01Confirmed
The installed base
1,034,738 subscribers (+10%)
Generation is moving from a few very large places to a very large number of small ones. The network already exists at scale — 8,732 MW of it, connected.
The Long Run Research DepartmentResearch status: Active
The Long Run
Wall Street sees a residential solar company.
We see a distributed energy network.
Anybody can be bullish for a day.
We're here for the LONG/RUN.
$LONG / $RUN
The meme is long.
The thesis is run.
Welcome to the research department
We started researching as a joke.
Unfortunately, we kept reading.
- 01Batteries.
- 02Virtual power plants.
- 03Grid constraints.
- 04AI power demand.
- 05A network of more than a million customers.
Somewhere along the way, the joke became a thesis.
So naturally, we made a memecoin.
The meme is long.
The thesis is run.
The thesis in one sentenceThe Long Run's interpretation
The LONG/RUN thesis is that Sunrun may gradually become more valuable as an orchestrator of distributed energy infrastructure than the market gives it credit for as a residential solar company — but proving that thesis requires the battery network, grid-services business and emerging data-centre opportunities to generate meaningful economics while the core business survives policy, financing and execution pressure.
Not “Sunrun is secretly an AI company”. The infrastructure already exists. The open question is how valuable the new monetisation layers become.
§ 01The thesis at a glance
Six strands, one argument.
Each is tracked separately, because they can fail separately. Every figure below is reported by the company or a named third party, and links to its source.
02Confirmed
Battery storage
74% attachment · 4.6 GWh
Solar generates when the sun allows it. Storage is what makes that energy controllable, and attachment is at a record.
03Third-party
Grid constraints
5+ years median to connect
New transmission is slow. Projects reaching operation in 2025 took a median of more than five years from interconnection request.
04Third-party
AI power demand
6.7–12% of US power by 2028
Compute has to be plugged into something. Data centre load is a new class of buyer arriving in a market that was not built with it in mind.
05Confirmed
Virtual power plants
425 MW · ~18 GWh dispatched
Not a future capability. Eighteen programmes ran in 2025 and dispatched real energy into real grids.
06Risk
The bear case
19,793 additions (−31% YoY)
Solar installed fell 23% and storage 15% year over year. This sits here on purpose — it is the strand most capable of breaking the rest.
§ 02Constraint
The grid has
a problem.
Electricity demand is changing faster than traditional infrastructure can always be built.
AI data centres require power. Electrification requires power. Grid interconnections take time.
- AI data centres require power
- Electrification requires power
- Interconnections take time
Fig. 01 — distributed supply against concentrated demand
The question
What if millions of distributed batteries and connected homes become part of the infrastructure?
§ 03The network
One roof is a purchase. A million is infrastructure.
The progression below is the whole mechanism. Nothing in it is novel on its own; the argument is entirely about what happens at the last two steps.
- N-011/6
Home
One roof. One meter. Statistically irrelevant.
- N-022/6
Solar
Generation arrives where the demand already is.
- N-033/6
Battery
Power becomes storable, and therefore schedulable.
- N-044/6
Distributed network
Individually small. Collectively dispatchable.
- N-055/6
Grid services
Capacity coordinated from assets already connected behind the meter.
- N-066/6
Power-hungry economy
Compute, electrification, and everything downstream of them.
Your neighbour's battery doesn't look like infrastructure. A million of them do.
Evidence // 2025 distributed power plant programmes
This isn't a future power plant.
It dispatched last year.
18
Programmes
100k+
Participants
~18 GWh
Dispatched
Sunrun's California distributed power plant alone now spans more than 80,000 households and 110,000 batteries, with up to 425 MW of peak dispatchable capacity.
What is demonstrated is the capability. What is not settled is the economics — how much this earns, and for how long. That distinction is the whole thesis, and we are not going to blur it.
§ 04Thesis monitor
What the desk is watching.
Every strand of the thesis, how well evidenced it currently is, and the source behind it. Two of the ten rows are marked RISK. They stay there.
LONG/RUN thesis monitor
Reviewed by handThesis factorStatusEvidenceReviewed
Million-home installed base
The network exists at scale. This is the foundation everything else in the thesis is built on.
ConfirmedReviewed07 SEPT 2026Storage-first transition
Batteries are what make the fleet controllable, and the mix keeps shifting toward them.
ConfirmedReviewed07 SEPT 2026VPP capability
Not a concept. Distributed power plant programmes are operating and dispatching today.
ConfirmedEvidenceCalifornia DPP: 80,000+ homes, 110,000+ batteries, up to 425 MW
Reviewed07 SEPT 2026AI power-demand macro
The demand backdrop is independently documented. Confirmed as a trend, not as a benefit to any one company.
ConfirmedEvidenceData centres ~4.4% of US electricity in 2023; projected 6.7–12% by 2028
Reviewed07 SEPT 2026Hyperscaler monetisation
An announced Voltus arrangement plus a non-binding three-party framework. Real, early, and not yet material revenue.
DevelopingEvidenceVoltus arrangement announced; 16.8 GW framework is non-binding
Reviewed07 SEPT 2026Distributed AI compute
Compute nodes at homes with Sunrun energy systems. A pilot, carrying no revenue expectation.
ExperimentalReviewed07 SEPT 2026Subscriber trajectory
The network must keep compounding for the infrastructure argument to strengthen. Additions are falling.
WatchEvidenceQ2 additions 19,793 (−31% YoY); solar 174 MW (−23%); storage 332 MWh (−15%)
Reviewed07 SEPT 2026Cash generation / financial execution
Installed volumes are declining while the model still requires funding ahead of returns.
WatchEvidenceInstallation volumes down across both solar and storage in Q2 2026
Reviewed07 SEPT 2026Capital costs
The model depends on continuous external financing, and its cost is set entirely outside the company.
RiskReviewed07 SEPT 2026Policy environment
Incentive and tariff changes, plus California concentration, are the two things most capable of overwhelming everything above.
RiskReviewed07 SEPT 2026
ConfirmedReported and citedDevelopingReal but early — outcome unprovenExperimentalPilot only — may never become materialWatchMoving against the thesisRiskCould break the thesis outright
Statuses describe the state of the evidence, not a view on any security. “Confirmed” means the underlying fact is reported and cited — it does not mean the thesis is proven. Nothing here is market data, a price signal, or investment advice. This table is reviewed by hand on the dates shown — it is not a live feed.
§ 05Invalidation
How we could be
completely wrong.
Stated in advance, so it cannot be quietly revised later. If any of these happen, it gets written up with the same prominence as anything supporting the thesis.
Read it in full- 01Capital intensity. The model requires continuous external financing to fund systems before they generate returns.
- 02Cost of capital. Financing costs are set by people with no interest in this thesis, and they feed directly into unit economics.
- 03Financing dependency. Access to tax-equity structures and securitisation markets is a precondition, not a detail.
- 04Policy exposure. Incentives, net-metering and tariff design can change, and have changed, faster than an installed base can adapt.
- 05Execution risk. Installation and servicing at scale is an operational business with real margins for error.
- 06Subscriber growth. Additions fell 31% year over year in Q2 2026, directly attacking the compounding the thesis assumes.
- 07Competition. Nothing about aggregating home batteries is proprietary — utilities, OEMs and other aggregators can pursue it too.
- 08California concentration. A large share of the customer base and the most advanced distributed power plant programme sit in one state, exposed to one regulatory regime.
The deskStill reading
§ 06Research department
Welcome to the department.
An anonymous research desk with an unreasonable interest in electricity, and a token attached to it for reasons that made sense at the time.
Research
“We read the filings so trench traders don't have to.”
Primary sources, in full, with the boring parts included. The boring parts are usually where the thesis is.
Thesis
“Good news gets recorded. Bad news gets recorded too.”
Every position we publish carries the conditions that would break it. A thesis you cannot falsify is not a thesis.
Memes
“Financial research should not be this entertaining.”
The distribution layer. It is how the reading gets read, and we are not going to pretend otherwise.
§ 07Questions
Plainly answered.
No jargon, no roadmap, no promises. If a question is not answered here it is probably answered on the verify or thesis page.
What is The Long Run?
An internet-native research project about distributed energy — batteries, distributed power plants, grid constraints and the electricity demand arriving with AI — with a memecoin attached to it. The research is sincere and sourced. The coin is a coin.
What is $LONG?
The project's token. It is a memecoin. It is not equity, it is not a claim on any company, and it does not entitle you to anything. It does not exist yet — verify the official contract on the verify page before touching anything claiming to be it.
What is $RUN?
It depends on the context, and we keep the two apart. In research context, RUN refers to Sunrun Inc. and to NASDAQ: RUN, its ordinary listed equity. In LONG/RUN trading context, $RUN refers specifically to the canonical Robinhood Stock Token representing economic exposure to the underlying Sunrun security. The Stock Token is a distinct on-chain instrument — it is not ordinary Sunrun equity. The canonical contract address is published on our verify page.
So is RUN the company or the token?
Both names exist and we keep them apart. In the research, RUN and Sunrun mean Sunrun Inc., the company listed on NASDAQ under the ticker RUN. In a trading context, $RUN means the Robinhood Stock Token. When it matters, we say which one we mean.
Why LONG/RUN?
Because the pair reads as a sentence. Long RUN. The long run. It is the rare case where the joke and the thesis are the same three words, and we were not going to walk away from that.
What is the thesis?
That Sunrun may gradually become more valuable as an orchestrator of distributed energy infrastructure than the market credits it for as a residential solar company — provided the battery network, grid services and emerging data-centre opportunities generate real economics while the core business survives policy, financing and execution pressure. Read it in full, with sources, on the thesis page.
Is buying $LONG the same as investing in Sunrun?
No, and this matters. $LONG is a memecoin attached to a research project. Sunrun Inc. is a listed company we have no affiliation with. Buying a memecoin is not investing in a company, and nothing on this site is an offer, a recommendation or financial advice.
Where can I verify official links?
The verify page. Every official address and channel is published there and nowhere else. If a link did not come from that page, treat it as hostile.
Where can I read the research?
The thesis page for the long-form argument, the research feed for individual notes, and the terminal for the monitor at a glance. Every quantitative claim carries a numbered source you can open yourself.
The pair
LONG/RUN
Anybody can be bullish for a day. The research keeps going either way.
Research last updated // 07 SEPT 2026