How Firm Is Millennium’s Book Value: Tipa and TripleW Under the Hood

Starting Point

The main article argued that Millennium’s real problem is not only the size of the discount, but also the long distance between paper value and value that can actually reach unitholders. This continuation isolates the question that sits underneath that argument: how much of the carrying value of Tipa and TripleW should be treated as firm value, and how much of it still rests on models and assumptions.

The reason to focus on these two names is straightforward. At the end of 2025 they were carried together at NIS 35.948 million, or about 63.3% of the private fair-value portfolio. On the partnership’s own asset view, Tipa and TripleW were also defined as material holdings and together accounted for about 52% of total assets. This is not a technical debate about valuation methodology. It is a debate about the core of Millennium’s balance sheet.

The first thing to understand is that this value is not being set by a fresh, clean price event. The valuation appendix itself describes the work as an indicative economic review based on data and representations supplied by Millennium and the portfolio companies, without an independent verification of the reliability, timeliness or completeness of that information. That matters because the two heaviest holdings in the portfolio are only as firm as the evidence base supporting them.

That leads to the key distinction in this follow-up. The question is not whether Tipa and TripleW are worth zero. The evidence set does not support that read. The real question is how much of this value has already passed through an external anchor that hardens it, and how much still sits in a zone where forecasts, discount rates, time to liquidity and FX translation do most of the work.

Tipa: A Detailed Model, But A Weak External Anchor

Tipa was carried at NIS 21.896 million at the end of 2025, or USD 6.864 million. It is the single largest mark in the portfolio, and it is also the one with the most explicit assumption set. In the valuation-disclosure section, Millennium says the mark was based on DCF and OPM, with a 21% discount rate, a 3.5% risk-free rate, 65% volatility and 2.5 years to a liquidity event.

That is a value built with a lot of internal mechanics, not with a fresh market price. There is no new financing round here that settles the debate. There is a model that has to assume what future cash flows look like, how long liquidity will take, and how much risk should be loaded onto that path.

The more important nuance is that the shekel move in 2025 can easily mislead. At the end of 2024 the holding stood at NIS 24.708 million. By the end of 2025 it was down to NIS 21.896 million. On a superficial read that looks like a sharp deterioration. But Note 12 breaks the move down differently: there was an FX loss of about NIS 3.244 million, offset by a fair-value gain of NIS 432 thousand. In other words, the core model did not mark Tipa down in 2025 because the business story collapsed. Quite the opposite. In dollar terms the value actually edged up from USD 6.775 million to USD 6.864 million. What fell was mainly the translation into shekels.

That is exactly where the read has to stay disciplined. FX is not proof of value, but it is not a refutation of value either. It mainly reminds us that the number in the balance sheet is a mix of valuation and currency translation. Anyone reading the shekel decline as operating deterioration is missing the point. Anyone reading the dollar increase as proof that the value has hardened is missing the point as well.

The real weakness in Tipa sits somewhere else. The valuation review says that, as of the end of November 2025, Tipa had about USD 18.9 million of cash and that management believed this was enough until the second quarter of 2027. The same review says the valuation relied on a profit-and-loss forecast supplied for a 409A employee-option exercise. But Millennium’s annual report presents the disclosure in a very different tone: Tipa says it had no going-concern note in its 2024 audited statements, that it was still not producing positive cash flow, and that as of the report date it refused to publicly disclose its cash-burn rate, its cumulative development spending and its cash balance, on the grounds that this was commercially sensitive information.

That is not a formal accounting contradiction, but it is a critical gap in how the mark should be read. To support Tipa’s carrying value, investors have to rely on data the company was willing to provide to the valuer, but not willing to fully disclose to the public. That makes the mark less anchored in rich public disclosure and more dependent on a private information channel to the valuation process.

There is another detail that weakens the external anchor. In 2024 Millennium bought 492,475 common shares and 21,320 Series A preferred shares of Tipa from an unrelated third party for a total of about USD 525 thousand. In theory that could have served as a real-world price reference. In practice, the valuation review explicitly says this transaction was not viewed as representative of fair value because it took place under seller pressure. So even the one actual transaction that could have functioned as an external anchor is disqualified inside the evidence set itself.

The result is that Tipa looks stronger in the balance-sheet number than in the type of evidence supporting it. There is a real company here, with strategic activity, new products, acquisitions in Europe and a credible commercial thesis. But Millennium’s year-end mark does not rest on an exit, a meaningful sale or a fresh financing event that hardens the anchor. It rests on a detailed model and on assumptions that still have to prove themselves.

TripleW: Not A Repricing, A Roll-Forward

If Tipa is a mark with many visible assumptions, TripleW is a mark with less disclosed model detail and more reliance on the prior value. At the end of 2025 Millennium carried TripleW at NIS 14.052 million, or USD 4.405 million. In the valuation-disclosure section, the method here is not a DCF with a stated assumption stack. It is described as "economic review plus the latest investment", and the assumptions line is simply left blank.

That is the heart of the TripleW story. The value did not break in 2025, but it was also barely re-tested by a fresh outside price. The valuation appendix says that because the convertible financing was close in time to the valuation date, the value of the holding was left unchanged versus June 30, 2025 at USD 4.255 million, with only an additional USD 150 thousand added for the investment made in the second half of the year. In other words, the year-end mark mostly does not reflect a fresh business repricing or a new financing round that sets an updated price. It reflects a carry-forward of the mid-year value plus fresh money.

Here too, the shekel move tells only part of the story. At the end of 2024 TripleW was carried at NIS 15.518 million. During 2025 Millennium added another NIS 483 thousand of investment, yet the year-end carrying value still fell to NIS 14.052 million. Note 12 explains why: an FX loss of about NIS 1.949 million. So, as with Tipa, most of the shekel decline came from currency. But unlike Tipa, there is no parallel disclosure here of a small accounting upside from a clearly specified model. What you have is mainly a rolled-forward value.

What sits underneath that roll-forward? In Millennium’s annual report, TripleW is presented as a company that has started to generate revenue, but still on a very small scale relative to the broader thesis. Its subsidiary ended 2025 with about EUR 310 thousand of revenue, EUR 3.8 million of cash and a monthly cash burn of EUR 350 thousand. In the same disclosure the company also says it expects to need additional investment during the second half of 2026.

The valuation appendix, meanwhile, presents slightly different operating inputs: about EUR 3.5 million of cash, a burn rate of about EUR 410 thousand per month, and about USD 1.1 million of convertible financing raised in September and November 2025, in which Millennium itself participated with USD 150 thousand. The gap is not huge, but it is large enough to remind us how sensitive this mark is to the quality of the operating data sitting underneath it.

That does not mean TripleW’s mark is detached from reality. Far from it. The evidence set includes real building blocks for why Millennium did not write it down: a concession agreement with the Port of Antwerp for a site of roughly 38 thousand square meters, an agreement with an engineering contractor for additional Basic Engineering work, an LOI with Hawkins for about 6,000 tons of lactic acid and derivatives, and a broad grant stack that includes NIS 68 million approved in 2024 of which NIS 34 million had been received, EUR 13.7 million from CBE JU of which EUR 4.8 million had been received, and another EUR 9.45 million grant from CINEA that had not yet been received by year-end 2025.

But grants, LOIs and engineering progress are not a price. They support the thesis, they extend the runway and over time they may turn into stronger external confirmation. At year-end 2025 they still do not harden the mark the way a meaningful new financing round or a monetization event would.

What This Means For Millennium’s Book Value

Putting the two holdings together produces a sharper conclusion than looking at either one alone. Their combined carrying value fell from NIS 40.226 million at the end of 2024 to NIS 35.948 million at the end of 2025. But once the move is broken down, most of it does not come from an operating collapse in the holdings themselves. It comes mainly from FX.

That chart matters because it blocks two opposite reading errors. The first is to say that the shekel decline proves the balance sheet is falling apart. It does not. Most of the move came from FX. The second is to conclude from that that the value must be firm. That is also wrong. If most of the holding still depends on models, an indicative review, management-supplied data and a roll-forward of prior value, then the fact that the number did not collapse is not the same thing as having a solid market anchor.

Put differently, this looks more like value that can be explained than value that has already been tested. In Tipa, there is a detailed model but it carries heavy assumptions around time, risk and liquidity. In TripleW, there is real industrial and financing progress, but the mark itself relies heavily on carrying a prior value forward with added money. In both cases the evidence supports the view that the companies are still alive and moving forward. It does not yet support the view that the balance-sheet number has turned into a market-cleared value investors can fully hold on to.

Bottom Line

Tipa and TripleW are not incidental holdings inside Millennium. They are the core. That means the lesson here is structural, not anecdotal: Millennium’s book value at the end of 2025 does not rest on an obvious fiction, but it also does not rest on an external anchor firm enough for the market to fully trust it.

In Tipa, firmness depends on whether the 409A-based forecast, the cash runway and the 2.5-year liquidity assumption can survive the tests of 2026 and 2027. In TripleW, firmness depends on whether grants, LOIs and plant progress turn into a new price event, rather than just a longer story on the way to one. Until that happens, these NIS 35.9 million should be read for what they are: a reasoned value, but not yet a hardened one.