Energy · Deep dive

Torus

Flywheel-plus-battery hybrid storage from Lehi, Utah — modular power plants that pair a spinning steel rotor's millisecond response with LFP battery duration, sold to utilities, data centers, and industry and aggregated into Rocky Mountain Power's Wattsmart virtual power plant.

emerging

The question that decides it: Torus's entire differentiation is the Nova Spin flywheel bolted in front of commodity LFP batteries. Does that layer actually pay — does absorbing the millisecond-scale load swings of AI GPU clusters and shaving high-frequency cycling off the battery measurably lower lifetime cost per MWh versus what Tesla and Fluence already do by oversizing inverters and LFP containers — or do the flywheel's ~12-minute energy duration, self-discharge, and steel-rotor bill of materials make it an expensive UPS welded onto a subscale BESS, leaving Torus to fight integrators with 10-40x its manufacturing capacity on price it cannot win?

My take ★★ Very interesting ★ Interesting − Maybe ✗ Not interesting

Snapshot

Torus is a Lehi, Utah company building modular hybrid power plants that weld two storage technologies together: the Nova Spin, a steel flywheel in a vacuum chamber that responds to grid signals in under 250 milliseconds, and the Nova Pulse, a lithium-iron-phosphate battery rack that carries four-to-eight-hour duration. Founded in 2021 by former Pluralsight product chief Nate Walkingshaw, it has raised roughly $304 million (PitchBook, 2026) — including $67 million led by Origin Ventures in April 2024 and $200 million from Magnetar in September 2025 — and claims 230+ US deployments (September 2025), a 70 MW demand-response partnership with Rocky Mountain Power, and a new 545,000-square-foot Salt Lake City factory running since January 2026. It matters because it is the best-funded test in a decade of whether flywheels belong on the grid — a category with a forty-year record of elegant engineering and commercial failure.

Founding story

Walkingshaw is a serial Utah operator, not an energy man. He began as an EMT, decided ambulance patient-evacuation equipment was dangerous, and founded Paraslyde in 2004 to fix it; Stryker Medical acquired it and he ran global R&D for Stryker’s EMS division. He then founded Brightface, co-authored O’Reilly’s Product Leadership, and in January 2015 became Pluralsight’s Chief Product Officer — expanded in 2016 to Chief Experience Officer over product, engineering, and content — through the company’s 2018 IPO. The origin story Torus tells is agrarian: on his family’s Salt Pyne tree farm in Sandy, Utah, Walkingshaw wanted to store micro-hydropower to run irrigation pumps, found that car batteries died in the cold, and went down a flywheel rabbit hole (KSL, 2023). He founded Torus in 2021 with Gilbert Lee, described in Utah press as an energy technologist; Lee keeps an unusually low profile for the technical co-founder of a hard-tech company. The honest read: this is a product-and-go-to-market founder who chose the single hardest category in cleantech hardware — one where the graveyard (Beacon, Pentadyne, Active Power) is full of teams with deeper rotational-machinery pedigrees.

How it works

The Nova Spin is a multi-ton steel rotor — press coverage cites around seven tons (Autonocion, 2025) — levitated on magnetic bearings inside a hermetically sealed vacuum chamber. A motor-generator spins it up to store energy kinetically and brakes it to give the energy back. Because there is no chemistry, response is under 250 milliseconds, discharge runs at roughly 10C — about twenty times the power-to-energy ratio of a standard BESS — and Torus rates it for 25,000 cycles with no capacity fade and a 25-year life, built from ~95% recycled or recyclable steel. Round-trip efficiency is 85-87% (Torus CEO, Battery Technology, 2025). The catch is physics: a full charge-discharge cycle takes about 12 minutes, meaning the flywheel stores minutes of energy, not hours, and standing losses (flywheel self-discharge industry-wide runs several percent per day at best) make it useless for shifting solar to evening. So Torus pairs it with the Nova Pulse, a conventional LFP battery system scalable to four-to-eight-hour duration. The design logic: the flywheel eats the violent, high-frequency power swings — AI GPU clusters can swing ~90% of their draw in milliseconds — so the batteries cycle gently and last longer, while the system runs natively at 800V DC, matching modern server racks. Units sit at the grid edge or on-site, are wrapped in Torus’s software and an AI-driven cybersecurity appliance (unveiled November 2024), and are orchestrated fleet-wide — the company claims 99.9% uptime across deployments (2025).

Product and business overview

Four named components. Nova Spin, the flywheel, sold for grid stability, power quality, and backup. Nova Pulse, the LFP BESS, for duration. A software and orchestration layer that aggregates units into what Walkingshaw calls a “distributed utility” or grid operating system, including bidding capacity into utility programs. And the cybersecurity appliance, pitched at utilities and military customers — KSL and Green Prophet (2025) note military interest. The original product was residential: the Torus Station, distributed through a November 2023 Vivint Smart Home partnership in Utah. The company has since pivoted decisively upmarket to commercial, industrial, utility, and data-center customers — the Vivint channel gets little mention after 2024. Manufacturing is vertically integrated in Utah: a 37,000-square-foot South Salt Lake facility, plus GigaOne, a 545,000-square-foot Salt Lake City plant online since January 2026 that now builds nearly the entire storage pod in-house (company, July 2026).

Business model and pricing

Revenue is booked three ways, none with published price points. First, hardware sales of Nova Spin and Nova Pulse systems to C&I and utility buyers. Second, energy-as-a-service: a subscription under which Torus owns, finances, and maintains the equipment and the customer buys outcomes — resilience, demand-charge reduction, backup — with no upfront capital (company materials, 2025). The Magnetar money, from an asset manager rather than a classic VC, is plainly aimed at financing this balance-sheet-heavy EaaS fleet. Third, grid-services revenue: under Rocky Mountain Power’s Wattsmart Battery program (commercial approval March 2025), Torus-equipped sites are paid for dispatchable demand response, with a January 2025 MOU targeting up to 70 MW of C&I capability within 12-18 months. No $/kWh system price has been published; the uncomfortable benchmark is that grid-scale flywheels have historically cost on the order of $400/kWh-plus against sub-$200/kWh 4-hour LFP systems (NREL-cited ranges, 2024-25), which is why the flywheel is sold as a power device, not an energy device.

Traction over time

Dated wins beyond the table: a late-2024 deal placing ~26 MWh of VPP-connected Torus Stations across Gardner’s Utah commercial real estate portfolio (Energy-Storage.news), and TIME naming Nova Spin a Best Invention of 2024 (October 2024). What is missing is equally clear: no disclosed revenue, no MWh-deployed total, no named data-center customer, and “230+ deployments” is a site count that mixes homes, buildings, and pilots — a number built for press releases, not diligence.

Market analysis

The tailwind is real and enormous. US battery storage installations hit a record 18.9 GW in 2025, up 52% year over year (Wood Mackenzie/ACP, March 2026), with ~93 GW and ~500 GWh forecast over 2026-2031 — though Wood Mackenzie projects an 11% contraction in 2026 and 8% in 2027 as tariffs and supply chains bite. The structural driver is AI: data-center load growth has utilities begging for fast, sited capacity, and PacifiCorp’s Wattsmart program shows regulated utilities will pay for aggregated behind-the-meter response. Torus’s specific wedge — the flywheel — addresses a far smaller pond: the global flywheel energy storage market is estimated in the low single-digit billions with ~12-13% CAGR (Grand View, Evolvance, 2025), and the US DOE’s 2024 long-duration storage assessment dropped flywheels from its ten-technology shortlist entirely, citing cost, safety, and space. Torus is therefore best read as a claim on the giant BESS market wearing a flywheel differentiator, not as a flywheel company riding a flywheel market.

Competitive intel

Scale asymmetry defines the set. Tesla ships Megapacks from a ~40 GWh/year Lathrop plant — one hundred times Torus’s capacity — and sets the $/kWh floor. Fluence (public; covered on this site) brings tens of GWh deployed and utility-grade software, and its persistent margin pressure is a warning about integrator economics even at scale. Against both, Torus argues the hybrid wins where power quality and millisecond response matter — AI load swings, microgrids, defense — and that Utah manufacturing dodges China-tariff exposure on finished systems (cells are another matter). Amber Kinetics and the ghost of Beacon Power (bankrupt 2011; its 20 MW/5 MWh plants cost ~$10,600/kWh of stored energy) are the flywheel record: technically sound, commercially dead. Piller, Vycon, and Active Power already own the profitable flywheel niche — rotary UPS for critical facilities — with decades of installs, attacking Torus’s data-center pitch from below. Base Power ($200M Series A, April 2025) shows the distributed-utility model works with plain batteries, no flywheel required. Torus’s genuine edges: speed-of-deployment for a utility partner headquartered in its backyard, vertical integration, and a security-and-sovereignty story for military buyers. None of those requires the rotor to be the winning reason.

History and evolution

• 2021 — Founded by Walkingshaw and Lee; flywheel concept born on the Salt Pyne tree farm.
• Nov 2023 — Vivint Smart Home partnership puts Torus Stations into Utah homes; residential-first strategy, Wattsmart-eligible.
• Apr 3, 2024 — $67M led by Origin Ventures (equity, note conversion, loan facility); David Bywater (ex-Vivint Solar CEO) and Origin’s Brent Hill join the board.
• Oct 2024 — Nova Spin named to TIME’s Best Inventions of 2024; full C&I product suite (Nova Spin, Nova Pulse, cybersecurity) unveiled at Utah’s 47G Zero Gravity Summit.
• Late 2024 — ~26 MWh VPP deal across Gardner’s commercial real estate portfolio; pivot upmarket is complete in all but name.
• Jan 27, 2025 — Rocky Mountain Power strategic partnership; MOU for up to 70 MW of C&I demand response.
• Mar 19, 2025 — Commercial approval into RMP’s Wattsmart Battery program.
• Sep 9, 2025 — $200M from Magnetar; 230+ deployments claimed; GigaOne factory announced (Bloomberg, BusinessWire).
• Jan 2026 — GigaOne comes online.
• Jul 22, 2026 — Company reports weekly production records at GigaOne, output up 76% in six months, headcount ~300.

What people say

The case for. TIME put Nova Spin on its 2024 Best Inventions list. Trade press (Battery Technology, 2025) credits the hybrid’s 99.9% uptime claim and the logic of using flywheels to absorb AI’s millisecond load swings — a real, documented problem for grid operators. Magnetar writing a $200 million check after diligence is the strongest external signal, and PacifiCorp’s willingness to build a 70 MW program around a four-year-old startup (2025) says the utility saw working hardware. Glassdoor reviewers (3.8/5 across a small base, 2025) praise mission, transparency, and the company-wide 100-day demo cadence.

The complaints. The technical criticism is blunt. CleanTechnica’s Michael Barnard (April 2026) surveyed fifty years of flywheel history and concluded the sector survives on “niches, hybrids, and investor patience,” calling Torus specifically a battery-heavy bet “with a side helping of flywheels” that only looks large because it is not a pure flywheel play — the polite version of saying the differentiator is marketing. The DOE’s 2024 safety strategy flags large-rotor failure containment as a core problem and excludes flywheels from grid-scale energy roles. Flywheel self-discharge — up to several percent per day even in good designs, far worse in bad ones — is physics, not engineering debt. On the employee side, Glassdoor’s small review base includes pointed negatives: mentions of the CEO as a problem, a “bully culture” comment, crunch around the 100-day events, and concerns about product delivery — worth weight precisely because the base is only a handful of reviews and they still surfaced. And the disclosure gap is itself a criticism: no revenue, no MWh totals, no named data-center customers, undisclosed valuations at every round.

Outlook: the open question

Torus works if the flywheel layer provably lowers lifetime cost per MWh for high-volatility loads — demonstrated by third-party-verifiable battery-degradation and uptime data from the 70 MW Rocky Mountain Power fleet and by a named hyperscale or colocation customer choosing the hybrid over a Megapack/Fluence bid on economics, not novelty. The bull case: AI load volatility is a new, genuinely unsolved grid problem; a 10C, 800V-native, 25,000-cycle buffer in front of LFP is a defensible answer to it; Magnetar’s capital funds an EaaS fleet whose recurring grid-services revenue compounds; and Utah gives Torus a friendly regulated utility, a manufacturing base, and a defense-adjacent customer set no coastal rival has. The bear case: the flywheel stores twelve minutes of energy, leaks charge daily, and adds a seven-ton precision-machined rotor to a bill of materials competing against the most brutally cost-optimized product in cleantech; every previous flywheel-grid company died or shrank into the UPS niche that Piller already owns; and if the hybrid premium doesn’t clear, Torus is a ~400 MW/year battery integrator fighting Tesla’s 40 GWh — at which point the $304 million bought a very good story. Watch three things through 2027: whether the 70 MW Wattsmart buildout hits its 12-18-month MOU timeline with published performance data; whether GigaOne’s ramp converts to disclosed revenue or MWh; and whether a single named data-center anchor signs. Absent all three, the CleanTechnica read — batteries with a side of flywheel — becomes the verdict.

How a challenger would attack it

Deliver the flywheel’s benefit without the rotor. Torus’s pitch reduces to one claim: something fast must sit in front of the LFP racks to absorb AI’s millisecond load swings. A challenger attacks the premise — oversized inverters, supercapacitor modules, or simply more parallel LFP strings deliver the same buffering with commodity parts, no seven-ton precision-machined steel, no vacuum chambers, no daily self-discharge. Tesla and Fluence already do versions of this at 10-100x Torus’s ~400 MW annual capacity, so the sharper attack is the model, not the hardware: run Base Power’s playbook against Torus’s home-state beachhead — plain batteries, energy-as-a-service, aggregated into utility demand-response programs like Wattsmart — at a bill of materials Torus’s hybrid can’t match. The disclosure gap is the sales weapon: Torus publishes no revenue, no MWh totals, no $/kWh price, and no named data-center customer, while “230+ deployments” mixes homes and pilots. A challenger that shows up to the same utility and colocation RFPs with third-party-verified degradation data and a public price per kWh forces Torus to defend a premium it has never quantified. And the profitable flywheel niche Torus gestures at — millisecond ride-through for critical facilities — is already owned by Piller and Vycon, with decades of installs; there is no undefended ground on either flank.

Same playbook, new buyer

The transferable asset isn’t the flywheel — it’s the stack Torus assembled around it: vertically integrated US manufacturing, an energy-as-a-service balance sheet (Magnetar’s $200M is financing, not product validation), a regulated-utility VPP partnership, and a cybersecurity-and-sovereignty story. That stack re-aims at buyers where the flywheel premium actually clears. Defense is the clearest: military microgrids pay for domestic supply chains, EMP-and-cyber-hardened systems, and power quality — the file notes military interest in the security appliance, and DoD procurement tolerates costs the C&I market won’t. Second, industrial power-quality buyers — semiconductor fabs, precision manufacturers — where a millisecond sag costs a production run and the rotary-UPS incumbents sell hardware, not an orchestrated EaaS fleet. Third, the Wattsmart template itself: mid-size regulated utilities in inland states want a local partner to stand up C&I demand-response programs, and Torus proved a four-year-old company can be that partner — a replicator running the same play with cheaper all-battery pods in other PacifiCorp-like territories faces no Torus response, because Torus’s capital, factory, and story are all committed to justifying the rotor.

Sources and further reading

• Torus Secures $67 Million in Venture Financing (BusinessWire, April 3, 2024)
• Torus Raises $200 Million to Accelerate Deployment of Modular Power Plants (BusinessWire, September 9, 2025)
• Magnetar Bets $200 Million on Torus Energy Storage Expansion (Bloomberg, September 9, 2025)
• $200 Million for Renewables-Friendly Flywheel Energy Storage (CleanTechnica, September 10, 2025)
• From Beacon to Amber, Flywheels Missed the Grid (CleanTechnica / Michael Barnard, April 19, 2026 — the flywheel-economics skeptic’s case, names Torus)
• Torus, Rocky Mountain Power sign MOU for 70-MW C&I demand response (Utility Dive, February 2025)
• Torus Receives Commercial Energy Storage Approval for Wattsmart Battery Program (BusinessWire, March 19, 2025)
• Torus Nova Spin: Best Inventions of 2024 (TIME, October 2024)
• Flywheel + Battery Storage Achieves 99.9% Uptime — Torus CEO interview (Battery Technology, 2025)
• Torus GigaOne Sets Production Records Six Months After Coming Online (Yahoo Finance / company release, July 22, 2026)
• Utah man creates company devoted to flywheel energy storage (KSL, 2023 — founding story)
• 2025 U.S. Energy Storage Installations Set New Record, Surpass 2024 by 52% (Wood Mackenzie, March 2026)