Chief Executive Officer / Singapore / Southeast Asia
Where the proposal changed before the commitment
- Owned solar capacity
- 148 MWp
- 315 roofs at June 2026; capacity is not energy delivered
- Company revenue
- US$31m
- 2025 electricity, installation and service income
- Colleagues
- 280
- Company-wide at June 2026
Decisions & consequences
01A portfolio approval that excluded eight of the proposed roofs
In 2021 a development proposal brought together twenty-two industrial roofs with a combined initial concept of 24 MWp. The portfolio looked efficient to build as one programme, and the commercial team wanted to secure equipment and installation capacity together. The site reviews, however, were at different stages. Some building owners had not settled future roof work, and some access arrangements depended on another party’s consent.
I asked the team to distinguish an attractive prospect from a site ready for capital approval. Engineering, operations and legal advisers reviewed the conditions with the host organisations. We spent US$390,000 on assessments and legal preparation before recommending the final commitment. The board approved fourteen ready roofs, representing 17 MWp and US$12.6 million of planned capital expenditure. Five sites remained in preparation and three were declined. The five were not quietly retained in the committed-capacity number.
The development director argued that separating the programme would weaken purchasing efficiency and disappoint customers expecting a common timetable. I accepted the loss of some package efficiency because the larger commitment depended on conditions we did not control. We revised the installation sequence and explained the distinction directly to the affected hosts. The ready group could proceed through its required site-specific approvals and delivery arrangements without using another roof’s uncertainty as a reason to defer the whole portfolio.
The outcome was an investment decision, not a claim that every installation subsequently performed exactly as forecast. We reported the fourteen committed sites, the five awaiting further preparation and the three declined opportunities separately. I would involve the operating team earlier in future development so that access and roof-renewal questions influence the initial commercial discussion. The enterprise lesson was to preserve the difference between opportunity, readiness and commitment, even when a single larger number would have made the growth story easier to tell.
A missing signal was not automatically lost generation
In 2022 the operating review showed a recurring problem with remote alerts: a site could appear unavailable because the monitoring connection had failed, while the equipment continued to operate. Conversely, a restored signal could make a dashboard look better without confirming that an underlying equipment issue had been resolved. The existing report did not consistently preserve that distinction.
I approved US$240,000 for a six-month programme covering ninety-two sites. Operations, monitoring specialists and field teams agreed a common diagnosis record. The work did not treat every alert as an instruction for an immediate site visit. It first established what could be confirmed remotely, which questions required field inspection and who owned the unresolved cases. Required access and safety procedures remained with the qualified operating team; the executive review did not replace them.
Across the review period, 121 alerts were recorded. Seventy-six were classified as communication interruptions, thirty-one as equipment issues and fourteen remained unclassified at the review date. Of the thirty-one confirmed equipment cases, twenty-two had been corrected and nine were awaiting parts. Those groups were reported separately rather than collapsed into one closure rate. A commercial colleague proposed translating the corrected cases into recovered generation for an investor presentation. We did not have a sufficiently reliable baseline to support that calculation, so I declined the claim.
The programme improved our ability to direct operating work and made the outstanding cases easier to examine. It did not establish a quantified generation gain or eliminate future faults. The spending also included investigation that produced no immediate repair, which was appropriate where the initial signal was ambiguous. If repeating the work, I would define the diagnosis fields when a site first enters the portfolio. Retrofitting a consistent record after alerts have accumulated is slower, and it invites people to interpret the earlier data more confidently than it deserves.
A smaller factory package still had a forecast to answer for
In 2024 a factory customer discussed a nineteen-building package with an initial 24 MWp solar concept and US$16.8 million proposed capital commitment. Early commercial work relied heavily on annual electricity consumption. Subsequent operating discussions showed that the customer expected changes in shifts and production, with less daytime demand available at several buildings than the original model assumed.
The development team wanted to preserve the full package, arguing that demand might recover and that installing everything together would be more efficient. I asked for a proposal that could stand on the revised operating assumptions rather than on an uncommitted recovery. Engineering and commercial colleagues reduced the initial package to 16 MWp, with planned capital expenditure of US$11.2 million. The remaining US$5.6 million was not committed. Any later expansion would require a fresh review of demand, permissions and the relevant site arrangements.
We explained the change to the customer as a different investment sequence, not a judgement that its business would never grow. The smaller installation preserved the possibility of a later phase without treating that possibility as current revenue. The first twelve operating months delivered 18.4 GWh against a 19.6 GWh forecast. Lower daytime offtake was one contributor, alongside site-specific operating and weather differences. We retained the original forecast in the review rather than rebasing it after the result was known.
The decision reduced the capital placed at risk under the original assumptions, but the first-year shortfall showed that the revised case still contained uncertainty. I would now bring the customer’s operating planners into the initial model review, alongside the energy and procurement contacts. An annual total is convenient for a first conversation; it is insufficient for deciding when equipment will produce, when energy will be used and which obligations the company should accept for the years that follow.