The Compute Gold Rush: How AI Data Centers, EVs, and Biotech Are Rewiring the Next Decade

The New Oil Barons: Compute Becomes the World’s Hottest Commodity

If the 20th century was defined by oil, the 2020s are being carved out by GPU-hours and data-center acreage. This week, SpaceX’s IPO filing pulled back the curtain on just how expensive the AI training arms race has become — and it is staggeringly, almost comically, large.

According to the S-1, SpaceX spent $12.7 billion in capital expenditures on AI infrastructure in 2025, roughly 61 percent of its total capex. That is more than the entire GDP of some small nations, poured into racks of accelerators in Memphis, Tennessee. Yet even with that scale, SpaceX ran into trouble: latency issues between Colossus I and two other sites more than 10 miles apart, compounded by aging local network infrastructure, forced Elon Musk’s company to rent out capacity to outsiders just to keep utilization healthy.

Anthropic’s $15 Billion Bet

The first tenant was Anthropic, which signed a deal to pay SpaceX $1.25 billion per month through May 2029 — roughly $15 billion annually. To put that number in context, it nearly doubles SpaceX’s total 2025 revenue of $18.7 billion. The agreement includes a 90-day mutual exit clause, a necessary hedge in an industry where competitive positions shift faster than semiconductor roadmaps.

For Anthropic, the deal is existential. The company is on the brink of its first quarterly operating profit, with projected revenue of at least $10.9 billion for the quarter ending in June 2026 — more than double the $4.8 billion it reported for the March quarter. Yet Claude, its flagship model, competes directly with xAI’s Grok. Paying Musk’s company for compute while simultaneously battling Musk’s AI is either the ultimate realpolitik of the AI era or a sign of desperate scarcity. Probably both.

Google’s Short-Term Pivot

Google followed Anthropic, signing its own deal to pay SpaceX $920 million per month from October 2026 through June 2029. In a statement, Google called it a “short-term” agreement designed to meet “surging customer demand” for its Gemini Enterprise agent platform. The phrasing is telling: Google, with its own Tensor Processing Units and massive cloud footprint, is still having to rent capacity from a rocket company because demand for AI agents is outpacing its ability to provision internal infrastructure.

This is the hidden infrastructure story of 2026. The model training and inference workloads that power chatbots, code assistants, and autonomous vehicles are eating the world’s semiconductor supply. Data-center construction is running into local opposition — a recent Gallup survey found 70 percent of Americans oppose having a data center built near them — yet demand shows no sign of slowing.

When Hollywood Meets the Machine: Calls for an AI “Rating System”

While data-center czars haggle over billions, AI’s impact is being felt in creative industries. Director Gore Verbinski, known for Pirates of the Caribbean, reignited the conversation about transparency in AI-generated content during this year’s Tribeca Film Festival.

Verbinski stopped short of condemning AI outright. He acknowledged that a young indie filmmaker with limited resources might reasonably turn to generative AI to produce a key scene they could not otherwise afford. But he drew a sharp distinction between tool use and authenticity. “You almost need a rating system,” he said. “If you use AI to write a script, you get an F. What people are most afraid of is that there is no transparency. People are afraid of what is real and what isn’t.”

The comment touches on a tension that will define the next few years: generative AI is no longer just a novelty — it is producing work indistinguishable from human craftsmanship. Yet the industry has no standardized disclosure norms. A “rating system” for AI-generated content is one proposal, but any solution will require cooperation between studios, streaming platforms, regulators, and the creative community itself.

Meta, meanwhile, is taking a different tack on accessibility. The company announced it is donating its AI-powered smart glasses to more than 130,000 blind veterans in the United States, using computer vision and audio description to give visually impaired users real-time scene understanding. It is a reminder that the same technology driving corporate profit motives is also enabling genuinely life-changing assistive applications.

Electrifying the Roads: EVs, Autonomy, and the Cybertruck’s Murky Future

The AI infrastructure boom has a direct corollary in transportation: autonomous vehicles require the same inference horsepower that data centers are scrambling to build. And the EV market, after years of hyped growth, is entering a more turbulent phase.

Tesla continues to dominate headlines, but not always for the right reasons. The Cybertruck, once the poster child for EV disruption, remains a niche product with pricing and range figures that keep it out of reach for mainstream buyers. Meanwhile, legacy automakers are accelerating their EV rollouts. Hyundai’s Ioniq 5 N has been praised for combining EV efficiency with genuine performance-car dynamics, signaling that the “EVs are slow and boring” narrative is finally dead.

More significant is the build-out of charging infrastructure. The United States has been slow to match the density of networks in Europe and China, but 2026 is seeing a wave of new Supercharger-compatible stations and government-backed corridor investments. For consumers still sitting on the fence, the charging experience — once the single biggest barrier to EV adoption — is improving rapidly.

Full autonomy remains the elephant in the room. Tesla’s Full Self-Driving beta continues to expand to new cities, but regulatory clarity in the United States is fragmented. The NHTSA’s framework for autonomous vehicle reporting is voluntary rather than mandatory, leaving consumers to navigate a patchwork of state-level rules. Meanwhile, Waymo’s robotaxi operations continue to expand in selected markets, offering a glimpse of what fully driverless urban mobility could look like — and highlighting just how far personal autonomous vehicles still have to go.

The Biology Revolution: Fungal Superhighways, Ancient Genes, and Gut-Brain Links

While technologists wrangle over compute and cars, biologists are making discoveries that would have seemed like science fiction a decade ago.

Researchers this week published the first comprehensive quantification of global arbuscular mycorrhizal fungal networks — the underground web of fungal threads that connects plant roots and facilitates nutrient exchange. The numbers are almost impossible to internalize: the combined length of these fungal threads is long enough to reach beyond the Solar System. These networks, invisible to us but essential to nearly every terrestrial ecosystem, are the original distributed systems, predating the internet by hundreds of millions of years.

Another study probing the origins of complex life found that the first complex cells did not acquire their genes from a single source. Instead, their genomes were assembled through successive waves of gene transfers from a diverse mix of species. It is a finding that reshapes how we understand the very nature of genetic inheritance, suggesting that horizontal gene transfer — once thought to be rare in eukaryotes — played a starring role in the emergence of everything from fungi to humans.

In medical biotech, the gut microbiome continues to yield surprises. New research strengthening the link between gut bacteria composition and Alzheimer’s disease risk is pushing the field toward a provocative question: if neurodegenerative diseases have a microbial component, could next-generation probiotics or microbiome-targeted therapies become standard preventive care? The research is still early, but the commercial and clinical implications are enormous.

The Common Thread: Physical Scarcity in an Age of Infinite Hype

Look at all these stories together and a pattern emerges. AI, EVs, and biotech are often discussed as purely software problems — better algorithms, smarter training data, more elegant models. But the breakthroughs of 2026 are being driven by physical constraints: data-center land, semiconductor fabrication capacity, charging-station density, fungal network lengths, gut bacterial diversity.

The winners in each domain are the ones who best navigate the boundary between digital ambition and physical reality. SpaceX is a rocket company at heart, and rocket companies understand logistics, supply chains, and site selection. Its AI subsidiary is succeeding not because of superior models but because it controls the physical substrate. Similarly, automakers that master battery chemistry and charging partnerships will outlast those that rely on software features alone.

For investors and technologists, the lesson is clear: the next decade belongs to integrators, not innovators in the narrow sense. The companies that can stitch together scarce physical resources — compute, energy, biological materials — into coherent products and services will define the era.

The AI compute crunch will eventually ease as new fabs come online and chip architectures improve. EV charging will reach parity with gas stations within a few years. Biotech discoveries take longer to commercialize but often deliver more durable impact. In every case, the current bottlenecks are temporary. The organizations that build resilience during this period of scarcity will be the ones holding the strongest hands when the next wave arrives.