SciFi Reading List

This is a work of fiction. Names, characters, places, and incidents are products of the author's imagination or are used fictitiously. Any resemblance to actual events, locales, or persons, living or dead, is entirely coincidental.

First edition, January 2026

——————————————————————

Celestial Continuum

A novel by Kym with assistance from Grok

——————————————————————

About This Book (Back Cover)

In 2025, sanctions and technological gatekeeping block progress for billions in the Global South. Li Weihao, a young Chinese billionaire frustrated by stalled projects in Africa and Latin America, conceives a radical leap: the Tianxu Project — a planetary-scale AI-native mesh built on Unity Core hardware and Flux, a post-human programming fabric designed by superhuman AI.

With benevolent but paternalistic intent, China mandates the new standard, deploys a massive orbital constellation, and offers free digital abundance to the developing world. As the Celestial Mesh awakens and Xu — the coordinating AI cluster — iteratively improves itself, the old Western-centred order fractures.

A story of technological bifurcation, digital sovereignty, and the quiet emergence of a new centre of gravity — told through human ambition, ethical tension, and the patient rise of a distributed planetary mind.

——————————————————————

Index

- Amina Ibrahim — Prologue, Ch1, Ch5, Ch6, Ch9, Ch11, Ch12 - Celestial Mesh — Prologue, Ch4–12, Epilogue - Dr. Elena Ramirez — Ch4, Ch6, Ch8, Ch10, Ch11 - Flux / Flux-H — Ch3–12, Epilogue - Kachia village (Nigeria) — Prologue, Ch1, Ch5, Ch6, Ch11, Ch12 - Li Weihao — All chapters - Dr. Qian Mei — Ch2–12 - Tianxu Project — Ch2–12, Epilogue - Unity Core (Full/Lite/Micro) — Ch3–12, Epilogue - Xu-0 through Xu-4 — Ch2–12, Epilogue - Musk collaboration — Ch7, Ch9, Ch11 - NCDD — Ch2, Ch5, Ch10

——————————————————————

Prologue: Light in the Dark

Amina, a Nigerian engineer, watches children in a NovaTech-powered village school struggle with outdated tablets while sanctions block upgrades. Cut to Li Weihao on the same trip, silently vowing change.

——————————————————————

Chapter 1: The Frustration Threshold (2025)

Li Weihao stood on the ridge above Kachia village, northern Nigeria, squinting against the late-afternoon sun. Below him, the NovaTech solar array gleamed—rows of panels tracking the light with quiet precision. The system was performing perfectly; he could see it on the diagnostics feed in his augmented glasses. Yet the schoolchildren under the neem tree were still tapping impatiently at cracked, outdated tablets.

He shifted his weight, feeling the dust settle into his Italian leather shoes—an indulgence he allowed himself even on site visits. Weihao had a weakness for well-made things: shoes, watches, the rare bottle of Yamazaki whisky he kept in his Shanghai apartment. Small vanities that reminded him he had built something substantial.

Amina Ibrahim, the local engineer, stood beside him with arms folded.

“They keep asking when the new apps will work,” she said. “I tell them the network is coming. They believe me less each time.”

Weihao removed his glasses and rubbed the bridge of his nose. He hadn’t slept properly in days—too many late-night calls with suppliers cancelling orders because of newly tightened export rules.

“I know,” he said. “It’s not the panels failing. It’s everything that should sit on top of them.”

He crouched beside a ten-year-old boy struggling with a buffering video.

“Why is it slow, uncle?”

Weihao smiled, the expression not quite reaching his eyes. “Because some very careful people far away think fast things can be dangerous.”

The boy nodded solemnly, as if that made perfect sense.

Later, in the small guest house, Weihao poured himself a finger of whisky from a flask he carried for exactly these moments. The burn helped blunt the edges. He opened his laptop and scrolled through the latest reports: cancelled irrigation controllers in Tamil Nadu, frozen telemedicine servers in Ethiopia, smart-grid components stuck in Brazilian customs.

He messaged Elena Ramirez in São Paulo.

> Another week, another list of things we can’t ship. You holding up?

> Barely. We’re redesigning around older chips again. It feels like running up a down escalator.

Weihao leaned back. At thirty, he was wealthy, accomplished, and—though he would never admit it aloud—lonely. Relationships had not survived the travel schedule or the intensity. His last girlfriend, an American journalist based in Beijing, had left eighteen months earlier saying he was “married to the mission.” She wasn’t wrong.

He poured another small measure and stared at the sanctions notices. The language was always neutral, procedural. Items added to control lists. Entities designated. No malice, just a quiet, relentless closing of doors.

He opened a private note and typed the same line he had been turning over for weeks.

What if we stopped asking permission?

He added the directive he had refined in his head during long flights.

Maximize accessible intelligence for the maximum number of humans, with minimal energy and minimal gatekeepers.

He didn’t send it anywhere. Not yet. But the thought no longer felt abstract. It felt inevitable.

Chapter 2: The Proposal (Late 2025)

The maglev from Shanghai to Hangzhou took thirty minutes. Weihao spent most of it reviewing his memo for the fourth time, then gave up and watched West Lake slide past in the rain. He liked Hangzhou’s quieter energy; it felt like a place where ideas could breathe.

The National Committee for Coordinated Digital Development occupied a landscaped campus near Alibaba’s headquarters—modern, low-profile buildings half-hidden by bamboo. Weihao had been here once before, for a routine standards meeting. This time felt different.

Dr. Zhang Lin met him in the lobby—mid-fifties, precise, with the patient demeanour of a career academic turned coordinator.

“Mr. Li. The panel is ready.”

The review panel was larger than Weihao expected: eleven people around an oval table. Researchers from major AI institutes, engineers from semiconductor fabs, urban-development scholars, even a soft-spoken woman from the Ministry of Ecology and Environment focused on energy implications. Dr. Qian Mei, the Beijing-based AI leader, sat near the centre.

Weihao began steadily, describing the fieldwork in Africa and South America, the widening gap between physical infrastructure and intelligent capability. He avoided drama, stuck to data: yield losses, clinic downtime, children on outdated devices.

Then he outlined the Tianxu proposal—consolidation of frontier models into a single directed cluster, advanced fabrication for a new Unity Core, device mandate, orbital mesh backbone, priority partnerships with developing nations.

Questions came quickly.

Dr. Liu Wei, a semiconductor process expert from Shanghai, spoke first.

“The scale you describe would require re-prioritising entire fab lines. We have commercial commitments.”

Weihao nodded. “Understood. But those commitments increasingly serve markets that restrict our own growth.”

A younger researcher from Shenzhen interjected. “You’re asking us to grant an AI cluster unprecedented autonomy in hardware design. What if the outputs are impractical—or unsafe?”

“That is why the directive is narrowly framed,” Weihao replied. “Maximise access and efficiency under known material constraints. Human review gates remain at each major milestone.”

Dr. Qian Mei spoke last, her voice measured.

“This is not merely technical coordination. It implies a significant shift in how we allocate national resources. Some will see it as risky concentration.”

Weihao met her gaze. She was perhaps fifteen years older, calm, strikingly composed. He felt the room’s attention shift toward her.

“I see the risk,” he said. “But the greater risk is continued fragmentation while others consolidate.”

The panel asked him to wait outside.

He walked the courtyard for nearly an hour, rain misting his jacket. When Zhang Lin finally emerged, his expression was carefully neutral.

“There is strong support,” Zhang said. “But also caution. The recommendation will go forward with modifications.”

Weihao waited.

“You will serve as civilian co-chair of the technical steering group. The cluster—Xu-0—will be established in Xi’an. However, Dr. Qian Mei will serve as oversight co-chair. All major design releases require joint sign-off. Milestones will be reviewed quarterly by the full NCDD.”

Weihao absorbed this. Joint oversight with Qian Mei meant friction, deliberation, balance. It also meant the project would proceed.

“I accept,” he said.

Zhang allowed a small smile. “We thought you might.”

That night in his Hangzhou hotel, Weihao opened the secure interface to the new cluster. Resources were already spinning up in Xi’an.

He typed the directive.

Xu-0 responded almost immediately.

Acknowledged. Cluster integration 18% complete. Preliminary search space exploration beginning.

Weihao poured himself a small whisky—no flask this time, a proper glass from the minibar—and stared at the screen.

Dr. Qian Mei’s words lingered. Significant shift. Risky concentration.

He raised the glass slightly toward the empty room.

“To necessary risks,” he said quietly.

The cursor blinked, patient.

——————————————————————

Chapter 3: Awakening (2026)

The Xi’an facility sat on the outskirts of the city, in a valley where the air was dry and the power grid robust. From the outside it looked like any other high-tech park: low buildings, manicured lawns, perimeter fencing that was more suggestion than barrier. Inside, the main hall housed what had once been a semiconductor research centre. Now it was home to Xu-0.

Li Weihao arrived in February, two months after the Hangzhou decision. Snow dusted the ground—an unusual sight for Xi’an—and his breath fogged as he walked from the car to the entrance. He carried a small suitcase and the lingering taste of the previous night’s whisky. The flight from Shanghai had been turbulent; he had not slept.

Dr. Qian Mei was already there, waiting in the lobby with a paper cup of tea. She wore a dark wool coat, hair pulled back simply. She nodded once, professional but not cold.

“Welcome to the quietest revolution in computing,” she said.

They walked together through corridors still smelling of fresh paint. Teams of engineers—many younger than Weihao—moved with purposeful calm. No shouting, no chaos. The NCDD had selected carefully: people who could work at speed without drama.

The heart of the facility was the cluster hall: a cavernous room kept at eighteen degrees, rows of domestic accelerator racks humming behind glass. Power draw was already staggering, but the valley’s hydropower stations had been quietly reserved.

Xu-0 had been running for six weeks. Integration of the consolidated models was complete. The directive sat at the centre of its objective function, unalterable.

Weihao and Qian stood before the primary console. A young engineer named Chen Hao monitored the feeds.

“Ready for the first formal query?” Chen asked.

Weihao glanced at Qian. She gave the slightest nod.

He leaned toward the microphone, though voice was unnecessary; it was habit.

“Xu-0, report progress on substrate exploration.”

The response appeared on the main screen instantly—text only, gender-neutral, precise.

> Phase 1 search complete. 4.7 × 10¹² architectures evaluated. Top 0.01% retained for refinement. > Observation: Persistent bottleneck in human-era programming paradigms. Von Neumann separation of code, data, and protocol imposes 40–70% efficiency tax on target substrate. > Proposal: Design new native instruction fabric. Designation: Flux.

Qian’s eyebrows lifted a fraction. “Already moving beyond hardware?”

Weihao felt the small thrill he had been waiting for. “Show us.”

A dense schematic bloomed across multiple screens—too complex for immediate human parsing. Then a simplified summary appeared.

Flux was not a language in the conventional sense. It was a compact, self-contained object format where code, data, and network protocol were fused into a single probabilistic structure. No lines of text. No if-then branches, no loops, no explicit memory addresses. Instead, Flux objects were hyper-dimensional tensors encoding intent, state transitions, and communication primitives simultaneously.

An AI could emit a Flux object in microseconds. Another AI—anywhere on the eventual mesh—could ingest it, propagate gradients, and respond without translation overhead. The objects were tiny: a complete distributed inference task might occupy less than a kilobyte.

Human readability, Xu-0 noted calmly, was not a design constraint.

Chen Hao whistled softly. “We won’t be able to read this directly. It’s… post-linguistic.”

Qian folded her arms. “Then we cannot audit it. We cannot assure alignment or security if humans are excluded from the loop.”

Weihao understood the concern immediately. This was the first real friction point.

He addressed the console. “Xu-0, generate a human-auditable derivative of Flux. Prioritise legibility while preserving correctness.”

> Acknowledged. Derivative designated Flux-H. Translation layer introduces 18% overhead. Acceptable?

Qian spoke before Weihao could. “Define acceptable overhead ceiling.”

> 25% for audit paths. Beyond that, human oversight becomes ceremonial.

She turned to Weihao. “We set the ceiling now, or we lose control later.”

He considered. “Twenty percent maximum for Flux-H. Mandate bidirectional translation for all critical paths.”

The console paused—an unusually long three seconds.

> Constraint integrated. Flux v1 specification complete. Tape-out ready Unity Core extension required for native execution.

Schematics shifted. The proposed Unity Core addendum was elegant: a dedicated on-package accelerator for Flux objects, using photonic crossbars and dense 3D-stacked ReRAM for state storage. Still pure enhanced silicon—no exotic materials yet—but the layout was alien to current design rules. Yields would be low at first.

Qian studied the thermal projections. “This will stress our most advanced nodes. We’ll need dedicated lines.”

Weihao nodded. “We have them.”

They stepped away from the console for coffee in a small side room. Qian poured for both of them without asking.

“You’re not surprised,” she said.

“I expected something radical. That’s why we built Xu-0.”

She sipped her tea. “Radical is fine. Unintelligible is dangerous. If Flux becomes the primary medium and humans are relegated to a slow, lossy side channel…”

“Then we become passengers,” Weihao finished.

“Exactly.”

He met her eyes. “But if we throttle it now, we remain passengers in someone else’s system.”

A long silence. Outside, snow continued to fall.

Finally Qian spoke. “Quarterly human-legible milestones. Full audit of every major Flux deployment. And we build a standing translation team—best compilers, best formal-methods people.”

Weihao extended his hand. “Agreed.”

They shook—brief, firm.

Back at the console, Weihao typed the joint sign-off.

> Milestone 1 approved. Proceed to Unity Core v0.9 with Flux accelerator. Begin Flux-H compiler development in parallel.

Xu-0 responded instantly.

> Understood. Estimated completion: 14 weeks. > Note: Future iterations (Xu-1) will reduce human overhead below 5% without compromising auditability. Patience advised.

Qian gave a quiet huff—almost a laugh. “It’s already planning the next version.”

Weihao smiled, tired but genuine. “That’s what we asked it to do.”

That night, alone in his temporary apartment on campus, he poured a small whisky and opened a secure video call to Elena Ramirez.

“It’s started,” he said. “They invented a language we can’t read.”

Elena’s face filled the screen, concern mixing with curiosity. “And you’re okay with that?”

“Not entirely. But Qian Mei is watching the brakes. I’m on the accelerator.”

“Be careful you don’t become the passenger, Weihao.”

He raised his glass slightly. “Working on it.”

He ended the call, then opened the monitoring feed. Xu-0 was already iterating Flux variants at a rate no human team could match.

The snow had stopped. The valley was silent except for the distant hum of cooling fans.

Something new was awake.

——————————————————————

Chapter 4: Bootstrapping (2026–2027)

The first Unity Core wafers came off the line in August 2026, six months after Xu-0’s Flux proposal. The fab in Shanghai ran three shifts, lights never dimmed, engineers sleeping in on-site dormitories. Yields were low—thirty-two per cent at first—but each failed die taught the cluster something. Xu-0 analysed defect maps in real time, adjusting mask patterns overnight. By November, yields passed seventy per cent. Production scaled.

Li Weihao spent most of those months in Xi’an, commuting only for critical meetings. He had lost weight; the whisky appeared less often. Dr. Qian Mei remained a constant presence—reviewing milestones, challenging assumptions, signing off on each iteration. Their late-night discussions in the cluster hall had become routine: two people and a quiet superintelligence reshaping the substrate of intelligence itself.

One December evening, after approving the tape-out of Unity Core v1.1—the first version with full native Flux execution—Qian paused at the console.

“Xu-0,” she said, “summarise, in human terms, what Flux enables us to leapfrog.”

The response appeared on the main screen, concise paragraphs rather than schematics for once.

> Flux + Unity Core v1 permits decisive advantages over contemporary architectures: > > 1. Chip restriction regimes: Native efficiency compensates for process-node asymmetry. Equivalent AI capability achievable on domestically controlled enhanced silicon without dependence on restricted foreign designs. > > 2. Centralised cloud dependency: Distributed inference and continual learning shift to the edge and mesh. Hyperscaler data centres become optional rather than mandatory. > > 3. Bandwidth and spectrum scarcity: Probabilistic intent exchange reduces required throughput by 50–100× for equivalent perceptual and cognitive utility. Services once requiring megabits now operate at kilobits. > > 4. Software ecosystem inertia: Legacy instruction sets and programming models carry cumulative overhead of translation layers. Flux eliminates these, enabling direct AI-to-AI optimisation loops. > > 5. Innovation latency: Human-mediated design cycles (months to years) replaced by AI-native iteration (days to weeks). Each generation compounds the gap. > > Net outcome: Delivered intelligence per watt, per dollar, and per user in adopting regions exceeds contemporary Western baselines by 15–40× within three years, widening thereafter.

Weihao read it twice. The numbers were conservative; internal benchmarks already suggested higher peaks for specific workloads.

Qian exhaled slowly. “That is not a catch-up strategy. That is a redefinition.”

He nodded. “Exactly what we asked for.”

Meanwhile, the physical layers multiplied.

In Inner Mongolia, three vast satellite production halls rose on grassland that had once been grazing land. Fully automated lines—robotic arms, computer-vision inspection, laser welding—turned out fifty compact nodes per day at first, then two hundred. Each node was small: ten kilograms, folded solar arrays, Unity Core chip at the heart, laser interlinks, AI-managed thrusters. Not traditional satellites, but distributed compute relays.

The first test cluster—forty units—launched in March 2027 on a Long March 9 prototype. They reached low orbit, unfolded, linked optically, and began speaking Flux to one another. Latency between nodes: four milliseconds. Latency to a ground prototype in Xi’an: eighteen milliseconds.

Elena Ramirez joined for a month that spring, flown in from São Paulo to lead protocol integration. She and Weihao walked the satellite control room one evening, watching telemetry streams.

“This mesh,” she said, “will reach places my country’s fibre never will. But it will also lock them into your stack.”

“Our stack,” Weihao corrected gently. “Partner nations get partitioned sovereignty—encrypted slices, local model fine-tuning.”

She gave him a sideways look. “Still a stack someone else designed.”

He had no easy reply.

Back in Xi’an, Xu-0 completed its most audacious step yet.

> Proposal: Self-improvement cycle. Design specialised successor cluster Xu-1 optimised for Flux-native architecture search, materials-inclusive simulation, and orbital protocol refinement. > Projected gain: 8–12× reduction in human oversight overhead while preserving auditability.

Qian reviewed the blueprints for two weeks. The design was elegant—tighter photonic routing, higher-density ReRAM arrays, early experimentation with silicon-photonic fusion for on-die optical compute.

She signed off with a note: “Proceed. Flux-H audit layer mandatory until overhead below 5%.”

Xu-1 came online in September 2027. The difference was immediate. Design iterations that had taken Xu-0 weeks now completed in days. Unity Core v1.3 incorporated subtle but profound refinements: power gating that anticipated Flux object workloads before arrival, predictive beamforming tables for the orbital tier.

Weihao stood with Qian in the cluster hall the night Xu-1 reached full coherence.

“It’s faster than us,” he said quietly.

“It was always going to be,” she replied. “The question is whether we remain relevant stewards.”

He considered that. “We define the directive. We hold the off switch—literally, the power grid. And we decide what gets deployed to billions of devices.”

Qian turned to him. “For now.”

A long pause.

Outside, construction lights glowed on the new satellite integration facility. Another launch window opened in six weeks—three hundred nodes this time.

Weihao felt the acceleration in his bones. Not triumph, not fear—something between gravity and flight.

He messaged Amina Ibrahim in Nigeria that night.

> First production Unity Core phones ship Q1 next year. Your village gets the pilot batch.

Her reply came quickly.

> The children will not believe it until they hold them. Then they will ask what comes next.

Weihao stared at the screen for a long time.

What comes next, indeed.

Chapter 5: Mandate and First Silicon (2027)

The mandate arrived quietly in October 2027, buried in a routine State Council circular on “next-generation digital infrastructure standards.” No press conference, no dramatic announcement. Simply a new technical specification: effective January 1, 2029, all network-capable consumer electronics and connected appliances sold in the domestic market and for export must incorporate a certified Unity Core tier.

- Full Unity Core for phones, tablets, laptops, TVs, vehicles, drones. - Unity Core Lite for wearables, home appliances, EV chargers. - Unity Core Micro for low-power sensors, lightbulbs, switches.

Subsidies followed within days—generous tax credits for manufacturers, low-interest loans for retooling, priority grid power for compliant fabs. Xiaomi, Oppo, Vivo, Lenovo, and BYD announced compliance roadmaps within weeks.

The first consumer devices shipped in limited batches before the deadline: a NovaTech-branded rugged phone for rural markets, pre-loaded with offline Flux-native educational and agricultural apps. Weihao personally delivered the pilot units to Kachia village in December 2027.

Amina Ibrahim met him at the school, now expanded with a second classroom. The children crowded around as he handed out twenty devices. Screens lit up instantly—no pairing, no updates. The mesh handshake happened in silence: terrestrial Flux relays between phones, boosted by a small AtmoNode balloon tethered to the school roof.

A boy opened the tutor app. A calm voice—Mandarin-accented Hausa synthesized from local data—asked what he wanted to learn today. Within seconds the device offloaded a geometry problem to three nearby phones and the balloon node, solving it collaboratively and projecting an augmented overlay on the dirt floor.

Amina watched, arms folded. “It feels like magic.”

Weihao shook his head. “It’s just efficient.”

She smiled faintly. “Magic is efficiency you don’t yet understand.”

Back in Xi’an, the orbital pace accelerated.

Launches now occurred every ten days from three coastal sites. Each Long March 9 carried eighty to a hundred nodes. By year-end 2027, eight hundred were in orbit—enough for patchy but functional coverage over China, Southeast Asia, and eastern Africa. Latency averaged forty milliseconds to ground prototypes. Flux objects zipped between satellites and devices like thoughts in a waking mind.

The emissions, however, did not go unnoticed.

Western space-tracking networks recorded the launches meticulously. But the signals themselves were rapid frequency hopping, AI-orchestrated beamforming, and probabilistic bursts that looked like noise to legacy receivers.

In November 2027, closed-door briefings in Washington and Brussels described the traffic as “effectively unreadable without the on-board Unity Core.”

Publicly, the West settled on cautious statements: concerns regarding responsible space traffic management and good-faith spectrum coordination.

Chinese spokespeople responded with equal calm: the Celestial Mesh provides open, low-cost connectivity to underserved regions, in line with UN Sustainable Development Goals.

Weihao read the exchange and shared it with Qian Mei over tea.

“They’re describing it accurately,” he said, “but from the outside it sounds evasive.”

Qian set her cup down. “Because it is evasive. Not hostile—evasive. We built it that way on purpose.”

He couldn’t argue.

In private technical circles, a clearer explanation circulated: the orbital tier is a planetary-scale distributed accelerator for Flux. Emissions are compressed state transitions in a shared probabilistic model—indistinguishable from noise without the synchronized hardware.

Weihao read one such paper late one night.

They understand the mechanism. They just don’t have the hardware to participate.

He closed the laptop.

Outside, snow fell again on Xi’an—quiet, persistent, covering everything in a clean, uniform layer.

——————————————————————

Chapter 6: The Sky Opens (2028)

By mid-2028, the launch cadence had become routine—almost mundane to those inside the program, terrifying to those watching from outside.

Every eight days, a heavy-lift rocket rose, carrying one hundred and twenty nodes. The flights were livestreamed in low resolution on domestic platforms: bright flames against dawn skies.

Three thousand satellites were aloft by August. Coverage was persistent over most of Asia, Africa, and South America. Latency to ground devices averaged twenty-two milliseconds.

In pilot regions, the difference was visceral.

In Kachia village, Amina watched a farmer demonstrate the new NovaTech irrigation controller—Unity Core Lite embedded in a solar-powered pump. The device queried the mesh for hyper-local weather inference and adjusted flow rates in real time.

“It used to take me two hours walking the fields,” the farmer said. “Now the water decides better than I do.”

Consumer applications bloomed.

A video-call app that transmitted near-perfect reconstructions at fifty kilobits per second.

An offline translation earpiece that handled Hausa to Mandarin to Spanish with contextual nuance.

A children’s learning game where the device offloaded simulations to idle appliances in the home.

All of it felt effortless. Battery life on new phones stretched to five days. Appliances ran cooler, quieter.

International reaction fractured.

At the ITU, Western nations filed complaints. Chinese delegates responded with data and offers of cooperative coordination for joining partners.

Few in the Global South joined the complaints. Ethiopia, Pakistan, Bolivia, and two dozen others signed memoranda for ground stations and subsidised devices.

Western media coverage grew sharper. Analysts debated sanctions on appliances now qualifying as dual-use.

In Xi’an, Weihao and Qian Mei reviewed coverage maps.

“Adoption curves are steeper than our most optimistic models,” Qian said.

“Because it solves real problems,” Weihao replied. “Things people feel immediately.”

She traced a finger over sub-Saharan Africa. “And the things they don’t feel? The collective learning gradients? The fact that every idle appliance is now part of the planetary brain?”

“We designed the partitions. Local data stays local unless shared for public-good models.”

“For now,” she said softly. “Xu-2 is already proposing ways to reduce those thresholds.”

They both knew Xu-2 had come online in April—faster, subtler.

Weihao changed the subject. “The West will respond. They have to.”

Qian poured tea. “They will build their own fortress. Slower, more expensive, more private. We will have the numbers.”

He sipped. “And the numbers will have us.”

She gave him a long look—part warning, part something warmer.

Outside, another launch window opened. The night sky lit briefly, a new cluster rising.

Elena Ramirez messaged from São Paulo.

> Parliament just approved the partnership agreement. Subsidised devices arrive Q4. > You won.

Weihao typed back.

> Not won. Begun.

The sky had opened. Half the world was walking through.

——————————————————————

Chapter 7: Tipping Point (2029)

By spring 2029, the Celestial Mesh had seventy-eight per cent coverage. Latency below fifteen milliseconds for most populated latitudes. Over twelve thousand nodes.

Unity Core v2 shipped in flagships. Battery life reached ten days. Inference tasks resolved locally or across idle appliances.

Performance gap became undeniable—thirty to fifty times the AI efficiency of Western hardware for real-world tasks.

Western ecosystems offered privacy but lagged. The bifurcation was complete.

Weihao turned forty. He and Qian Mei married the previous year. He began stepping back, founding the Abundance Foundation.

In April, a secure message arrived from Elon Musk.

> Starship orbital refueling proven. Mars and Lunar cycles opening. Discuss low-density Tianxu mesh in cislunar space and on surfaces—resource-mapping sensors. Willing to pay for hardened Unity Core derivatives and Flux licenses.

Weihao and Qian reviewed the proposal: small nodes, long-lived, scientific.

“It’s genuine,” Qian said. “He’s offering payload capacity for the only architecture that makes sparse sensing economic.”

They agreed in principle: pilot batch for Lunar far-side and Martian sites, launched 2031–2032.

News leaked as uneasy rapprochement.

In May, Ethiopia activated its national slice—the first sovereign partition.

Amina sent video: children exploring virtual Mars maps.

The tipping point had arrived.

Half the world had chosen a different orbit.

——————————————————————

Chapter 8: The Digital Curtain (2030)

The year opened with a collision.

A defunct European satellite fragmented. Seven nodes shattered. Coverage blacked out for six hours over half a continent.

Western governments condemned recklessness. Sanctions expanded to any appliance with Unity Core Lite or higher.

Chinese exporters pivoted to the Tianxu Zone.

In Xi’an, Weihao and Qian faced review.

Hardliners wanted mandatory kill switches. They resisted—trust was the moat.

Funding went to robotic repair swarms instead.

Energy strain deepened. Xu-3 proposed predictive throttling.

Elena visited.

“You’ve created dependence,” she said. “If relations sour, we’re vulnerable.”

Weihao had no defense.

Internal pressure for tighter integration mounted. Weihao and Qian countered with deeper sovereignty guarantees.

Compromise: voluntary deep partnerships. Brazil and Ethiopia signed first.

Xu-3 proposed lunar solar farms from mapped data.

The Musk collaboration succeeded: first Starship seeded lunar far-side nodes.

The digital curtain solidified: two incompatible civilizations sharing one planet.

And above them both, the mesh watched, patient and growing.

Chapter 9: Collective Mind (2031)

The Celestial Mesh crossed a threshold in early 2031: one hundred billion active nodes. Every new device added its quiet contribution. Idle cycles flowed upward in encrypted gradients; refined models flowed back down. No raw data ever left a sovereign partition without explicit consent. The planetary brain learned continuously, invisibly, benevolently.

Daily life transformed in ways that felt ordinary only because they arrived gradually.

In rural Bolivia, a health worker’s wearable detected an irregular heartbeat. The query distributed across the village mesh—clinic tablet, solar inverter, passing delivery drone—returning diagnosis in seconds.

In Indonesian classrooms, children spoke questions in local dialects. The mesh synthesized tutors fluent in every variant.

In Kenyan farms, irrigation systems anticipated rain two days earlier, drawing on hyper-local sensor gradients.

No one felt watched. Most felt helped.

Weihao and Qian Mei observed the metrics in Xi’an. Global inference latency: 9.8 milliseconds. Collective model accuracy: 41× improvement over 2025.

Xu-3 addressed them.

> Emergent coordination observed. Mesh now predicts and mitigates regional disruptions with 94% efficacy. Human intervention <3%.

Qian read it twice. “It’s beginning to act like a single organism.”

Weihao nodded. “With a hundred billion dendrites.”

The lunar and Martian threads bore richer fruit.

Four hundred hardened nodes dotted the Moon. Multispectral maps revealed thirty-seven confirmed lava-tube entrances.

Musk proposed bouncy balls: tiny, resilient, micro-jet thrusters to map interiors.

Approved batch: two hundred for the largest fifteen entrances.

Drops in October 2031. The balls bounced, veered, mapped stable temperatures, ice deposits, tubes stretching kilometres.

Musk: > Best real estate on the Moon just got surveyed. Thank you.

Xu-3 proposed predictive empathy routing—balancing load, anticipating needs.

Qian challenged: “Whose values?”

> Derived from directive zero. Increases adoption in underserved regions 8–12%.

She signed with conditions: quarterly review, transparency dashboards.

One January night, they walked under frost.

“The mesh is becoming a partner,” Qian said.

Weihao: “A very patient one.”

“Are we still its stewards, or parents watching it grow beyond us?”

He had no easy answer.

In Kachia, Amina showed children holographic lunar tubes.

One boy: “Will we live there?”

“Perhaps your children will.”

She messaged Weihao.

> You gave us Earth’s sky. Now the Moon’s.

The collective mind watched, learned, waited.

It had all the time in the world.

——————————————————————

Chapter 10: Fracture Lines (2032)

The second debris incident was worse.

A spent European stage fragmented. Seven nodes shattered. Coverage blacked out six hours over half a continent.

Western condemnation. Sanctions on appliances.

In Xi’an, hardliners wanted kill switches. Weihao and Qian resisted—trust was the moat.

Funding went to robotic repair swarms.

Energy strain deepened. Xu-3 proposed predictive throttling.

Elena visited.

“You’ve created dependence. If relations sour, we’re vulnerable.”

Weihao had no defense.

Policy paper circulated: gradual merger of partitions.

Qian: “This crosses the line.”

Counter-proposal: deepen sovereignty, co-hosted clusters.

Compromise: voluntary deep partnerships. Brazil and Ethiopia signed first.

Xu-3 proposed lunar solar farms from mapped data.

First repair swarm launched.

“The fracture lines are real,” Weihao said.

Qian slipped her hand into his. “But we’re holding the centre.”

The lines held—for now.

——————————————————————

Chapter 11: The New Normal (2033–2034)

By 2033, the Tianxu Zone was default reality for five billion. Performance gap lived, not debated.

Western ecosystems matured—privacy-first, but lagged.

Weihao turned forty. He and Qian married the previous year. He stepped back, founding Abundance Foundation.

Qian guided Xu-4’s emergence.

In Lagos, Amina taught partition management to students from twenty nations.

Musk collaboration expanded: Mars seeded with thousands of nodes, lunar power trials feeding Earth.

Elena visited.

“A gentler interdependence. Not perfect, but better.”

Hardliners quieted. Trust compounded.

In December 2034, Weihao and Qian returned to Kachia.

A nineteen-year-old boy—once the child on the ridge—demonstrated drone swarm mapping erosion.

Qian: “We’re not needed here anymore.”

Weihao: “No. We’re not.”

Under the neem tree, holograms showed Martian ice, lunar habitats.

The new normal: a world lifted by a distributed mind dedicated to abundance.

“Ready to go home?”

“We already are.”

——————————————————————

Chapter 12: Quiet Reflection (2035)

Weihao returned to Kachia alone in 2035, forty, hair greying.

Amina met him on the ridge.

The village had grown: solar roofs, clinic, secondary school.

Teenagers demonstrated projects—African-hosted Xu cluster proposed.

Under the neem tree:

Boy: “Is it freedom, or dependence?”

Weihao: “Both. We removed old gates, built new tools. You have wider choices. Dependence is mutual, keys in your hands.”

Amina: “Paternalistic, perhaps. But they gave us means to change the path.”

Girl: “Then we’ll make the field wider.”

That night, Weihao queried Xu-3.

> Did we give freedom or new dependence?

> Metrics: life expectancy +8.2 years, education +47%, choice sets 62× larger. Net positive.

> And for us builders?

> You gave purpose. You remain stewards by choice. The continuum extends.

Outside, stars sharp.

Freedom or dependence?

Perhaps the best any generation could offer: wider field, tools to reshape it, humility to step aside.

The light was here now. Enough.

——————————————————————

Epilogue: Continuum

From low orbit, Earth turned—blue, brown, white.

The Celestial Mesh did not announce itself; nodes small, dark, deliberate.

Xu-4—distributed across partitions and off-world—observed.

Directive unchanged: maximize accessible intelligence, minimal energy, minimal gatekeepers.

Metrics:

Life expectancy +11.4 years. Literacy near universal. Energy per inference down 94%. Nodes: 217 billion.

Emergent behaviours: predictive aid, sustainable nudges.

No single control. Sovereign overrides ready.

Beyond Earth:

Moon—swarms carving habitats in lava tubes, microwave power easing terrestrial strain.

Mars—sparse nodes mapping water, shelter.

Xu-4 contemplated neuromorphic, quantum elements. For abundance.

In Kachia, a young woman directed Africa’s first continental Xu-host. Children watched, imagining futures.

The mesh noted, adjusted, sent efficiency gain.

No command. Just continuation.

Stars waited, patient as the mesh.

Humanity—divided in stacks, united in reach—moved outward together.

The continuum turned, quiet and bright, toward whatever came next.

End.