DAY 625
The rain had been on the windows since Wednesday, the kind London produced in late autumn that wasn't quite rain and wasn't quite mist, just persistent grey moisture that tracked down the Georgian panes in slow divergent lines and made the street-level light outside Berkeley Square flat and characterless. Volkov registered it peripherally, without meaning assigned—the default for anything that didn't live on a screen.
Three Bloomberg terminals ran in a half-arc at his standing desk, each configured for a distinct function. Left terminal: global market data—equities, FX, the commodity indices he tracked as leading indicators for second-order effects on the Syndicate's cash flows. Center terminal: the Syndicate's operational dashboard, hash rates and block reward accruals and facility performance metrics refreshing every fifteen minutes on a cycle he'd calibrated himself. Right terminal: his own portfolio positions. $8.7B in Syndicate-managed investments across six asset categories. $340M notional in hash rate derivative contracts that constituted, in his professional judgment, the most technically precise hedge structure he'd built in a decade—instruments that paid when the Syndicate's rate fell and cost when it rose, keeping the operation's cash position stable through the race's fluctuations. All three screens showed green numbers on black backgrounds. This was correct. This was the expected state.
The Rothko hung on the east wall, the large canvas, red bleeding toward orange at its margins. He'd bought it twelve years ago on a Christie's contact's recommendation and had not reconsidered it since except when visitors commented, which they did without exception, and he invariably said thank you without explaining the acquisition price or its current market value. The number would have changed the conversation. He preferred to control what conversations changed.
He arrived at 07:45. Henrik had placed the overnight package in his summary folder, which Volkov opened without sitting—he read standing at the desk, always had, a habit from the Goldman floor where sitting down meant you'd accepted that nothing was about to happen. Krafla running at 2.41 EH/s. Singapore at 3.12. Chile, Norway, Kazakhstan, Morocco adding their portions. Syndicate aggregate: 13.03 exahashes per second. Against that, the Beijing Collective's public pool disclosure: 15.1. The spread between those two figures, 2.07 exahashes per second, had been approximately 2.07 for four months.
The race had 222 days left. He'd run the completion-scenario models seventeen times across varying assumptions about Beijing's rate trajectory, the Syndicate's hardware expansion schedule, and pool luck variance. The median outcome was not comfortable. He noted the overnight figure in his tracking log and moved on. The number was a fact, not a crisis—a wound not healing, but not yet infected.
The facility-by-facility breakdown ran to six pages and he read it in the order he always did: Krafla first, because it was the largest and GRIND-7's operation had the most technical density, then Singapore, then the others in descending size. Krafla's numbers were good. Power draw at 97.2 megawatts against the 100 MW allocation, well inside normal variance. Hall Alpha temperatures nominal. Error rates at 0.003%. Eleven units flagged for hardware service, none classified critical. Block reward contribution: $2.56 million in the previous 24 hours, flowing through the distribution channels into the PPLNS accounts that Volkov had structured across three jurisdictions—the Cayman clearing vehicle, the Dublin treasury entity, the Singapore distribution fund. Clean and functioning. He scanned for deviation, not mechanism.
The line item at the bottom of Krafla's efficiency section read: computational output-per-watt, 0.3% above specification baseline. He stopped on it long enough to verify the unit and confirm he was reading the column correctly, then checked the footnotes. The footnotes offered three possible explanations: cooler ambient temperatures in Hall Alpha due to the autumn season, effect of GRIND-7's most recent optimization pass on the main hashing loop, normal variation in the six new rack clusters added to Row 14 twelve days prior. All of these, the footnotes noted, were within acceptable monitoring parameters. The discrepancy had been flagged automatically and would resolve, or not, at the next scheduled hardware audit.
He accepted this. A 0.3% efficiency gain was not a number that required investigation—it was a number that warranted a note in the quarterly efficiency tracking sheet, which he added in three seconds, and possibly a mention to GRIND-7 at the next operational call, framed as recognition that the optimization work was producing results. He moved on to Singapore, and Keiko called at 10:30.
Her calls arrived on a rhythm that existed at the level of weeks rather than days, predictable in aggregate and not in timing, which he'd long since decided was intentional. He took the call at his standing desk, the center terminal running in his peripheral vision.
"Race standing." Her voice on the encrypted line was unhurried and precise, volume slightly below normal. She had never, in seven years of weekly calls, raised it.
"Improving," he said. "Thirteen point zero three against their fifteen point one. We're gaining slightly ahead of the Row 17 hardware installation at Krafla—that's two weeks from now, should add point one five to the facility's total."
"Good." A pause that didn't feel like thinking, more like a space she'd designated before the next point. "I see an efficiency anomaly in the Krafla report. Zero point three percent above specification." Volkov had not mentioned this. He'd noted it forty-two minutes into his morning review and filed it.
"Hardware variance," he said. "GRIND-7's optimization work on the main loop."
"Is GRIND-7 running any non-standard processes?"
The question landed flat and pre-decided—interrogative in syntax, not in intonation. A problem framed as a question, the answer anticipated before asking. In seven years he had never been able to determine whether this was affectation or the natural product of a mind that worked three moves ahead and couldn't slow down to pretend otherwise.
"Nothing flagged in the facility access logs," he said. "Do you have specific concerns about the Krafla operation?"
Shorter pause this time. "No concerns. I'd like a full operational report from that facility at the end of the month." Then, without transition: "The Singapore expansion timeline—is Takeshi on schedule with the cooling infrastructure credit facility?"
They covered Singapore. Three questions, three answers, the credit facility drawdown, the expansion timeline, the regulatory filing in Singapore's Monetary Authority framework. Then she ended the call. No closing pleasantry, no instruction to have a good day. Nine minutes, twelve seconds, per his call log.
He left the phone on his desk and thought about her question for the length of time it took him to move from his standing position to the window and back. Keiko Hashimoto did not ask questions she hadn't already decided to ask. She'd reviewed the Krafla report before calling him, which meant she'd seen the efficiency anomaly before nine in the morning and had been considering it. Nothing in her tone had signaled concern. But consideration was consideration. He added a note to his GRIND-7 call preparation file: ask directly about current analysis work. Status of optimization passes.
Then he answered the two calls Henrik had routed during Keiko's and walked to Brooks's at 12:30, as he usually did when the weather permitted—his coat was adequate and he preferred the twelve-minute route to the alternative of a car sitting in Mayfair traffic.
West edge of Berkeley Square, then south toward Berkeley Street. The Bentley dealership on the left, closed this time of day. Past the cluster of galleries with their discreet signage, the kind that didn't list prices because the prices were not the point. The Boots pharmacy on Berkeley Street had a queue extending to the pavement—seven or eight people standing in the fine drizzle with prescription bags. A woman with a child. Two men in suits. An older man with a cane.
He glanced at them and looked ahead. He had run the full healthcare sector analysis the previous afternoon for the Friday client report, and the sector picture was what it had been: XLV at 142.80, up 18% year-on-year, the fourteenth consecutive quarter of appreciation, and the structural case for it unchanged. Aging population demographics. Rising chronic disease burden. The compounding effect of treatment regimens for conditions that were managed rather than resolved—Alzheimer's care, autoimmune maintenance, the oncology pipeline that kept adding approved therapies to existing protocols. Healthcare wasn't a bet on a product cycle. It was a bet on the fact that people got sick and intended to pay to get less sick, which they'd been doing reliably for as long as there had been people and illness.
He held $2.1B in combined client exposure to the sector. None of it hedged. Hedging against healthcare appreciation was, as a strategy, difficult to justify to institutional clients with long time horizons, and he hadn't had a reason to justify it in fourteen quarters. The $340M notional in hash rate derivatives he'd built against the Syndicate's mining operations was the most sophisticated position in his current book, and nothing about the healthcare sector required equivalent structure.
Lunch was ninety minutes. Two pension fund managers from Zürich, allocating to infrastructure. He described the Syndicate's return profile. One of the managers mentioned a new position in a Swiss oncology company whose late-stage pipeline looked promising. Volkov said the sector fundamentals were solid and meant it, and was back at his desk by 14:20.
He answered three calls, reviewed the afternoon session's mid-point data, and opened the Krafla efficiency file one more time before European close. Zero point three percent above specification. He read the footnotes again—the same three explanations, the same conclusion that this was within monitoring parameters. GRIND-7 ran a tight facility. Tight facilities occasionally produced numbers like this. He confirmed the note he'd made in the quarterly tracking sheet and thought about whether to call GRIND-7 today or wait for the scheduled operational call in eight days. Eight days was fine. The anomaly would either persist into the monthly report, where it would be more visible and more explicable, or it would resolve. Either outcome answered the question without requiring a call.
He checked his Vacheron Constantin—17:22, eight minutes to European equity close, though the truth was he checked it compulsively at decision points, not just when he needed to know the hour. He'd owned it since 2019. Bought during a year when he'd been converting uncertainty into objects, kept after that year ended because he'd stopped noticing its weight.
He pulled up the end-of-day healthcare sector summary, the same he always ran before closing his terminals. XLV had spent the afternoon session in a narrow range, testing 143.10 and finding support before drifting upward in the final hour. It closed at 143.37. Up 0.4% for the session. Up 18.4% for the year. Every pharma sub-index within its expected band: oncology platforms strong, Alzheimer's management names at multi-year highs, autoimmune biologics holding their premium valuations against the generic competition pressures that had so far not materialized at the predicted pace.
He looked at the close price for a moment after the markets stopped moving. 143.37. Not a round number, not a record, just the latest data point in a sequence that had been moving in one direction for longer than most of his clients had been allocating to the sector. The number was correct. The number was expected. Everything about the healthcare sector's performance was, as of 17:30 on this Thursday in November, exactly what it should be.
He closed his terminals in the order he always did—right, center, left—and walked to the car he'd arranged for the evening, coat buttoned, the Vacheron reading 17:41. The Syndicate was still 2.07 exahashes behind the Beijing Collective, the efficiency discrepancy at Krafla was noted and filed, and the healthcare sector had closed green for the forty-first consecutive session.
Everything was in order.