The Bottleneck
Is the Wire
Fifteen transmission nodes placed by FCC ground-conductivity data, serving six gigawatt-class demand anchors through ground-coupled surface waves where the earth conducts and beamed corridors where it doesn't, selling what the incumbent grid cannot: connection without the queue, blackstart without a grid tie, timing without GPS. Fueled, at the horizon, by helium-3. You are building the largest machines in history on someone else's grid. Tesla started this delivery layer 125 years ago. This page finishes the argument with the physics, data, and hardware of 2026.
Compute is scaling exponentially.
Transmission is scaling like it's 1975.
The numbers are federal, not speculative. Berkeley Lab's congressionally mandated report found US data centers consumed 176 TWh in 2023, about 4.4% of national electricity, with 2028 projections of 325 to 580 TWh, as much as 12% of the country's power.[01]
The delivery system is not keeping up. At the end of 2024, roughly 2,290 GW of generation and storage sat in interconnection queues, nearly twice the entire operating US fleet. The median project now takes over four years from request to operation. Only 13% of capacity that applied between 2000 and 2019 was ever built.[02]
If you are siting a gigawatt-class facility, your critical path is not chips, land, or talent. It is a wire you do not own, permitted on a schedule you do not control. The companies solving this today do it by buying their own generation. Almost no one is rethinking delivery. That is the open layer.
What this proposal is not
It is not free energy, and it is not a claim that one tower can power the planet. The Earth-ionosphere cavity is real (Schumann resonance, 7.83 Hz) but lossy. Tesla's single-tower global vision does not close, and any pitch that says otherwise should be ignored.
What does close: guided surface-wave delivery reported at kilowatt scale over a kilometer (company-era accounts, never independently published[05]), optical power beaming at 800+ W over 8.6 km (DARPA, 2025[04]), and through-earth transmission (US Navy, operational for two decades[06]). This proposal connects proven pieces. Every claim on this page carries a source.
I spent eight years producing video for brands you know, and my industry has been running this exact experiment for a decade: capture always outruns the pipe. You can buy a 12K cinema camera off the shelf today, and 17K bodies are shipping. But try to deliver that signal and the platforms fold: 4K is the practical streaming ceiling, and 8K, standard in capture for years, still buffers, gets crushed by compression, or simply can't be posted at native resolution. The sensor was never the bottleneck. The delivery layer was, and it still is.
Energy in 2026 is the same graph with bigger numbers. Generation is the sensor: 2,290 gigawatts of it sitting in the queue, more than double the entire installed US fleet. Transmission is the platform that can't buffer it. Cameras taught me you don't fix that by building better sensors. You fix it by building a better pipe. This page is the pipe.
"If anyone can draw on the power, where do we put the meter?"
Here is what the record does document. Morgan put $150,000 into Wardenclyffe in 1901 to build transatlantic wireless communication. When Marconi crossed the Atlantic first with a cheaper system, and Tesla revealed that the tower's true purpose was transmitting power itself, Morgan declined all further funding, and other financiers followed his lead. Tesla's own letters to Morgan survive, including his 1903 plea: "Will you help me or let my great work — almost complete — go to pots?" Morgan never funded another dollar. The tower was dynamited for scrap in 1917.[16]
Morgan was not wrong about the business. He was early about the customer. In 1903 no buyer on Earth needed unmetered power delivery badly enough to fund it. In 2026 the buyers are named at the top of this page, they are spending tens of billions of dollars a year on metered electricity, and the meter question finally has an answer: the meter goes at the gigawatt campus, where the delivery contract is signed. The technology did not fail a century ago. The market had not been born yet. It has been now.
What Tesla's sites actually had in common
The historical record is specific. Colorado Springs was chosen because attorney Leonard Curtis, a shareholder in the local El Paso Electric Company, offered Tesla free electricity and discounted land, at a 6,000-foot altitude where thin, dry air made high-voltage experimentation easier.[07][08] Wardenclyffe was cheap acreage on the Long Island Rail Road near New York, where Tesla sank a 120-foot shaft and drove iron pipes 300 feet further down to reach the saline groundwater table, in his words, to "get a grip of the earth." Strip away the era and four criteria remain. They are the same four that decide where you site a gigafactory today.
Water in the ground
Conductive, water-saturated strata are the transmission medium. Saline aquifers, coastal boundaries, and karst limestone aquifers (the Edwards Aquifer under central Texas is a limestone system) hold the dissolved minerals that carry current. Dry rock does not. This is why the FCC has mapped US ground conductivity, from 0.5 to 30 mS/m, for a century of AM broadcasting.[03]
Power at the source
Tesla went where electricity was free. The modern equivalent: site nodes where generation is abundant and stranded, Pacific Northwest hydro, Great Plains wind, Texas solar and gas, and deliver it where the queue can't.
Iron and infrastructure
Tesla's earth connection was sixteen driven iron pipes; his supply line was a railroad. Grounding arrays, rail access, and heavy industrial capacity are still prerequisites. It is no accident the Navy built its through-earth ELF system next to the Great Lakes iron ranges.[06]
The right rock for the mode
Two valid modes, opposite geology. Surface-wave delivery wants conductive saturated ground (aquifers, coasts, plains). Through-earth ELF wants the opposite: resistive shield granite that forces current miles deep, which is exactly how the Navy reached submarines through the planet. Site each node for the physics it runs.
Fifteen nodes. One network.
Six gigawatt customers.
Every node is scored against the four criteria using FCC M3 ground conductivity data and the documented geology under each city.[03] Cyan arcs are the unified grid backbone; red dashes are demand corridors delivering to each customer. Red markers are gigawatt-class demand: Terafab, Colossus, Hyperion, Stargate, TSMC, and Intel Ohio. Small cyan points are today's operating data center markets, sized by CBRE-reported inventory.[11] Select any node, primary or second tier.
Why the second tier exists
Memphis and Gulf Louisiana earn their place by serving live gigawatt customers (Colossus and Hyperion) from conductive embayment alluvium and salt dome geology. Red River Valley is top-tier FCC conductivity with wind-belt power and no story required. Niagara pairs 2.7 GW of working hydro with the site where Tesla's AC system first ran at scale in 1895.
Waxahachie is the anomaly worth saying out loud: 14+ miles of abandoned Superconducting Super Collider tunnel already bored through the Austin Chalk limestone, one town from the standing Viziv tower, on I-35. Great Salt Lake stays on the bench as the wildcard, hypersaline brine giving seawater-class conductivity a thousand miles inland.
What the load map shows
Toggle the market layer and one fact jumps out: Northern Virginia alone runs 4,039.6 MW of data center inventory, roughly three and a half times all secondary US markets combined, at a 0.5% vacancy rate with $155–185/kW asking rates.[11] The demand is already continental. The delivery layer under it is not.
Note where the red anchors sit: Terafab, Stargate, TSMC, and Intel Ohio all occupy dry or middling ground. They are demand, not transmitter sites, and the architecture serves them by beamed corridor from the nearest conductive node. Siting each function where its physics works is the entire discipline of this map.
What the energy bill actually is
Run the federal numbers and the problem prices itself. US data centers consumed 176 TWh in 2023; at EIA's ~9¢/kWh industrial rate that is roughly a $16 billion national electricity bill, today. Berkeley Lab's 2028 high case is 580 TWh: $52 billion a year at industrial rates, $84 billion at the 14.4¢ commercial rate many facilities actually pay.[01][12] Be clear about what that number means here: wireless delivery adds conversion loss, so it does not cut the per-kWh price. The product is access. Queue bypass, blackstart without a grid tie, and GPS-independent timing are what this network sells, and none of them can be bought from the incumbent grid at any price.
Per facility: one gigawatt of continuous load draws about 8.8 TWh a year. At 9¢, that is roughly $700 million a year, about $2 million a day, per gigawatt, before a single rate increase. And rates are moving the wrong way: EIA reports industrial electricity up 8.6% year over year, with the steepest increases in Virginia and Ohio, the two states at the center of the data center boom.[12] A multi-gigawatt campus like Terafab books an electricity line item measured in billions per year. That line item is the scale of customer this grid exists to serve.
The land ledger
Toggle the generation layers and read the acreage in the tooltips. Generation takes land because physics says so: Roscoe Wind spreads 781 MW across roughly 100,000 acres. Solar Star packs 579 MW onto about 3,200 acres. Palo Verde puts 3.94 GW on ~4,000 acres. Energy density decides the footprint, and no delivery technology changes that.
What can change is the footprint of delivery. A 500 kV transmission corridor consumes a right-of-way roughly 200 feet wide, about 24 acres per mile, so a 500-mile line permanently occupies on the order of 12,000 acres of easement, which is precisely why new lines take a decade to permit. A grid node builds vertically instead: Tesla put a 187-foot tower and its entire earth system on a plot measured in acres, and a modern node targets the same. The claim is not that towers replace power plants. It is that delivery can stand up instead of sprawling out, and delivery is where the decade of permitting lives.
Four elements generate.
The fifth delivers.
The ancients organized nature into earth, wind, fire, and water, with a fifth element binding them. Read it as an organizing lens rather than a physics claim, and it maps cleanly onto modern energy engineering: the four classical elements are the four families of generation we actually build, and the fifth, the one the Greeks called aether, maps onto the electromagnetic field and the atomic nucleus, the physics this entire page is built on. Every figure below is federal or peer-reviewed.[18]
Earth
Wind
Fire
Water
Aether
The honest physics, first: the 19th-century luminiferous aether died by experiment in 1887. But what the ancients were reaching for survived and got a real name. The electromagnetic field is the actual medium of every layer on this page: the surface wave bound to the ground, the beamed corridor, the resonant coil, the grid itself. The fifth element is not generation. It is delivery.
And the nucleus is where density lives: fission delivers roughly two million times the energy of burning the same mass of fuel, from ~8% of US capacity, and helium-3 fusion runs an order beyond that.[14] Four elements make the power. The fifth moves it, and concentrates it. That is the element this grid is built from.
Three of Tesla's mechanisms
survived a century of review
One engineering statement underneath everything here, and it is not a metaphor: in a ground-coupled system the earth itself is the wire, the conductive boundary that guides the wave, the return path that closes the circuit, the medium the Navy pushed signal through for 26 years. Here is the physics, mechanism by mechanism.
High-Q slow-wave resonators
Tesla's "extra coil" was a base-driven quarter-wave helical resonator. Wave velocity along the helix drops to 1–10% of c, letting a compact coil sustain a standing wave that multiplies base voltage by its quality factor. The topology is standard transmission-line physics. Modern silicon carbide drivers replace his spark gaps with precise, efficient switching.
V_top = V_base × QGuided surface waves
A Zenneck-type wave bound to the earth-air boundary decays exponentially upward instead of radiating to space, and attenuates slower over conductive ground than broadcast radiation. Kilowatt-scale delivery over kilometer distances was demonstrated in Texas before the operating company's 2021 bankruptcy.[05] Grid scale is unproven. That gap is the program.
Z_s = √(μ₀ / ε₀(εᵣ − jσ/ωε₀))Coherent phased networks
The capability Tesla lacked. GPS-disciplined clocks phase-lock every node to nanosecond precision, so seven stations operate as one coherent continental array, steering constructive interference toward a receiver instead of hoping the planet resonates on its own. The same discipline gives the network a second product: GPS-independent timing and positioning.
Σ nodes → one apertureWardenclyffe, rebuilt with 2026 parts
Chemical grounding array
Copper-clad steel rods in bentonite-carbon backfill with buried radial counterpoise, targeting R_g ≤ 2 Ω contact resistance into the local aquifer. Replaces Tesla's 120-foot timber shaft and sixteen 300-foot iron pipes with standard utility-grade practice.
High-Q helical extra coil
Single-layer oxygen-free copper on a low-loss dielectric form, toroidal terminal, slow-wave velocity 0.02c – 0.1c, quality factor Q > 1000. Tesla's exact topology, executed with modern materials and tolerances.
Solid-state resonant inverter
Full-bridge silicon carbide MOSFETs with zero-current-switching feedback replace the spark gap, the loudest and lossiest component of 1901, delivering high conversion efficiency and exact frequency agility across 20 – 150 kHz.
Network coherence and tuning
Vector network analyzers null the reactance at each feedpoint. GNSS-disciplined timing locks every node's phase so the grid behaves as one instrument. Tesla tuned by watching sparks. This tunes by the picosecond.
Dual-mode delivery
Series-resonant ground-coupled receivers for regional distribution and blackstart resilience. DARPA-class optical beaming for line-of-sight high-power corridors.[04] Two physics, one network, each used where it wins.
Every layer of this has run
somewhere, at some scale
Colorado Springs and Wardenclyffe Origin
Tesla lights lamp banks in open soil miles from his Colorado station, records terrestrial standing waves from passing thunderstorms, and begins the 187-foot Wardenclyffe transmitter over Long Island's saline aquifer before funding collapses in 1905.[07][10]
US Navy Project ELF Operational
From Clam Lake, Wisconsin and Republic, Michigan, the Navy drives extremely low frequency current through the resistive Laurentian granite, using the planet's crust as an antenna to signal submerged submarines worldwide. Through-earth transmission, in continuous military service for nearly two decades.[06]
Viziv Technologies, Milford, TX Reported Defunded
A Wardenclyffe-class Zenneck surface-wave tower rises on I-35 south of Dallas. First-person site accounts describe roughly 1 kW received wirelessly at over a kilometer, running a one-horsepower motor. No peer-reviewed measurement was published before the 2021 bankruptcy, so this page treats it as the strongest unverified data point in the record, not as proof.[05] What the tower proves beyond dispute: someone built Wardenclyffe-scale surface-wave hardware in Texas this decade, and it is still standing on the Phase 1 corridor.
DARPA POWER program Federal record
More than 800 watts delivered by laser across 8.6 km at White Sands, with over a megajoule transferred during the campaign, the standing distance record for optical power beaming. Relay-chain architecture is now in Phase 2.[04] The beamed backbone of this grid is being de-risked with federal money right now.
The Continental Resonance Grid The open seam
No one has connected these proven layers into one network: ground-coupled regional delivery, beamed high-power corridors, and continent-wide phase-locked timing, sited by conductivity data and anchored to real industrial demand. That integration is the program.
Five phases. One build-out.
Feasibility first, frontier last.
The questions this answers, in order: how do you build an energy grid across the US (dual-mode delivery, ground-coupled where the earth conducts, beamed where it doesn't); what has to be produced (the node stack in Section 06, built from the fifth element in Section 04); where does it complete (the fifteen-node mesh on the map); and what is most feasible first (the Texas corridor, where a standing tower, a limestone tunnel network, and a $16.8B customer already share one highway). The phases below are ordered by feasibility, each one funding and de-risking the next.
The Texas Corridor
- Two-node testbed: Austin + Waxahachie on I-35, extending toward Grimes County
- Kilowatt-class surface-wave delivery over tens of km on blackland prairie
- GNSS-disciplined two-node coherence demonstration
- Instrumented receiver vault at the Terafab perimeter
The Heartland Spine
- KC hub live on the best ground in America, linked south to Texas
- Colorado Springs heritage node: high-altitude testing where Tesla tested
- Memphis node serving Colossus, the first live AI-campus customer
- Green Bank calibration station receiving and verifying every emission
Water & Coasts
- Saltwater nodes: Seattle, San Diego, Shoreham, Gulf Louisiana
- Hoover node anchors the desert; first beamed corridor to the TSMC cluster
- Hyperion service from salt dome geology in Louisiana
- Niagara node closes the heritage loop where AC power began
Continental Unification
- All fifteen nodes phase-locked into one coherent continental instrument
- GPS-independent timing and positioning sold as a resilience product
- Blackstart service: restart any downed campus with no grid connection
- Minneapolis, Fargo, and Great Salt Lake complete the interior mesh
The Fusion Era
- Megawatt-class beamed corridors to fab, launch, and data campuses
- Helium-3-cycle fusion plants siting at nodes, direct electricity to delivery
- Lunar He-3 supply chain feeding the generation layer per Section 09
- Grid-scale wireless delivery: the unproven summit, approached with a decade of data
Why Terafab is the forcing function
The largest building on Earth is rising in Grimes County, Texas: a $16.8 billion first phase of a vertically integrated fab whose stated purpose is to close the gap between global chip supply and a compute demand its builders measure in terawatts.[09] Facilities in this class, and the campuses Google and Meta are building alongside them, will carry the largest point loads the American grid has ever served.
Phase 1 of this program is a rounding error against that capital plan, and it buys three things no transmission line offers: a blackstart resilience layer independent of the grid, GPS-independent precision timing for facilities that cannot tolerate drift, and a founding position in the power-delivery layer every gigawatt campus after this one will need.
Tesla died with the model right and the century wrong. The century is no longer the problem.
The grid delivers power.
Helium-3 is where the power is going.
Every layer of this proposal so far concerns moving energy. This section concerns what the next generation of energy runs on, and it is the least speculative part of the page: the US government has already made its first purchase of a natural resource from the Moon, and Microsoft has already signed a contract to buy fusion electricity. Both transactions run through helium-3.
The scarcity, and who is already buying
Earth has no natural standing supply of helium-3. What exists comes as a byproduct of tritium decay in nuclear stockpiles and trace separation from natural gas, a flow measured in kilograms while demand from quantum computing, medical imaging, and neutron detection for weapons screening keeps climbing. The US government has been searching for a scalable source since the shortage it confronted around 2010.[13] The Moon, blasted by solar wind for 4.5 billion years with no atmosphere to stop it, holds an estimated million tonnes locked in its regolith at 20 to 50 parts per billion in titanium-rich mare soils.[14]
Interlune, the Seattle company founded by former Blue Origin president Rob Meyerson, Gary Lai, and Apollo 17 moonwalker and geologist Harrison Schmitt (the man who collected the samples this entire field is built on), has converted that geology into signed contracts: the DOE Isotope Program's purchase of three liters of lunar helium-3 by 2029, the first government purchase of a non-terrestrial natural resource in history; Maybell Quantum buying thousands of liters annually from 2029 for dilution refrigerators that hold quantum processors below 10 millikelvin; and Bluefors at up to 10,000 liters a year through 2037, the largest commercial space-resource agreement to date. In July 2026, Interlune's Cold Capture system demonstrated 99% pure helium-3 from domestic helium, technology that could triple US supply while the lunar chain gets built.[13]
The honest caveats, stated plainly: USGS astrogeologists note that recovering one kilogram means processing on the order of 100,000 to a million tons of regolith, a copper-mine-class operation, and D-He3 fusion is not yet commercial.[14] But NASA has put contract money behind the extraction hardware through its CLPS and SBIR programs, and no one disputes the resource is there and measurable.
The fusion endgame, and why it closes the loop
Deuterium-helium-3 fusion is the reaction engineers have called the perfect fuel for forty years: its primary products are charged particles, not neutrons, meaning less irradiated equipment and the possibility of converting fusion energy to electricity directly instead of boiling water. Helion, in Everett, Washington, is building exactly that machine, a pulsed system with direct electricity recovery whose fuel cycle runs on deuterium and helium-3 the company breeds in-house. Microsoft signed the world's first fusion power purchase agreement with Helion in 2023, targeting at least 50 MW from 2028, and ground broke on the Orion plant at Malaga in July 2025. Commonwealth Fusion Systems followed with a 200 MW agreement with Google for the early 2030s.[15]
The launch industry has now aligned behind the same destination. SpaceX is targeting an uncrewed lunar landing in March 2027 and holds a roughly $4 billion NASA contract to land Artemis III astronauts on the Moon, and Musk's February 2026 pivot puts the world's highest-cadence launch company on the lunar route full time.[17] Interlune's own founding team came out of Blue Origin's presidency and the Apollo program itself. The transportation layer the helium-3 economy needs is being built by the exact companies this page addresses.
Read those names again: Microsoft and Google, the same companies on this page's map buying gigawatts for data centers, are the first customers of helium-3-cycle fusion. The demand anchors and the fusion buyers are the same companies. The race back to the Moon is not nostalgia; through the DOE's own purchase order, it is a procurement program for the most valuable material per kilogram that humans have ever priced.
And this is where the whole page becomes one system. Fusion plants like Orion produce electricity directly, at whatever site the reactor stands. A delivery layer that moves power without a decade of corridor permitting is the natural counterpart to generation that no longer needs to sit beside a coal seam or a dam. Tesla's grid was a delivery architecture waiting for a worthy source. Helium-3 fusion is a source waiting for a worthy delivery architecture. The Seattle node of this grid sits in the middle of both: Interlune in Seattle, Helion in Everett, Orion on the Columbia.
Sources
Every load-bearing claim on this page traces to the record below. Anything that cannot be sourced does not belong in an infrastructure proposal.
- US DOE / Lawrence Berkeley National Laboratory, 2024 United States Data Center Energy Usage Report. Data centers at 4.4% of US electricity in 2023 (176 TWh); projected 6.7–12% (325–580 TWh) by 2028.
eta-publications.lbl.gov - Lawrence Berkeley National Laboratory, Queued Up: 2025 Edition. ~2,290 GW active in interconnection queues at end of 2024; median request-to-operation over 4 years; 13% of 2000–2019 applicant capacity ever built.
emp.lbl.gov/queues - Federal Communications Commission, M3 Map of Effective Ground Conductivity in the United States (47 CFR 73.190). US ground conductivity 0.5–30 mS/m; seawater 5,000 mS/m.
fcc.gov/media/radio/m3-ground-conductivity-map - DARPA, Persistent Optical Wireless Energy Relay (POWER) program. 800+ W delivered by laser at 8.6 km, White Sands, 2025; program Phase 2 pursuing integrated relays.
darpa.mil/news/2025 - Viziv Technologies Zenneck surface-wave program, Milford, TX (2018–2021), including first-person site accounts (unverified; no peer-reviewed measurements were published) of ~1 kW wireless delivery at >1 km, and reporting of the company's 2021 bankruptcy.
texashighways.com · howstuffworks.com - US Navy Project ELF, Clam Lake, WI and Republic, MI. Extremely-low-frequency through-earth transmission to submerged submarines over the Laurentian shield, operational 1989–2004 (Clam Lake test transmissions from the 1980s).
en.wikipedia.org/wiki/Project_ELF - Nikola Tesla, Colorado Springs Notes, 1899–1900. Primary experimental record of the Colorado station, terrestrial stationary-wave observations, and magnifying transmitter development.
- History Colorado, "The Wizard in the Mountains." Leonard Curtis, shareholder in El Paso Electric, offered Tesla discounted property and free electricity to bring his experiments to Colorado Springs.
historycolorado.org - SpaceX / Tesla Terafab announcement, Grimes County, TX, August 2026. $16.8B initial investment, 100M+ sq ft planned, at least 3,000 jobs.
techcrunch.com - Tesla Science Center at Wardenclyffe. Site history of the Shoreham, NY facility: the 187-ft tower, 120-ft shaft, and grounding system over Long Island's glacial aquifer.
teslasciencecenter.org - CBRE, North America Data Center Trends H2 2025. Northern Virginia at 4,039.6 MW inventory (largest market globally, 0.5% vacancy); Atlanta 1,459 MW; Dallas-Fort Worth 1,067 MW; primary-market supply up 36% year over year to 9,432 MW; record 2,497.6 MW net absorption in 2025.
cbre.com · press release - US Energy Information Administration, Electric Power Monthly and annual sales/revenue data. Average retail electricity ~9.0¢/kWh industrial and ~14.4¢/kWh commercial (early 2026), with industrial rates up 8.6% year over year and the largest increases in Virginia and Ohio. Cost projections on this page are computed from these rates and the consumption figures in [01]; they are estimates with stated assumptions.
eia.gov/electricity/monthly - Interlune (Seattle; founders Rob Meyerson, Gary Lai, Harrison Schmitt). DOE Isotope Program agreement: three liters of lunar helium-3 by 2029, the first government purchase of a non-terrestrial natural resource. Maybell Quantum: thousands of liters annually, 2029–2035. Bluefors: up to 10,000 L/yr, 2028–2037. Cold Capture demonstration of 99% pure helium-3 from domestic helium (July 2026) with ~$500M in binding purchase agreements reported. Price ~$20M/kg per CEO Meyerson.
interlune.space (DOE) · interlune.space (Maybell) · Cold Capture release - NASA / University of Wisconsin Fusion Technology Institute / USGS. Lunar helium-3 resource assessments: ~1 million tonnes total lunar inventory (order-of-magnitude, NASA-cited); 20–50 ppb concentrations in titanium-rich mare regolith; ~25 tonnes of He-3 estimated to supply one year of US electricity via D-He3 fusion; Mare Tranquillitatis alone estimated at 15,000+ tonnes. USGS scale caveat: ~100,000 to 1 million tons of regolith processed per kilogram recovered. NASA CLPS/SBIR contracts fund Interlune extraction payloads for 2028 lander missions.
Wisconsin FTI UWFDM-879 · NASA NTRS (lunar He-3 mining concepts) · SpaceNews (USGS) - Helion Energy / Microsoft / Commonwealth Fusion Systems. World's first fusion power purchase agreement (May 2023): Microsoft to buy ≥50 MW from Helion's first plant, targeted 2028, Constellation as power marketer. Helion's D-He3 fuel cycle with in-house helium-3 breeding and direct electricity recovery; Orion plant groundbreaking at Malaga, WA, July 2025. CFS-Google 200 MW agreement for the early 2030s.
helionenergy.com · CNBC - Morgan-Wardenclyffe record. Morgan's $150,000 investment (1901) for transatlantic wireless communication; funding declined after Marconi's 1901 transatlantic signal and Tesla's disclosure of the power-transmission goal; tower dynamited for scrap, 1917. The "where do we put the meter?" line is the traditional attribution (used by PBS American Experience among others) but historians note Morgan's July 1903 letter declining further advances has never been published verbatim and the quote is likely apocryphal. Tesla's 1903 letters to Morgan, including "go to pots," are documented in the Tesla-Morgan correspondence; see W. Bernard Carlson, Tesla: Inventor of the Electrical Age (Princeton, 2013).
Long Island Press · Wardenclyffe Tower (overview) - Elon Musk, X post, February 8, 2026, and subsequent reporting. SpaceX shifts focus from Mars to a self-growing lunar city, citing Moon launch windows every ~10 days with 2-day transit versus Mars alignments every 26 months with 6-month transit; Moon city achievable in under 10 years versus 20+ for Mars. SpaceX targeting an uncrewed lunar landing March 2027; ~$4B NASA HLS contract for Artemis III.
SpaceNews · CNN · TIME - Generation shares, capacity factors, and resource potentials. EIA Electric Power Monthly and Energy Explained (2025 data): nuclear ~20% of US generation at 92%+ capacity factor; wind ~11% at ~34%; hydro ~6% at ~34% with ~90% water-to-wire conversion; solar ~23% capacity factor, +34.5% generation growth in 2025; geothermal <1% at ~65% capacity factor. Physics limits: Betz (59.3%, wind) and Shockley-Queisser (33.7%, single-junction PV). Resource scales: ~44–47 TW continuous terrestrial heat flow; ~173,000 TW continuous solar delivery to Earth; NREL US wind technical potential ~10 TW onshore; DOE Enhanced Geothermal Shot. D-He3 energy yield per [14].
eia.gov/energyexplained · energy.gov (Enhanced Geothermal Shot) · nrel.gov
About the Author
