AI demand will rise. That part is easy.
The harder question is who captures the value.
My view: the winners may not simply be the frontier labs. A lot of AI usage will become “good enough” and get absorbed into existing distribution: search, phones, office software, CRMs, ticketing systems, and enterprise workflows.
That puts pressure on frontier model companies. If every task does not need the smartest model, enterprises will route usage: cheap/local models for basic work, secure cloud models for normal workflows, frontier models only for the hardest coding, reasoning, and agentic tasks.
This is the dot-com lesson again: first movers create the category, but durable profits often go to the companies that own distribution, workflow, infrastructure, and customer budgets.
$GOOGL $META $AAPL $MSFT $AMZN $ORCL $PLTR $SNOW $NOW $DELL $HPQ $LNVGY $SPCX $TSLA $HPE $NVDA $ALAB $ANET $VRT $CRDO $GEV
Article coming soon:
https://t.co/xEIQ0k6kCs
$VMI reports tomorrow. Valmont makes steel transmission poles, lighting structures, telecom towers, and irrigation equipment. Their connection to the buildout is through the electrical grid: as data centers demand more power, utilities need more transmission infrastructure to deliver it.
Last quarter Valmont reported record first-quarter earnings per share of $5.51, up 28% from a year earlier. Their North American Utility segment grew 27% —
driven by both higher prices and more volume. Management said "we have not seen it like this for decades." They pointed to utility capital spending plans now totaling $1.4 trillion through 2030, revised upward by roughly 27%, driven by load growth, grid modernization, and data-center demand. Full-year
Infrastructure revenue guidance was raised and the company guided full-year earnings of $21.50 to $23.50 per share. Lead times for their products are 42 to 44 weeks, the best in the industry. The drag on results was the Agriculture segment, where sales fell 15%.
What to watch: is the Utility segment still growing at the same pace? And is the Agriculture business stabilizing? Longer lead times are worth
watching — if they stretch further, it signals demand continues to outpace supply.
A good report from Valmont matters for the buildout because transmission infrastructure is the physical layer that connects power generation to data
centers. If utilities are ordering more poles and structures, it means they are actually building the grid capacity that the AI buildout requires — not
just planning it.
$MLI reports tomorrow. Mueller Industries makes copper, brass, and aluminum products used in plumbing, air conditioning, refrigeration, and electrical
applications. Their connection to the buildout is indirect: copper tubing is used in the liquid cooling systems that keep AI servers from overheating, and
their wire and cable subsidiary serves electric utilities.
Last quarter Mueller did not hold an earnings call or issue a press release with operational results. The quarterly filing stated there were no material
changes in risk factors. Since then, the company declared a 2-for-1 stock split and maintained its quarterly dividend — actions that typically signal
board-level confidence in the trajectory of the business.
What to watch: this is the first real data point in months. What did revenue and margins look like? And — importantly — will management say anything
about demand from data centers or electric utilities? Until now, the AI link has been entirely outside analysis, not confirmed by the company.
A good report from Mueller would add one more data point to the broader picture: that the companies supplying basic industrial materials — copper, wire,
tubing — are seeing demand tied to the physical infrastructure that AI data centers require.
$RRC reports tomorrow. Range Resources is a natural gas and natural gas liquids producer focused entirely on the Marcellus Shale in Appalachia. Like $EQT ,
their gas feeds the power plants that are being planned to serve data-center electricity loads.
Last quarter Range produced 2.2 billion cubic feet equivalent per day. Free cash flow was roughly $400 million. The standout: they achieved the best
pricing premiums in company history — their natural gas sold at a premium to the benchmark Henry Hub price, and their natural gas liquids sold at more than
$4 per barrel above the main trading hub price. Management raised their full-year pricing guidance and disclosed that they signed a 75 million cubic feet
per day gas supply agreement with a new power plant in the Midwest, with service starting in late 2027. They also said they are in active discussions with
more than a dozen data-center and power-generation developers.
What we're watching: any additional supply agreements signed since last quarter? One deal is a start — more would signal that the data-center gas demand thesis is broadening beyond a handful of projects. Also: have those premium pricing levels held, or was the first quarter unusually strong due to weather and temporary supply disruptions?
A good report from Range matters for the buildout because gas producers in Appalachia sit on some of the lowest-cost supply in the country. If multiple
producers start signing long-term deals with power plants and data centers, it confirms that the buildout is pulling demand across the entire energy supply
chain, not just a few large players.
$EQT reports tomorrow. EQT is the largest producer of natural gas in the United States, operating in the Appalachian Basin. Natural gas is the primary fuel
for the power plants that are being built to supply the growing electricity demand from AI data centers.
Last quarter EQT generated more than $1.8 billion in free cash flow — a record. They reduced net debt to below $5.7 billion and received a credit rating
upgrade to investment grade. Management said they now see data-center gas demand of 10 billion cubic feet per day as "the new base case" — a significant
increase from prior estimates. They disclosed that they are in active negotiations for multiple large supply agreements and expect to sign deals in the
second half of the year. Full-year free cash flow was guided to roughly $3.5 billion.
Since that report, gas prices have been weak during the spring season, and EQT disclosed they would voluntarily curtail 10 to 15 billion cubic feet of
production in the second quarter rather than sell into low prices.
What we're watching: have any of those supply agreements moved from negotiation to signed contracts? The data-center demand story has been a narrative — signed deals turn it into something measurable. Also: progress on the balance sheet — management targeted $5 billion in net debt by year-end.
A good report from EQT matters for the buildout because it would confirm that data-center operators are securing baseload power through long-term contracts with gas producers, not just talking about it. That makes the infrastructure buildout more concrete.
$VICR reports tomorrow. Vicor makes the power delivery systems that sit underneath AI chips — their technology converts and delivers electricity at the
point of load for chips that are too powerful for conventional power architectures.
Last quarter they reported revenue of $113 million, up 20% from a year earlier. New orders came in at more than twice what they shipped — a book-to-bill
ratio above 2. Their backlog of orders scheduled for delivery within 12 months jumped 70% to $301 million. Management said the company is "essentially sold
out" and gave their first full-year revenue target of nearly $570 million. They also revealed they can expand their existing factory to support $1.5
billion in annual revenue, up from a previous ceiling of $1 billion.
Since that report, their lead customer — Cerebras, which makes wafer-scale AI chips — went public and reported revenue up 92% from a year earlier. Every
Cerebras system ships with Vicor power delivery inside. Cerebras guided to $865 million in revenue for the year and disclosed $25 billion in remaining
performance obligations. Vicor has made no public statements since April.
What we're watching: did Vicor hit its second-quarter revenue target of nearly $126 million? Is the order book still building — another book-to-bill above
2? And is their next-generation Gen 5 power delivery on schedule to begin ramping with the lead customer in the second half of the year?
A good report from Vicor tells us something real about the physical buildout: that the companies making the most advanced AI chips are ordering the
specialized power infrastructure needed to deploy them at scale, and that the supply chain for that infrastructure is functioning.
$TBI released a survey finding that experienced workers are driving renewed interest in skilled trades. The survey indicates a shift in workforce sentiment toward construction, electrical, and other hands-on roles—areas that have faced persistent labor shortages in recent years.
This could signal a potential expansion in the available labor pool for sectors like data center construction and renewable energy installation. If the trend continues, it may help reduce project delays and labor cost pressure for companies that rely on skilled trades workers to build physical infrastructure.
Suppliers and customers in this ecosystem could include $CARR , who make cooling systems for high-density data center racks; $CAT , who manufacture construction equipment used in site development; and $AMZN, whose data center and renewable energy projects depend on the same skilled labor pool that TrueBlue helps supply.
https://t.co/KI37OehAQa
Not investment advice.
$PWR — Quanta Services — could be a potential infrastructure partner as BTQ Technologies and TIDAL PWR announce a strategic collaboration to advance trusted quantum data center architecture. The partnership aims to develop secure, high-performance computing environments for quantum workloads.
This could signal increased investment in specialized data centers with unique power, cooling, and security requirements. If such designs gain traction, Quanta’s negotiated MSA model might position the company to capture work on emerging quantum-ready facilities.
Suppliers here could potentially include $BE , who provide solid oxide fuel cells for backup power [documented], and $APH, who make connectors for data center power distribution [inferred]. Competitors like $ACM , who design power infrastructure for data centers, could also be involved.
https://t.co/x0KYcpqXuq
Not investment advice
$NXT completed its acquisition of Prevalon Energy and launched an advanced energy storage business. The deal integrates Prevalon’s battery storage platform with NextPower’s solar trackers, power conversion, and software. The new business targets utility-scale solar-plus-storage and data center applications.
This could signal a broader shift for NXT beyond solar tracking toward an integrated energy storage platform. If Prevalon’s technology scales well in NextPower’s existing project pipeline, it might increase wallet share per site and diversify revenue streams away from pure tracker sales. The data-center voltage-support use case mentioned by management could become a more material growth vector.
Suppliers in this space could include $ABB, who make inverter-transformers used in utility-scale storage, and $ENPH, who provide inverters for solar-plus-storage systems, though their focus is more residential. Competitors like $CSIQ offer integrated solar and storage solutions, while downstream customers like $AES could potentially deploy Prevalon-based systems on large projects.
https://t.co/M0I5JKCYcE
Not investment advice.
$NNBR announced it has expanded its contract manufacturing of weapons components and secured new awards with a leading firearms provider. The company did not disclose the contract value or the customer's identity in the release.
This could indicate NNBR is deepening its position within the defense supply chain, a segment it has been targeting as part of its portfolio shift away from automotive. If these awards are material, they may contribute incremental revenue to the company's Power Solutions segment, which already serves aerospace/defense end markets.
Suppliers here could potentially include companies like $RGR, a major firearms manufacturer that may use similar precision components, and $CLF , who supply steel for such parts. Competitors in precision machining for defense applications include $KAMN, who make aerospace components, and $HEI, who produce electronic components for defense.
https://t.co/2qfuD94l5d
Not investment advice.
$NKLR just announced a commercial letter of intent with Waiken ILW to deploy SOLO™ microreactors at DIRECTV Latin America data center facilities. The agreement covers initial deployment of multiple 1 MWe units to provide behind-the-meter, zero-carbon power. No firm order value or timeline was disclosed.
This LOI moves Terra from the non-binding MOU stage into a commercial letter of intent, which could signal stronger customer commitment than earlier pipeline agreements. If the deployment proceeds, it may provide a reference case for microreactors serving telecom and media data center operators in Latin America — a region without established nuclear regulation for small modular reactors.
Suppliers here could potentially include $AMRC , who has a documented framework with Terra for siting, development, and deployment of up to 50 reactors. Cooling systems would likely involve SPG Dry Cooling, who already have a cooling agreement with Terra for the SOLO design. Connector and power distribution suppliers like $APH, who make interconnect solutions for data center infrastructure, could potentially see downstream demand if the deployment scales.
https://t.co/tvU4VHNx7z
Not investment advice.
$HUT announced full commercialization of its 1 GW Beacon Point AI data center campus with a second 352 MW IT lease, bringing the campus-level base-term contract value to $19.6 billion. The lease follows the earlier River Bend agreement and adds to Hut 8’s contracted AI infrastructure backlog.
This could indicate the company’s repeatable development model is gaining momentum. If the Beacon Point counterparty carries a similar credit profile as the River Bend lessee, it might imply stable, long-term cash flows from investment-grade tenants. The rapid scaling from 1 GW to a larger pipeline may suggest they are positioning as a leading developer of hyperscale AI campuses.
Suppliers here could potentially include $AEP , who provide power infrastructure for the campus [documented]. Others in the supply chain for high-density AI data centers might include $CARR, who make cooling systems, and $CAT, who manufacture backup generators. Competitors developing similar facilities include $APLD and $BTBT.
https://t.co/JAElEiEEQ0
Not investment advice.
$HPQ announced its ENDURA+ Gen4 21700 cell platform has achieved UL 1642 safety certification. The certification covers the lithium-ion cell’s compliance with a key safety standard for rechargeable batteries. HP disclosed the milestone in a release, but did not specify intended applications or volumes.
This certification could indicate HP is progressing toward commercializing its own battery cell technology. If HP plans to integrate these cells into products—such as uninterruptible power supplies, energy storage, or even PC batteries—it might represent a new vertical for the company. The move could also suggest HP is positioning to secure its own supply of cells, rather than relying solely on external battery vendors.
Suppliers in adjacent power and thermal management spaces could potentially see demand if HP scales production. Companies like $APH , who make power connectors, and $CARR , who manufacture cooling systems, could be part of the broader ecosystem. Competitors in the battery cell space include established manufacturers, though HP’s entry would be as a new participant rather than a traditional battery supplier.
https://t.co/8CGdmecxGh
Not investment advice.
$IREN just announced $2.8B in new AI cloud contracts and raised its 2026 annualized run-rate revenue target from $3.7B to over $4B, with ~85% now under contract.
Customer list now includes Microsoft, NVIDIA, Perplexity, Figure AI, Together AI, and others across bare metal and managed cloud. Contracted pricing continues to strengthen, with recent contracts including customer prepayments covering ~45% of associated GPU capex. Weighted average contract term ~4 years. $7.6B cash on hand.
The scale of this ramp could be significant for the broader AI infrastructure stack. Companies like $APH , who make high-speed interconnects, and $VRT , who provide power and cooling for GPU-dense environments, could potentially see demand as IREN scales from 480MW this year to a targeted 1.2GW in 2027.
Co-CEO Daniel Roberts: "In the past 12 months we have expanded from approximately 3MW of self-built AI Cloud capacity to 480MW being delivered this year, with 1.2GW targeted for 2027."
3MW → 480MW → 1.2GW in two years. That quote alone tells the story.
https://t.co/nqFq7lAMxY
Not investment advice.