI and field-deployable AI is still mostly a packaging and certification problem, not a compute problem. ARBOR is solving the packaging problem.

📡 Telecom: Deutsche Telekom’s MINDR goes live

Deutsche Telekom activated MINDR, a multi-agentic AI platform that spans radio, transport, and core network functions, in live production across its German network. MINDR is not a monitoring overlay. It is an operational system: AI agents with defined authority to take actions across network domains, coordinated without a human-in-the-loop for routine decisions.

The architectural significance is in the cross-domain scope.

• Carrier networks have historically been managed in silos, with radio, transport, and core teams operating separate toolchains.

• MINDR treats the network as a single AI-addressable system.

• That changes the economics of network operations: fewer specialized teams, faster fault isolation, and optimization that can trade off resources across domains rather than locally.

This is the first publicly confirmed production deployment of cross-domain agentic management at Tier 1 scale. The question for other carriers is not whether to adopt agentic operations; it is how many quarters behind they can afford to be.

⚡ Edge Computing: NVIDIA IGX Thor reaches general availability

NVIDIA announced the general availability of IGX Thor, the successor to IGX Orin, delivering 8x the compute performance of its predecessor at the industrial edge. IGX Thor carries dual safety certifications: ISO 26262 for automotive-grade functional safety and IEC 61508 for industrial safety integrity. Connectivity runs at dual 200 GbE, sufficient for real-time inference over high-bandwidth sensor arrays.

The certifications matter more than the compute multiplier.

• ISO 26262 and IEC 61508 are gatekeeping requirements in automotive, robotics, surgical systems, and critical infrastructure.

• A platform that arrives pre-certified compresses the qualification timeline for system integrators by months.

• The 8x compute headroom means IGX Thor is not just a replacement for Orin deployments: it is the substrate for the next generation of edge AI applications that Orin could not run at production latency.

The fog and edge reference architecture published as IEEE 1934 specified exactly this convergence point: compute, safety certification, and high-bandwidth connectivity in a single validated platform deployable at the machine edge. IGX Thor is that specification made silicon. Integration work starts now.

🚗 Autonomous Systems: Aurora and McLane go commercial

Aurora Innovation and McLane Company launched commercial driverless freight operations on the Dallas-to-Houston corridor on May 6. McLane, a Berkshire Hathaway subsidiary with a $50 billion revenue base and one of the largest private trucking fleets in the United States, is not a pilot partner. It is a reference customer at scale.

The milestone is operational, not technical.

• Aurora’s system has logged more than 280,000 autonomous miles.

• The Dallas-Houston corridor is one of the highest-volume freight lanes in North America.

• Running it without a safety driver, under a commercial contract, with a Berkshire entity absorbing the operational risk, is a different category of validation than a technology demonstration.

Three signals stand out.

• Berkshire’s involvement means liability and insurance economics have reached a threshold acceptable to a major corporate risk manager, which is the institutional validation the AV industry has been waiting for.

• The corridor choice confirms that geographic density matters: high-volume, well-mapped, high-frequency routes are the economic foundation for AV freight. And Aurora’s lead with trucking rather than robotaxis signals where the near-term unit economics actually work.

• Consortia like AVCC spent years building the governance table that brought trucking OEMs and compute stack suppliers into the same room; McLane’s commercial launch is the outcome that the table was convened to produce.

🤖 AI Infrastructure: $1.5B JV, $4B JV, and $3.2B fiber, same week

On May 4, Anthropic launched a $1.5 billion joint venture targeting enterprise AI deployment. OpenAI announced a separate $4 billion enterprise joint venture the same day. On May 6, NVIDIA and Corning announced a $3.2 billion fiber infrastructure deal to address the physical data center connectivity constraints that AI training and inference at scale are exposing.

The sequencing matters.

• The JV announcements signal that both major AI labs have concluded that enterprise deployment requires a structured investment vehicle, not just an API.

• The Corning deal signals that the constraint is no longer compute in isolation: it is the fiber connecting compute to storage to network at data center scale.

• NVIDIA is buying its way into the physical plant problem because the next performance ceiling is not on the chip.

The week of May 4 marked the AI industry’s investment thesis visibly shifting from software services to physical infrastructure. The $8.7 billion combined JV commitment and the $3.2 billion fiber deal are the numbers to anchor that thesis for the next four quarters.

💡 3 Non-Obvious Takeaways

• The certification gap is the real moat.
  ARBOR’s MIL-SPEC and NVIDIA’s ISO 26262 + IEC 61508 certifications are not marketing points; they are two-year qualification timelines that competitors cannot shortcut. Hardware that arrives certified owns the design win before the RFP closes.

• Berkshire’s risk calculus is the AV industry’s clearest institutional signal yet.
  McLane’s decision to run commercial driverless routes without a safety driver means Berkshire’s actuaries and lawyers reviewed the liability structure and signed off. That imprimatur is worth more than any autonomous miles figure.

• The fiber deal is the tell.
  When NVIDIA commits $3.2 billion to Corning for data center optical fiber, it is signaling that the compute bottleneck has migrated one layer down the stack. The next AI performance constraint is physical plant, not silicon. Operators who are not auditing their fiber topology now will be constrained in 18 months.

🧭 Where to start

If you are an industrial operator evaluating edge AI hardware, start with ARBOR’s ARES-2100 datasheet and map it against your environmental qualification requirements before evaluating compute specs. Certification fit eliminates most of the evaluation cycle.

If you are a carrier executive watching the MINDR deployment, the first question for your operations team is cross-domain authority: which AI actions are your teams willing to authorize without human confirmation, and under what conditions. That scoping exercise takes longer than the technology integration.

If you are on the finance side of an AV or AI infrastructure investment, pull the McLane-Aurora commercial terms structure and the NVIDIA-Corning deal structure before your next board conversation. Both are public and priced, and both tell you where the industry’s counterparties think the risk sits.

❓ Question for you

Which of this week’s five signals is most relevant to a decision your organization is facing in the next 90 days?

Reply and tell me. The best responses show up in future issues.

🗺️ The arc so far

• The Pulse™ (May 5): “AI Capex Sets the Clock”: the hyperscaler commitments that built the physical substrate this week’s deployments are running on.

• The Sextant™ (May 6): “The Hidden Bottleneck in AI Compute”: board-level framing of the infrastructure constraints that this week’s deployments begin to resolve.

• The Vector™ (May 7): “The AI-RAN Architectural Fork”: the carrier-grade AI analysis of the Deutsche Telekom MINDR deployment now makes concrete.

📅 Board Advisory Session

I work with a small number of deep-tech executives on a fractional basis: as a CxO who has sat on standards boards (AVCC, OpenFog, TIA, EIA), run cross-functional GTM in Germany and Japan, and led the technical and commercial integration that produces IEEE standards from industry consortia.

If your board or leadership team is navigating an AI, edge, or autonomous systems decision in the next two quarters, a focused advisory session may be the fastest way to pressure-test the options.

Schedule a Board Advisory Session

By Armando Pereira | Founder, PVentures Consulting | Senior Member IEEE | Co-founder, OpenFog Consortium (IEEE 1934) | President, Autonomous Vehicle Computing Consortium | Former VP/GM Optical BU, Centillium Communications (CMOS PON SoC, NTT-qualified)

👋 Welcome to The Pulse™

The Pulse™ is your weekly operating brief for execs, founders, and investors. What changed across Industrial IoT, Telecommunications, Edge Computing, Autonomous Systems, and Artificial Intelligence, why it matters, and what to do next, all in one place.

Subscribe now

🩺 This Week’s Pulse™

The week’s hyperscaler earnings reset the AI capex baseline above one trillion dollars and, for the first time, named the constraint. Microsoft booked an explicit $25 billion for higher component pricing in its 2026 guide. Alphabet, Meta, and Amazon raised or held at scale. The substrate, silicon, transformers, copper, and grid interconnect have moved from background condition to disclosed line item, and every adjacent industry that orders from the same supply pool now plans against the same clock.

🏭 INDUSTRIAL IOT: Hannover ratifies industrial AI as platform

Coverage of Hannover Messe 2026 carried into the window with Siemens and NVIDIA committing to the world’s first fully AI-driven adaptive manufacturing site at the Siemens Electronics Factory in Erlangen, Germany. ABB extended its Genix Industrial IoT and AI Suite with Microsoft Azure and NVIDIA Omniverse for AI-agent root-cause analysis.

• Siemens Digital Twin Composer integrates NVIDIA Omniverse, CUDA-X and Nemotron open models for physics-grounded simulation.

• Cadence, Dassault Systèmes and Synopsys joined the Omniverse commitment across industrial design software.

• Industrial AI moves from pilot funding to platform commitment; OT vendors should expect Omniverse-grade simulation as the new RFP baseline.

Where this is heading over the next 30 to 90 days: industrial software pricing repositions around digital-twin and agentic-AI tiers, and large factory operators begin renegotiating OT licensing on those terms.

📡 TELECOMMUNICATIONS: AI-RAN forks; carriers compete for power

Ericsson and Nokia diverged this week on AI-RAN. Nokia tied its 5G and 6G strategy to deep NVIDIA GPU integration; Ericsson held to a vendor-independent path while remaining about 41% exposed to North American RAN, where T-Mobile and Verizon spend pacing has tightened.

• Verizon launched a dedicated 5G SA slice for first responders, with priority access during major incidents.

• VodafoneThree confirmed Ericsson and Nokia as joint vendors for its UK 5G build, citing the AI-RAN roadmap as a fit.

• Carriers and hyperscalers now bid for the same grid interconnect; Open RAN economics depend on power positioning.

Where this is heading: through Q3, every Tier 1 carrier RFP that touches AI-RAN will be answered against an explicit power-and-grid-interconnect schedule, not just radio performance.

⚡ EDGE COMPUTING: Edge inference goes commercial grade

Cloudflare and NVIDIA moved edge inference from research to a product line. Cloudflare announced Mesh, unifying AI agents, users, and multi-cloud infrastructure into a single secure system and shipped Gen 13 servers redesigned around AMD high-core-count CPUs for parallelism over cache.

• NVIDIA IGX Thor reached general availability for industrial-grade real-time physical AI with functional safety.

• Cloudflare’s GPU-at-the-edge rollout targets up to 300 data centers for generative AI inference.

• Edge inference shifts from cloud-rented GPU hours to fixed-asset deployment with audited safety and latency budgets.

Where this is heading: 2026 enterprise AI budgets begin reallocating from centralized cloud GPU rental toward edge inference fixed assets in regulated and latency-sensitive workloads.

🚗 AUTONOMOUS SYSTEMS: Federal AV cap loosens; deployment race opens

Tesla’s fully unsupervised robotaxi service in Dallas and Houston, launched April 18, continued to expand into the window. Tesla also confirmed Cybercab production via FMVSS self-certification, engineering the vehicle to comply directly with updated standards rather than seeking exemption.

• NHTSA issued the first-ever demonstration exemption to American-built automated vehicles.

• Congress is debating the SELF DRIVE Act to raise the AV exemption cap from 2,500 to 90,000 units.

• Waymo reached 10 cities in February; Tesla is matching two-thirds of that footprint in months.

Where this is heading: federal rules of the road, not city-by-city pilot negotiation, set deployment scale through 2027; AV roadmaps anchored on local regulatory cadence misprice the curve.

🤖 ARTIFICIAL INTELLIGENCE: Capex past one trillion meets component pricing

Hyperscaler earnings on April 29 reset the capex baseline and exposed the supply constraint. Evercore and Bank of America placed 2027 hyperscaler capex above $1 trillion; 2026 estimates moved into the $800–$900 billion range.

• Microsoft’s 2026 capex of $190B includes $25B explicitly for higher component pricing.

• Alphabet raised guidance to $180B-$190B; Meta raised to $125B-$145B; Amazon held $200B.

• Google Cloud revenue grew 63% to $20B; backlog of $462B, up roughly 400% year over year.

Where this is heading: stock divergence (Alphabet up roughly 7% after-hours, Meta down 6%, Microsoft flat) signals investors are now pricing the substrate, not the model, as the gating function.

🔥 3 Non-Obvious Takeaways

1. Component pricing has joined the AI BOM as a disclosed line item

Microsoft’s $25 billion is the first time a hyperscaler has disclosed component pricing within an annual capex guide. IIoT, edge, and AV programs that order from the same silicon, transformer, and interconnect pool now compete for the same allocation; their procurement clocks tighten quietly, before any RFP is issued.

2. Tesla’s self-certification beats Tesla’s geofence

The Dallas and Houston launches are visible; the strategic move was the FMVSS-compliant Cybercab path that entirely bypasses the 2,500-exemption cap. If the SELF DRIVE Act passes at 90,000, federal regulators set the deployment ceiling, not city governments. AV roadmaps anchored on city-by-city negotiation are pricing the wrong variable.

3. Erlangen and AI-RAN are the same bet at two altitudes

Siemens-NVIDIA-Omniverse for the factory and Nokia-NVIDIA for the radio access network are the same architectural decision: tie the operating substrate to a single AI-physics stack. Operators that keep substrate optionality preserve negotiating leverage; operators that lock in early get faster deployment but inherit the licensing curve for a decade.

🧭 Where to start

If component pricing inside hyperscaler capex is now the upstream variable on your IIoT, telecom, edge, and AV roadmaps, the Board Advisory Session on our website is the right starting point. We walk the board through the substrate dependency map, name the procurement decisions that need to move this quarter, and frame the 2026 to 2027 capex curve in terms the directors can vote on.

❓ Question for you

Which of your four critical operational verticals, IIoT, Telecom, Edge Computing, or Autonomous Systems, has the weakest instrumentation for component pricing and substrate scarcity as a forward indicator of 2026 deliverability?

If you are not tracking GPU and NPU allocation contracts, transformer and switchgear lead times, or grid interconnect queue position, you are making 2026 procurement calls without the leading indicators that will separate early movers from laggards over the next two quarters.

👋 Welcome to The Pulse™

The Pulse™ is your weekly operating brief for execs, founders, and investors. What changed across Industrial IoT, Telecommunications, Edge Computing, Autonomous Systems, and Artificial Intelligence, why it matters, and what to do next, all in one place.

Subscribe now

🩺 This Week’s Pulse

The week’s signals share one source of gravity: artificial intelligence capital expenditure. Google committed up to forty billion dollars to Anthropic. Tesla started Cybercab production by self-certifying past NHTSA’s robotaxi cap. Siemens shipped its first autonomous engineering agent at Hannover. Vodafone moved Open RAN from pilot to a live German city. Dell crystallized the small language model thesis as a 2026 procurement default. Each vertical now runs on AI capex timing, not its own.

Share this: AI capex is no longer adjacent to deep-tech procurement; it is the procurement clock that runs telecom, edge, IIoT, and autonomous decisions in parallel.

🏭 INDUSTRIAL IOT: Siemens makes the engineering agent commercial

Siemens used Hannover Messe to introduce Eigen, the first commercially available industrial AI agent that plans and executes automation engineering tasks autonomously. It is the opening deliverable of the company’s one-billion-euro industrial AI program announced in November 2025.

• Eigen targets up to fifty percent efficiency gains in PLC programming and automation workflows

• Siemens, ABB, Schneider Electric, Microsoft, and NVIDIA staged a coordinated industrial AI push across the show floor

• Cedrik Neike, who runs Siemens Digital Industries, framed agentic AI as the new competitive lever for flexible manufacturing

The next thirty to ninety days will reveal whether buyers procure Eigen as software, embedded with hardware, or as a managed service. The packaging answer will tell us how the industrial AI margin gets distributed across the vendor stack.

📡 TELECOMMUNICATIONS: Open RAN moves from pilot to a live German city

Vodafone confirmed that Wismar will become the first city in its German footprint to be fully equipped with Open RAN this spring, anchored by Samsung as the primary vendor. Several hundred towers go live in 2026; thousands more are contracted across European markets, with a 30% Open RAN target for Vodafone’s masts by 2030.

• Samsung’s vRAN and Open RAN footprint puts Ericsson and Nokia on a smaller share of new builds

• Wind River provides the cloud-native platform underneath the deployment

• AT&T’s parallel Open RAN program now carries roughly half of the network traffic on open-capable hardware

The signal moves Open RAN from the pilot column to the procurement column. Operators delaying RFP redesign now risk a vendor lock-in that the rest of the market is actively dismantling.

⚡ EDGE COMPUTING: Small language models become the enterprise default

Dell formalized 2026 as the year of the small language model, and the enterprise data backs the framing: serving a seven-billion-parameter SLM runs ten to thirty times cheaper than a comparable LLM, and Gartner now projects that seventy-five percent of enterprise data will be processed at the edge.

• Microsoft Phi, Google Gemma, Mistral 7B, Meta Llama 3.2, and Qwen 2.5 lead enterprise deployment shortlists

• Siemens’ Industrial Edge ecosystem expansion this week pulls SLM inference directly into the factory floor

• Cloudflare, AWS, and Azure positioned edge GPU capacity for inference, not training

Edge AI procurement is now a 2026 line item, not a 2027 strategy. Boards that defer edge architecture decisions miss the cost curve and the latency arbitrage simultaneously.

🚗 AUTONOMOUS SYSTEMS: Tesla unlocks Cybercab volume by self-certifying to FMVSS

Tesla confirmed Cybercab volume production at Giga Texas this week and disclosed the regulatory path: full self-certification under existing FMVSS, bypassing the NHTSA 2,500-vehicle exemption cap that constrains Waymo and Cruise. Tesla’s Robotaxi service simultaneously expanded to Dallas and Houston at roughly half of Waymo’s per-ride price.

• Lars Moravy confirmed Cybercab carries standard federal compliance stickers, no waiver required

• Waymo cleared 500,000 paid rides per week and now operates in eleven cities

• Five additional Tesla Robotaxi launches slipped from the first half of 2026 to “preparations underway.”

The market now has two viable robotaxi economics: Waymo’s high-margin urban operating model and Tesla’s mass-produced, lower-priced fleet. The question is no longer whether AV scales; it is which unit economics survive.

🤖 ARTIFICIAL INTELLIGENCE: Google commits up to $40B to Anthropic, plus five gigawatts

Bloomberg confirmed Google’s investment of up to $40 billion in Anthropic: $10 billion in cash now at a $ 350-billion valuation, $30 billion more contingent on performance milestones, plus 5 gigawatts of compute. The deal closes the week after Amazon’s twenty-five-billion commitment to the same company.

• Anthropic’s annualized revenue run rate hit thirty billion dollars in April, up from nine billion at year-end 2025

• US hyperscaler AI infrastructure spend tracks above six hundred billion dollars in 2026

• OpenAI shifted UK Stargate plans to rented compute, citing energy cost and regulatory friction

Compute is now the binding constraint on AI economics. Power, land, and grid interconnect position will determine which model providers scale and which stall.

🔥 3 Non-Obvious Takeaways

1. Telecom is in a power-and-land bidding war with AI

The Google-Anthropic 5 GW commitment and Vodafone’s German Open RAN ramp draw from the same constrained substrate: power, land, and grid interconnect. Carriers that do not lock in power positioning ahead of hyperscaler buying lose Open RAN scale economics, regardless of vendor choice.

2. Self-certification is a regulatory moat, not a technology one

Tesla did not out-engineer Waymo this week. It out-architected the regulatory pathway by self-certifying Cybercab to existing FMVSS, bypassing the NHTSA exemption cap. The same play is open to Siemens Eigen, edge AI agents, and any commercial autonomy product that can comply with existing standards rather than waiting for waivers.

3. Two AI capex curves are forming; budget for one explicitly

Google’s 5 GW Anthropic commitment funds the centralized frontier-model curve. Dell’s SLM thesis funds the distributed inference curve, where seven-billion-parameter models run on commodity edge hardware at ten to thirty times lower cost. Most boards are budgeting as if there is one AI infrastructure decision; there are two, and they reward different capital allocation patterns.

🧭 Where to start?

If your leadership team is still treating AI capex as an adjacent line item rather than the procurement clock for telecom, edge, IIoT, and AV decisions, the timing for an Introductory Session is now. We map your portfolio against the synchronized capex curve and identify the two or three procurement decisions that need to move forward this quarter, not next year!

Free 30-min Intro Session

❓ Question for you

Which of your four critical verticals, IIoT, Telecom, Edge Computing, or Autonomous Systems, has the weakest instrumentation for detecting how AI capital expenditure is reshaping its procurement timing?

If you are not tracking your suppliers’ AI-tied compute commitments, edge inference cost trajectories, or carrier and OEM exposure to hyperscaler buying decisions, you are making 2026 procurement calls without the leading indicators that will separate early movers from laggards over the next two quarters.