← All notes

After a decade of connecting systems that were never designed to talk to each other, you start to notice where things actually break. It is almost never the component. The pump works. The sensor works. The database works. What fails is the space between them — the handoff, the timing, the assumption one system made about another that turned out to be wrong.

I am a software engineer. For more than ten years my work has been systems integration: making independently built subsystems behave like one coherent thing under real operating conditions. I am not an electrochemist. I mention that early because it matters for everything that follows.

When I started reading seriously about stationary energy storage, I expected the hard part to be the chemistry. What I found instead was a familiar shape — a field where the components are improving quickly and the integration between them is treated as a solved afterthought. That is the pattern I have spent my career inside. So I kept reading.

Where grid storage actually struggles

The public conversation about batteries is almost entirely a conversation about cells. Energy density, cycle life, chemistry. Those things matter. But a battery installation in the field is not a cell — it is a cell stack wrapped in power conditioning, thermal management, charge-cycle logic, and state estimation, all expected to cooperate for years.

Consider one well-documented fact: lithium-ion cells lose a meaningful fraction of their usable capacity in the cold, and prolonged cold can crack the cathode and permanently reduce capacity (SLAC National Accelerator Laboratory). Operators in cold regions know this and compensate by oversizing packs and adding active heating — more components, more parasitic draw, more seams.

The question that interests me is not “why does cold hurt lithium?” That is a chemistry question, and chemists have answered it well. The question is: why do we keep architecting grid storage as though the conditioning layer around the cell were trivial? In the systems I have worked on, the layer everyone assumes is trivial is usually where the failures live.

“Failure comes from a failure to imagine failure.”Josh Wolfe, Lux Capital — Farnam Street

A lot of field failure in storage looks, to me, like failure that was never imagined at the integration layer.

Why sodium-ion, and why now

I am not here to argue that sodium-ion is better chemistry than lithium-ion. That is not my claim to make. What I will say is that sodium-ion has reached a specific and useful moment.

The cells are arriving commercially. CATL's Naxtra sodium-ion line is moving toward full-scale mass production by the end of 2026 (Graphene Uses), and the company has now introduced a sodium-ion cell built specifically for grid and commercial storage, rated for 2-to-8-hour utility duty (ESS News). In the United States, Peak Energy has raised a $55 million Series A to build domestic sodium-ion systems and a planned giga-scale factory (Peak Energy). Sodium-ion cells also operate across a wider temperature band than typical lithium chemistries: CATL's storage cell is rated for a -40°C to 70°C range (ESS News).

Here is the part I find genuinely interesting. A new substrate is becoming commercially available faster than the architecture around it is being rethought. The cells are new; much of the conditioning and control logic they get dropped into still carries assumptions inherited from lithium. The moment a substrate is available but architecturally unexplored is exactly the moment a systems integrator has something to contribute.

Architecture over chemistry

This architecture sits on the shoulders of years of cell development done by other people. My contribution is not at the cell level. It is at the system level — how the pieces are sequenced and how energy is handled between them.

The system Attune Labs is designing pairs commodity sodium-ion cells with a regenerative power-conditioning stage. Every battery installation loses some energy during conversion and switching; that loss is normally dissipated as heat. The conditioning stage is designed to recover a portion of those losses before they leave the system, rather than letting them disappear.

I want to be exact about what that is and is not, because the difference is the entire point.

The system requires solar or grid input. It reduces losses — it does not create energy. It is not self-powering, and claiming otherwise would be scientifically incorrect. Recovering a loss is not the same as generating a surplus. The open research question is whether the energy recovered at the conditioning stage exceeds the energy the conditioning stage itself consumes — its own parasitic load. If the recovery does not clear that bar, the architecture has to be redesigned. I do not yet know the answer. That uncertainty is not a marketing problem to be smoothed over; it is the actual research.

Even the hardware sizing is still a variable. An early estimate suggests something on the order of 148 cells to support a 100 kWh configuration, but the real number depends on topology and operating profile, and pinning it down is itself part of the work — not a settled specification.

In June 2026, Attune Labs submitted a first-gate NSF SBIR Phase I Project Pitch (NSF SBIR). I want to be precise about what that is: it is a submission to the beginning of a formal evaluation process, not an award, not a grant, and not an endorsement. What it represents is a discipline — the act of turning a hunch into a researchable hypothesis with defined success and failure criteria.

What a software engineer brings to hardware

The honest version of my qualification is narrow, and narrow is the point.

A decade of systems integration trains a specific habit of mind: thinking in interfaces rather than components, looking for failure at the boundaries between subsystems, and refusing to assume that two things that each work in isolation will work together. Those are exactly the instincts that the integration layer of an energy system rewards — and exactly the instincts that get underweighted when a storage system is designed cell-first and everything else is treated as plumbing.

What I do not bring is electrochemistry, materials science, or power-electronics design at the device level. That work belongs to people trained in it, and Phase I research is structured to engage that expertise rather than pretend I already hold it. What I provide is the integration thesis and the direction: a specific architecture, built to test a specific question.

“There's a narrative, and you need to discern what parts of it are authentic.”Sarah Guo, Conviction — Forbes

I find that boundary clarifying rather than limiting. The authentic part of mine is that battery storage has a large, under-examined integration surface, and integration is the thing I actually know how to do.

This is a beginning, not a finish line. The cells are someone else's achievement. The chemistry is not in question. The open question is architectural — and that is a question worth building a company to answer carefully, in public, one verified result at a time.

Frequently asked questions

Is this an over-unity or self-powering system?
No. The system requires solar or grid input. The regenerative stage reduces losses within the round-trip cycle; it does not generate net energy. A system that created energy from nothing would violate physics, and Attune Labs makes no such claim.
Does the technology work yet?
No. Attune Labs is pre-prototype and in the R&D stage. There is no finished product and no revenue. The NSF SBIR Phase I Project Pitch is a submission to investigate whether the architecture delivers a net efficiency gain — not a demonstration that it does.
Why sodium-ion instead of lithium-ion?
Not because the chemistry is universally superior. Sodium-ion offers a wider operating-temperature range, lower cost per cell, and no dependence on the lithium supply chain, which makes it a useful substrate for the architectural experiment Attune Labs is running. Lithium-ion remains excellent for many applications.
What is the NSF SBIR Phase I Project Pitch?
The NSF SBIR program funds early-stage small businesses doing research with commercial potential. The Project Pitch is the program's first-gate submission, used to determine whether a team is invited to submit a full proposal. Attune Labs submitted its pitch in June 2026.
Why is a software engineer building an energy-storage company?
Because the part of the problem that drew me in is an integration problem — how independently built subsystems behave as one system under real conditions. That is the domain I have worked in for over a decade. The chemistry belongs to chemists; the architecture is where I can contribute.

Sabir Foux is the founder of Attune Labs Inc., a Delaware corporation based in Atlanta, Georgia, working on sodium-ion stationary energy storage and regenerative power conditioning.