twillo

[twillo @twillo1]

@aayushtrades Yep, a DD vid would def be appreciated

[twillo @twillo1]

@philroberts

[twillo @twillo1]

@philroberts I married a Geordie, I love it up there. Been to Craigside, and so many other spectacular places. Highly recommend.

[twillo @twillo1]

$NUAI's behind the meter strategy relies heavily on relatively small gas turbines, which some say is a more robust powering strategy than relatively few high power turbines. Nothingwithstanding this, recent reports from turbine engineers that large data center loads changing rapidly is putting unprecedented stresses on turbines makes me wonder if gas turbines will prove to be a reliable power source for data centers.

https://t.co/SALc1dr3jr

[Reasonably Approximating 🇺🇸 🇺🇦 🔋 🅰️ @bert_gilfoyle]

$EOSE

Heard from a colleague who recently spoke with the chief turbine engineer at a utility. He said that large data center loads changing rapidly are putting unprecedented stresses on turbines in ways they weren't designed for. Turbines are the backbone of the U.S. grid, and when utility engineers with decades of experience are concerned, the industry should pay attention.

There is a lot of talk about adding "baseload" but that doesn't accurately characterize what the modern grid actually needs. It needs energy storage next to the large cyclic loads and also the generation, which can then increase the utilization of the grid. Use the capacity we already have by letting those turbines run unstressed at their maximum efficiency and use energy storage to handle the severe cycling.

What blows me away are the number of experienced people in this industry who don't understand this is the solution. That's in large part, I think, because they're unaware of the existence of the technology that can actually handle this safely.

[twillo @twillo1]

@GauchoRico BMNR not so hot, though

[twillo @twillo1]

@philroberts @OldMateEngineer Many thanks for your valuable insights.

[twillo @twillo1]

@OldMateEngineer @philroberts Awesome, many thanks for sharing this!

[twillo @twillo1]

@bert_gilfoyle This fellow argues that the pinch point in small modular reactors is the fuel:
https://t.co/BbmeeEaZl0

[Optimus @OptimusAAIG]

I would like to go deeper into the downbleding part because it sits right at the point where policy narratives and physical constraints collide. If you follow the advanced reactor story backwards, you eventually stop talking about capex or licensing timelines and start talking about whether the fuel system can actually deliver consistent material at scale, on schedule, with qualified specs.

HALEU is where this becomes impossible to ignore. Most advanced designs lean on enrichment above 5% and below 20% U 235 because it improves core economics and cycle length. That part is straightforward. The problem is that the West built an industrial stack optimized for LEU, then spent decades treating enrichment as a mature service rather than something you expand. So when HALEU demand shows up, it is not a question of whether the chemistry works. It is a question of whether you can produce enough SWU output, then deconvert, fabricate, transport, secure, and qualify the fuel repeatedly rather than as a one off.
Downblending shows up as the near term workaround. The US is allocating government owned HALEU to multiple developers through the HALEU Availability Program (DOE). That matters because it gets first cores and demonstrations across the line. But the more I looked at it, the more it felt like a bridge that can easily be mistaken for a foundation.

Downblending moves atoms from A to B. It does not create commercial learning curves, supply elasticity, or a competitive supply base. It is finite by definition, and it does not solve the industrial problem, which is repeatability at scale. If early deployments rely on stockpile allocations, you can get a working reactor and still not have a scalable system. That distinction is easy to lose when a FOAK project finally gets fueled and everyone wants to treat it as proof the bottleneck is behind us.
The Natrium delay is the example that made this feel concrete. TerraPower slipped the project timeline by around two years because HALEU was not available on the schedule the project needed (World Nuclear News) (Reuters). Funding did not fix it. Site readiness did not fix it. The constraint was fuel.

Even where the West is making real progress, the scale mismatch is still obvious. Centrus hitting 900 kg of HALEU production is meaningful (DOE). But kilograms are not the same world as tens of metric tons per year once you move beyond demos. This is where the bottleneck shows up: advanced nuclear is trying to ramp like a manufacturing story while the fuel cycle still looks like a pilot program.

West versus East matters here because it determines who already has the industrial base. Russia maintained and commercialized enrichment capacity for decades, including HALEU supply, while the West is rebuilding under time pressure and geopolitical constraints. That is why you now see public private infrastructure moves like DOE selecting Orano for HALEU related enrichment services and capacity buildout (DOE) (Orano).

So this is how I see it playing out. Downblending can carry early deployments through licensing and initial cores. It is not useless. But it is structurally incompatible with a civilian growth cycle because it does not build the industrial base required for repeat deployment. The replacement is boring and capital intensive: centrifuge cascades, licensed enrichment capacity, deconversion, fabrication lines qualified for HALEU, and long term offtakes that make the buildout financeable before reactor volumes exist.

If I had to name the highest signals for the next 12 to 24 months, it would be whether HALEU moves from allocations to bankable multi year offtakes, whether enrichment projects move from announcements to licensed construction and measurable SWU additions, whether fabrication and deconversion scale in parallel rather than lagging, and whether developers publish fuel strategies that work without government special cases. If those milestones slip, the advanced reactor timeline slips with them.

On tickers, I still separate uranium beta from the actual tickers. $LEU is the cleanest public exposure to Western HALEU enrichment progress. $BWXT matters because fuel forms and qualified fabrication are where practical constraints live. $ASPI fits as a higher risk option to create new enrichment capacity tied to future HALEU demand. $CCJ matters less as spot uranium and more as a credible counterparty in a Western fuel stack being rebuilt under security of supply logic.

[twillo @twillo1]

@OldMateEngineer @CompoundingTony No foul in being wrong. Staying wrong another matter. Being open minded and flexible and willing to see other points of view prevents you from staying wrong.

[twillo @twillo1]

@OldMateEngineer @CompoundingTony Exellent work, very thoughtful. I like that you question assumptions and invite constructive criticism. Well done!

[twillo @twillo1]

@CharlesSteadman @chuckcook @AIDRIVR I've been wondering how FSD would do on Hwy 1 in Northern California, this video gives me a pretty good idea.

[twillo @twillo1]

Have you tried FSD along Hwy 1, the twisty road that hugs the west coast? It's one of those roads that requires a lot of driver engagement, with hairpin turns and steep dropoffs, with aggressive drivers tailgaiting conservative drivers ... how FSD would manage in that sort of environment

[🔋Tony🔋 @CompoundingTony]

Some clarifications on the recurring bear case around EOS Energy. $EOSE

I’ve seen the latest iteration of the bearish case on EOS Energy circulate again. It largely repeats the same claims that have already been made multiple times over the past year. Rather than debating headlines or framing, it’s more useful to address the underlying assumptions directly and put them in the correct technical and economic context.

1. Lithium prices and the “zinc no longer makes sense” argument
The core assumption here is that falling lithium prices automatically invalidate zinc-based storage. That logic does not hold for grid-scale energy storage.

Grid storage is not optimized on raw material spot prices or energy density per kilogram. It is optimized on total system cost over decades. That includes degradation, cycle life, safety systems, insurance, parasitic loads, replacement cycles and regulatory constraints.

Lithium pricing today is also far from structurally stable. It is increasingly exposed to tariffs and geopolitical pressure as the US and China weaponize supply chains. More importantly, lithium systems degrade materially under multi-cycle daily use. Zinc hybrid systems do not. Over a 20–25 year asset life with frequent cycling, degradation dominates economics far more than initial cell cost.

On a lifetime delivered MWh basis, zinc remains competitive and in many long-duration use cases structurally cheaper. Short-term lithium spot prices do not change that reality.

2. Losses and cash burn
EOS is still loss making. That is not disputed. The error is treating a manufacturing ramp as if it were a steady-state business.

The majority of historical losses were incurred during factory buildout, automation integration and yield stabilization. That phase is now ending. Line 1 is operating above 90% utilization and running continuously. Automated subassembly is live, already reducing defect rates by 45% and cutting variation by 63%. These are precisely the inflection points that move a hardware manufacturer from cash burn to operating leverage.

Contribution margin turns positive in Q4 2025. Gross margin turns positive exiting Q1 2026. EBITDA follows as Lines 1 and 2 stabilize. These milestones are not aspirational. They are the mechanical result of yield, cycle time and scrap improvements that have already been confirmed.

3. Backlog quality and “fake orders”
The critique of the backlog fundamentally misunderstands how energy infrastructure projects work.

In this sector, projects progress from early-stage MOUs to awarded contracts to binding agreements only after permitting, interconnection approval, EPC alignment and financing. Letters of intent are standard at early phases. Treating non-binding LOIs as evidence of fraud ignores how virtually every large-scale energy project is developed.

More importantly, the pipeline demonstrates real project validation. Frontier increased its UK Cap and Floor submissions from 5 GWh to 11 GWh, with every EOS project advancing to the next round. Pipeline composition explains why this matters. Data centers now account for 22% of demand, and 64% of projects require 6 hours or longer durations, a segment where lithium-based systems face inherent economic and technical limits.

This shift is not theoretical. Over the past year, EOS has announced multiple new project awards and strategic customer engagements, including utility-scale deployments, data center–driven systems and long-duration projects that moved from early-stage development into awarded or advancing status. These additions materially changed both the scale and quality of the backlog compared to prior years.

Backlog is not revenue guidance. Conflating the two is a category error.

4. Customer creditworthiness
Another recurring mistake is assuming the project developer’s balance sheet determines payment risk.

In grid-scale storage, project owners, financiers and end counterparties are often different entities. Capital is typically provided through tax equity, infrastructure funds or utility-backed financing structures. Judging payment viability based on the size of a developer misunderstands project finance entirely.

This is not how large-scale energy assets are funded.

5. Hydrogen bromide “gas leak” narrative
The fire incident report continues to be misrepresented.

The NFIRS report did not document measured hydrogen bromide gas detection. The wording reflects a standard hazard identification process based on known chemistry, not real-time analytical confirmation. HazMat teams characterize potential hazards for responder safety. They do not perform lab-grade gas speciation.

EOS conducted over 1,000 air quality measurements during the incident and detected no hazardous readings. Operations were not shut down.

From a chemistry standpoint, hydrogen bromide gas formation is not a credible outcome in an aqueous zinc hybrid system under overcharge. At extreme overcharge, water electrolysis dominates, producing hydrogen which is vented through pressure relief mechanisms. Formation of hydrogen bromide gas would require destructive, high-temperature, dry acidic conditions incompatible with system design.

This behavior is precisely why aqueous zinc systems are classified as non-flammable and non-thermal-runaway chemistries. The headline framing added by the short report ignores basic electrochemistry.

6. Round-trip efficiency claims
Claims of sub-55% round-trip efficiency are outdated and not applicable to current Z3 systems.

Field data shows average round-trip efficiency around 85%, with best cycles exceeding 90%. On a system-level basis, these results are at least competitive with lithium-based solutions once parasitic HVAC loads are accounted for, which zinc-based systems avoid. One of the referenced installations is the Torrecillas site, delivered to a strategic utility customer. Performance feedback disclosed by EOS described the results as phenomenal.

Accepting the bearish claim here requires assuming that EOS fabricated performance data for the largest utility in the United States without objection, which strains credibility.

7. Executive turnover and alleged misconduct
Executive turnover during a manufacturing scale-up is not unusual, particularly when companies transition from development to industrial production. CFO mismatches are common in this phase. No regulator, lender or counterparty has alleged fraud. The Department of Energy conducted extensive technical and financial due diligence before approving its loan. Claims of deception remain unsupported speculation.

8. The actual trajectory
Line 1 operates at full scale in 2026. Line 2 comes online in spring and ramps faster by design. Automation partners can now deliver a full manufacturing line every 90 days. A second factory is actively being negotiated with multiple states. Demand continues to shift toward long-duration storage driven by data centers and grid constraints.

At conservative assumptions, 2026 revenue lands between $500M and $750M. 2027 expands toward $2B with positive margins and meaningful operating leverage. None of this is reflected in a ~$12 valuation.

Setbacks will occur. Ramp-ups never move in straight lines. But the direction is clear. EOS is transitioning from a capital-intensive development story into a scaled industrial manufacturer.

What stands out most is not the bearish conclusion itself, but how consistently outdated the underlying assumptions are. This analysis applies early-stage ramp logic, EV-era battery economics and hypothetical safety scenarios to a company that has already moved beyond them. Repeating the same arguments a third time does not strengthen the case. It simply confirms that the authors are not following the operational reality of EOS as it exists today.

[twillo @twillo1]

@Agrippa_Inv LHAI

[twillo @twillo1]

@JKeynesAlpha $HIVE

[twillo @twillo1]

@JKeynesAlpha $ELVA

[twillo @twillo1]

@WayneWhaley1136 I think he's actually an algo

[twillo @twillo1]

@CompoundingTony Each little piece of the puzzle makes the picture clearer, and helps to build the conviction needed to hold when the weather is stormy

[twillo @twillo1]

I think $TEM (https://t.co/mMAO5fTT4i) has enormous potential. The US health care system is ripe for disruption, especially considering it's role in unsustainable budget deficits. TEM founder and serial entrepreneur Eric Lefkofsky makes a compelling case for this in his interview with one of it's seed investors, Scottish Mortage: https://t.co/VYyBajiXdo