The hardware is commoditised. The software that runs it can't prove its claims.
Inverters, batteries and EV chargers have converged on similar round-trip efficiencies and similar price bands across vendors. The differentiator now sits above the hardware: the home energy management system (HEMS), the software layer that monitors, schedules and dispatches these assets against tariffs, weather and grid signals. It is where solar-battery-EV value is now captured. The uncomfortable finding across the independent record is that vendor claims about this layer far outrun what can be demonstrated, and that where dispatch logic is weak, high-efficiency hardware delivers little of its promise.
A third of products claim the top tier. One in eight can show it.
The most systematic independent survey available is the IEA 4E market scan, prepared by the Institute for Sustainable Futures at UTS for the IEA's 4E technology programme (February 2025). It reviewed 51 residential HEMS products against a four-tier capability ladder: monitoring only (visibility, no control); basic (schedules devices against known factors such as time-of-use tariffs); sophisticated (optimises devices using live solar/battery/EV data against pre-defined prices); and orchestrated (the top tier, curtailing, diverting or activating resources on dynamic external market, network or aggregator signals).
Under vendors' own "highest stated" functionality, 33% of the 51 products claimed the orchestrated tier and 51% claimed sophisticated. Under a cautious, evidence-based assessment of the same products, the distribution inverted: 34% were assessed monitoring-only, 30% basic, 24% sophisticated, and just 12% orchestrated. A product's stated tier and its demonstrable tier are frequently not the same tier.
The same scan quantifies why the ceiling is hard to reach: of 51 products, only 21% use open standards, 53% rely on bespoke brand-by-brand integrations, and around one in five are closed single-vendor ecosystems, with "no clear, single, fully comprehensive and universally accepted protocol" for device interoperability. Closed ecosystems can lock a home out of the very VPP and demand-response programs that make the orchestrated tier pay.
What the control layer is worthPassive information saves 4–7%. Integrated automation saves around 26%.
How much the software tier is worth, separate from the hardware, is the market's central economic fact, and it has a long evidentiary baseline. A PG&E-commissioned characterisation of home energy management technology (Karlin et al., 2015) compiled the savings literature: passive, information-only management, corrected for statistical bias across a meta-analysis of 42 studies, delivers a defensible 4–7%; utility feedback portals of the Opower type only 1.5–2.5%; appliance-level feedback 12–20%. Integrated systems, those that combine monitoring with automated appliance and climate control, were estimated to save around 26% (Williams and Matthews, 2007). That roughly four-to-sixfold gap between knowing and automating is the difference the control layer makes.
Sophistication is not the same as outcome"AI-driven" dispatch beats well-tuned rules by a margin too small to sell on.
The marketing frontier is "AI-driven" optimisation, but the controlled evidence keeps it modest. A 2024 BNAIC benchmarking study (Leiden / Shell Global Solutions) compared deep-reinforcement-learning battery-dispatch algorithms against a rule-based heuristic and a mixed-integer-linear-programming optimum with perfect foresight. The best learning method edged out the rule-based heuristic by a meaningful margin in the model, but every method saved only up to around nine euros a month combined over the test window, and a well-tuned rule-based controller stayed competitive with all but the strongest learning agent. Read plainly: control-logic quality matters far more than whether the label says "AI," and the sector is mid-transition from rule-based to machine-learned dispatch rather than past it.
What we can't yet citeThe most-anticipated independent HEMS lab evaluation is not public, so its numbers are not here.
California's CPUC Emerging Technology program lists an "Advanced HEMS" performance evaluation (project ET22SWE0055) that has circulated in industry discussion with specific per-system dispatch and efficiency figures, including a striking result that a prominently marketed system was out-dispatched by a less prominent one. We went looking for the final report to source those figures and could not find it publicly published: only the project's listing in a program priority map. Under our sourcing rule, an unpublished dataset does not ship, so none of those specific numbers appear here. When CalNEXT publishes the report, it will earn its own piece. What stands today is the peer-reviewed and program-published record, the capability audit, the savings literature, and the dispatch benchmarks above.
V1G is mainstream; V2G is notThe bidirectional-EV future is real, and mostly not yet buyable.
The US Department of Energy's Vehicles-to-Grid Integration Assessment (January 2025) draws the line cleanly. Managed one-way charging (V1G) is mainstream, with no OEM warranty prohibitions. True vehicle-to-grid (V2G), discharging the car back to home or grid, remains demonstration-stage: only two EV manufacturers currently permit V2X operation under warranty, for specific models, and one OEM's own disclosure to the DOE estimated V2G could cause "unnecessary battery charging and discharging power losses of >10%" plus "charger losses of ~10% (round-trip)."
| V1G: managed charging | V2G: vehicle-to-grid | |
|---|---|---|
| Power flow | One-way (grid → EV) | Two-way (EV ↔ grid / home) |
| Deployment | Mainstream | Demonstration / niche |
| OEM warranty | No prohibitions | Two OEMs only, specific models |
US DOE, Vehicles-to-Grid Integration Assessment (Jan 2025). Orchestrating V2G rather than managed charging needs both a bidirectional-capable charger, still rare and costly, and an OEM warranty position permitting the discharge. Neither is broadly satisfied today.
Why this bites hardest in AustraliaLocal tariff and grid rules make control quality directly worth money.
Dispatch software is only as valuable as the price and grid signals it is tuned to, and Australia's are unusually specific. Under WA's Distributed Energy Buyback Scheme, Synergy customers from 1 July 2024 earn 10c/kWh for exports in the 3pm–9pm peak window and 2c/kWh off-peak, a direct, quantified signal that a battery scheduler either exploits or wastes. From 1 May 2026, WA raises the standard-connection ceiling for inverter systems toward an aggregate 30 kVA and ties export participation to remote disconnect/reconnect capability, with non-participating systems accepting a hard 1.5 kW export limit. Nationally, CSIP-AUS (built on IEEE 2030.5) is becoming the interface for dynamic export limits and emergency backstop control, mandatory in South Australia since July 2023, in Victoria's emergency-backstop form since October 2024, with an ANU/ARENA national certification service standing up from late 2025. Note the division of labour often garbled in marketing: the inverter standard AS/NZS 4777.2 governs grid-support modes and export limitation at the device; the dynamic, network-driven limits ride on CSIP-AUS on top.
The commercial layer sharpens the point. Amber Electric dispatches customer batteries against real-time five-minute NEM wholesale prices rather than fixed tariffs (a vendor-run service, not an independent benchmark), a control problem no monitoring-only product can touch. And the aggregation prize is now real money: in July 2025 AGL acquired South Australia's Tesla-built virtual power plant, taking control of roughly 7,000 Powerwall batteries. A HEMS that can't answer a CSIP-AUS signal or arbitrage a five-minute price is, in these markets, leaving the orchestrated tier's value on the table.
Where the category is headingThe commoditised layer is growing; the control layer is growing faster.
Independent market estimates disagree on size but agree on shape, and should be read separately rather than blended: Grand View Research puts the global HEMS market at about US$4.9bn in 2025 rising to US$13.5bn by 2033; Mordor Intelligence estimates US$3.8bn in 2025 to US$9.5bn by 2031; Statista's narrower Australian smart-home energy-management segment is put at about US$143m in 2024 rising to US$217m by 2028. Different base years and definitions produce one consistent read: software is the faster-growing layer. Regulation is pushing the same way. Germany's §14a EnWG, in force since January 2024, requires residential loads above 4.2 kW (EV chargers, heat pumps, electric heaters) to allow utility throttling during congestion, and Eurelectric's 2024 "Grids for Speed" study argues grid-friendly flexibility can materially cut the ~€67bn a year the EU needs for grid upgrades through 2050.
The evidentiary record converges on one conclusion for anyone specifying a residential or commercial solar-battery-EV system: they are evaluating two products bundled together, hardware whose efficiency differences are now narrow and well-characterised, and control software whose quality varies far more widely than vendor positioning suggests, and which the independent record shows most vendors claim more of than they can demonstrate.
Sources12 references
- IEA 4E EDNA / Institute for Sustainable Futures, UTS: Residential HEMS and controllers, global market scan (Feb 2025)
- Karlin et al.: Characterization and Potential of Home Energy Management Technology, PG&E (2015; incl. the 42-study savings meta-analysis and Williams & Matthews 2007 integrated figure)
- Schonenberg et al.: Benchmarking Deep Reinforcement Learning for Battery Dispatch Optimisation, BNAIC (2024)
- US Department of Energy: Vehicles-to-Grid Integration Assessment Report (Jan 2025)
- pv magazine Australia: AGL acquires South Australia virtual power plant from Tesla (Jul 2025)
- Eurelectric: Grids for Speed (2024)
- Amber Electric: wholesale-price battery dispatch (vendor-run service, not an independent benchmark)
- Energy Policy WA / Synergy: Distributed Energy Buyback Scheme rates; WA connection and Emergency Solar Management rules
- CSIP-AUS (IEEE 2030.5): SA and Victorian mandates; ANU/ARENA national certification service
- Grand View Research, Mordor Intelligence, Statista: HEMS market sizing (separate bases; not blended)
- Bundesnetzagentur: §14a EnWG controllable-loads framework (in force Jan 2024)
- CPUC / CalNEXT: Advanced HEMS evaluation (ET22SWE0055): listed program project, final report unpublished at time of writing; its figures are deliberately not cited in this piece
Read our ownership, funding and editorial disclosures.
Stay across the benchmark.
New scorecards, score changes and market signals, published as they happen.