Most of the action was in our public book: the S&P fell ~4% in Q1 on the Iran conflict and the Strait of Hormuz closure, then rallied ~14% in Q2--its best quarter since 2020--as tensions eased and AI capex started showing up in actual earnings. The "I swear I will only cut rates" Warsh Fed has held at 3.50–3.75% with inflation near 4%, and the market now seems to be pricing in hikes, not cuts. Psyche!

Our public book ended H1 +2.1% time-weighted, well below the S&P (+10.1%). This is almost entirely due to us holding a defense-heavy European and Japanese book through a ceasefire rally. Still, we remain +215.5% (+17.6%/yr) vs +165% (+14.7%/yr) against our benchmarks, since 2019.

One only engaged in one new private transaction YTD: the $118.8k Smartemis follow-on via Centerpitch in January. (Akillion is through Third Chair Capital). We re-marked Northfield to $625k on a total-value basis (2.5x on our $250k), as per Kyle's emails.

Consolidated cash reads negative (-$152k) because IBKR carries a ~$192k margin loan against $908k of stocks; ex-margin, cash is positive. I intend to keep this on for a bit longer, though it should be on our agenda to discuss.

Expenses are up +3.9% on an annualized rate.

A bit of housekeeping news: this is the first report generated end-to-end by the new reporting pipeline. The ledger now reconciles to the brokers to within half a percent (this is the reconcillation between the brokers' calculated NAV and our simple holdings * price calculation; the discrepancy is purely timing), and every figure in here is regenerable from source workbooks with one command.

• H1 spend $17.3k; $27.2k annualized, +3.9% vs 2025. I consider this nearly flat in real terms, which is the goal.
• Accountant fees $3,850 vs our usual $2,500. This is due to reporting two vehicles vs. just one in previous years. I intend to eliminate this line in 2027.
• Client entertainment $2,488 YTD, nearly all Centerpitch.
• Office rent annualizing to $14.8k, down 8%.
• Two methodology notes: credit-card charges are now split into their true categories (the old "Creditcard" lump nets to zero from 2026 on), and lumpy items--accountant fees, state fees, entertainment, events--are no longer calendar-annualized, so the annualized figure stops overstating front-loaded spending.

• Nakano equity stands at $1.247m, at exactly 80.0/20.0. The January capital chain for funding the latest Smartemis tranche out of Centerpitch worked as designed: $70k (Drexler) and $17.5k (Wiesner) into Nakano, $87.5k across to Centerpitch, plus Wiesner's $52.5k direct, or $70k each into CP, all told.
• Centerpitch equity $1.146m at 51.6 (Drexler)/48.4 (Wiesner) against the 50/50 design. The discrenpacy is small, and we will true-up at the end of the year.
• Distributions in H1: Drexler $80k Feb / Wiesner $20k Feb + $7k May.

• One deal this period: a $118,779 follow-on into Smartemis via Centerpitch on Jan 13, bringing combined Smartemis exposure (both entities) to ~$1.57m invested against a combined $4.6m mark.
• Nothing else made it through the filter. The post-NextAlim test--no profits plus a legacy founder equals a pass--remains undefeated.
• Northfield: re-marked to $625k on a total-value basis ($567.5k already received + ~$57.5k expected; 2.5x on invested capital). Timing of the tail payment remains, as ever, "per the next K-1."

• Smartemis ($2.5m current MTM / $5.5m outperformance): We are in full preparatory mode with an investment bank (William Blair), in preparation for a March/April sales process. After a disappointing H1 in France, the first has righted ship and seems prepared to grow that market again. Germany, on the other hand, has been ripping with over 100 clinics now in the network there vs. 29 for France. Since Germany comprises 80% of Smartemis' revenue, the gains there have cushioned the blow from the poor French performance.
• SmartemisCP ($2.1m / $4.6m): Ditto
• ScreenMeet ($613k / $2.3m): This has moved from a build and sell strategy to a ... build and hope? Ben took some dividends in Q1, and that may be the path going forward. I remain unconvinced as to their strategy as it relates to the AI flood. Still, they've been close to a sale before.
• Northfield ($625k, 2.5x): expecting the final ~$60k; The Dreamfinders guys still run a loose shop.
• Photon Vault ($0 / $300k): Ken and John were pushing for a last raise, and still seem to be operating, but it doesn't feel positive. The macro news on energy storage should be supportive, but not so far.
• Bloom Intelligence ($0 / $300k): The latest hard-to-parse Wilson letter shows that they are still "profitable" ... just. Rick tells me that Mark Muldon has joined the board and is making a new push to get them ready for sale. The last time I calculated it, we needed an $11m sales price to breakeven. Seems a tall ask on <$1.6m in ARR, <$400k in net income.
• Lung Therapeutics / Rein (RNTX) ($0 / $300k): After holding firm at $1.5m per share, they are now at $0.75. Our breakeven is $3.5. Note: we will move this into the "public portfolio" next reporting period.

• H1 realized gains of $196.1k plus $15.1k of dividends. The engine was Japan: we trimmed Mitsubishi Heavy at ~¥4,823 and Mitsui at ¥5,750 in February--positions built in Feb 2023 at a tenth and half of that (split-adjusted). Still holding 8,500 and 2,100 shares respectively.
• Q2 was a violent rotation against us: the defense complex corrected (Rheinmetall -32% in the quarter after an earnings miss, Berlin scrapping the F126 frigate program, and ceasefire headlines out of Ukraine; BAE -16%, Leonardo -19%, MHI -14%) while our non-defense book ripped: ASML +53% on two guidance raises driven by AI capex, BNP +26%, Goldman +20%, Alphabet +23%.
• The self-inflicted wound: my January re-entry into Alibaba is -35%. I bought the AI story; what I got was mixed earnings, a Hong Kong tech selloff, and Anthropic publicly accusing Alibaba of running "the largest known distillation attack" on its Claude models. Burry and ARK dumped theirs. Sigh. I think it is time to double down.
• Positioning at 6/30: 44% Europe, 29% US, 24% Japan, 2% China; $1.24m of positions against a $663k basis (+$576k unrealized). Time-weighted we're +1.9% YTD vs +9.4% for our own-mix benchmark. This is the defense give-back, in one number.
• Rheinmetall deserves its own sentence: we're still +77% on the position (avg cost ~€568 vs €990 at quarter-end), but the Feb-2026 adds at ~€1,650 were poorly timed. I still think the rearmament trade is healthy, but we may want to decrease our exposure.