TFX

We thought what is going on in Iran (this time) is different. Having at least 10% cash seems appropriate if what we are hearing from the Energy guys is true with respect to Oil & Gas disruptions and price impact. If Europe is hit hard, there will be opportunities to get into stocks (US too) at a double digit discount. It’s easier to judge in hindsight, but it is about being positioned right most of the time prior to price changes. That said, if $DOCN hadn’t become ~60% of our portfolio and we thought near-term inference economics were better, would we have 20% in cash now? Maybe not. So each investor should think for themselves given what they own and how they want to be positioned relative to what’s out there.

Our $DOCN model below. Thread includes: - Thesis Summary - Revenue Schedule - Cost of Sales and Operating Expenses Schedule - Income Statement - CashFlow Statement - Balance Sheet - PPE, Finance and Operating Leases Schedules We have not sold after the $800M share issue. Position currently 29% of our portfolio, we have been explicit about the risks going forward. All our decisions are based on how we see the business performing in the future, and how certain we are of the projections. We would caution against comparing $DOCN to $PLTR. $PLTR needed little to no CapEx to grow, if $DOCN profits show chronic pressure the stock could take a dive, quickly.

Today we reduced our $DOCN position by 44%, achieving a 148% return in less than 12 months on the block sold at $66.5 per share. Before we explain why, it is worth noting that our remaining position makes up 1/4th of our portfolio. We trimmed because: - Inference per MW economics mean very poor incremental gross margins during capacity ramps and on a fully MWs available basis (36.6% vs 60% average over last 5 years). - On this basis, we don't think the company is worth much more than today by 2030 - As a result of the sale, our cash is now 20% of our portfolio which makes us feel good given the uncertainty out there (we think the ongoing situation in the middle-east carries a different risk profile from prior geopolitical events given the incentives, and effective residual capabilities of Iran showcased in their ability to keep the Strait closed). We have not exited completely because: - If revenues per MW increase like they did for traditional cloud over the last 15 years, we estimate 5-6% Gross Margin improvement for each 10% improvement in the revenue per MW side - On this basis, the company could produce operating income in excess of $600M/year by 2030 which would put it on a path to $20B+ Market Cap. Given the uncertainty, we feel that this action protects our downside (story not longer as straightforward as it was prior to Q4) but allows us to participate in any outsized surprise that could occur given a future feasible improvement in inference economics. More detail: The company will add 31MW of capacity in 2026. We derive the 36.6% incremental GMs as follows: Investor Supplement from March 3, 2026: - ~$13M+/MW of Revenue, monetized for 6 years - $20-25M/MW of CapEx paid over 4 years - Rent expense as ~40% of total COGS (ex. Other) Our assumptions: - $13.5M/MW of Revenue, monetized for 6 years - $25M/MW of CapEx paid over 4 years - $19.3M/MW of co-location paid over 6 years - 12.3% Other COGS % of above 2 COGS components A note on potential revenues We think $DOCN will capture at least 0.1% of the total estimated inference demand of around 110GW by 2030, per McKinsey estimates. That will drive $2.7B of annual revenues by 2030, but $3.1B ARR on a fully ramped basis (MW capacity /= MW available). That's based on an assumption of $13.5M of Revenue per MW for inference MWs, and $23.5M of Revenue per MW for non-inference MWs.