Compute is the New Collateral: What January's Mega-Rounds Tell Us About 2026

The biggest funding rounds in January weren't about technology. They were about access.

Access to GPUs. Access to power. Access to data center capacity. Access to roads and permits for autonomous vehicles.

If 2023 was about "who has the best model," 2026 is about "who has the most compute, and who can prove their technology works in the real world."

This is the story of three sectors that defined January 2026: AI infrastructure, data centers, and autonomous vehicles. And why the market is paying unprecedented premiums for physical-world assets over pure software plays.

Part 1: The AI Infrastructure Land Grab

DayOne Data Centers: The $2B Premium for Power

$2B raised at 100% premium to prior round.

Let that sink in. A data center company just got valued at double what it was worth months ago. Why?

Because they have something that money increasingly can't buy: power contracts and permits.

The Math That Matters

DayOne's metrics tell the story:

~1GW of customer commitments (signed, contracted demand)
500MW already in service or under construction
500MW in future development across Finland, Singapore, Thailand, Japan, Hong Kong

In AI infrastructure terms, 1GW of capacity is worth more than 1GB of training data. Because without the power to run the compute, the data is useless.

Why Coatue is Betting on Bricks and Mortar

Coatue Management leading this round is fascinating. They're known for late-stage software growth investing. Now they're funding concrete, cooling systems, and electrical substations.

The thesis is simple: AI labs will spend billions on compute, but they can't build data centers themselves. Too slow. Too complex. Too many regulatory hurdles.

So they'll pay whatever it takes to secure capacity. And whoever owns that capacity has pricing power.

The Southeast Asia Angle

Indonesia Investment Authority's participation signals something bigger: Southeast Asia is positioning itself as the AI infrastructure corridor for APAC.

Why? Three reasons:

Proximity to manufacturing (chips, servers, networking gear)
Favorable regulatory environments (faster permitting than EU/US)
Cost advantages (land, labor, energy)

If US/EU data centers are constrained by power and permitting, APAC data centers become the release valve. DayOne is betting that constraint creates a structural pricing premium.

Global Technical Realty: $1.9B Says Europe is Underbuilt

$1.5B from KKR + $400M from Oak Hill Capital

KKR doesn't write $1.5B checks for fun. They're betting that Europe's AI infrastructure is 3-5 years behind the US, and that gap represents a massive opportunity.

The European Constraint

Europe's problem isn't demand. It's supply:

Permitting delays: 18-36 months for new data center construction
Energy policy: Aggressive renewable mandates that complicate power procurement
Density limitations: Restrictions on high-density, high-power facilities in urban areas

GTR is betting they can solve the permitting problem faster than competitors by:

Acquiring existing sites with permits already in place
Retrofitting older facilities for AI workloads
Securing long-term power contracts before breaking ground

For AI labs with European customers (GDPR, data sovereignty requirements), this is critical infrastructure.

The Private Equity Playbook

KKR's involvement also signals a broader shift: PE firms are treating AI infrastructure like they used to treat telecom towers.

The playbook:

Buy or build critical infrastructure
Secure long-term contracts (10+ years)
Generate stable cash flows
Exit via infrastructure fund, REIT, or strategic sale

If that sounds boring compared to "disrupting X with AI," that's the point. Boring infrastructure with monopolistic characteristics generates better returns than exciting software with competitive markets.

Part 2: When Autonomy Got Real

Waymo: The $110B Proof Point

$16B at $110B valuation. $13B from Alphabet.

Waymo's round isn't the biggest in dollar terms. But it might be the most important for what it signals: autonomous vehicles are no longer a science project.

The Metrics That Unlocked $110B

Waymo didn't get to $110B on potential. They got there on performance:

20M+ autonomous trips completed
$350M+ annual recurring revenue
Operating profitably in San Francisco (per reports)
Expansion into Los Angeles, Austin, and other markets

That ARR number is key. It's not hypothetical. It's actual customers paying actual money for actual rides.

At $350M ARR growing at (estimated) 100%+ YoY, Waymo is on track to cross $1B in revenue by 2027. For a technology that was written off as "10 years away" just five years ago, that's a remarkable inflection point.

Why Alphabet is Writing a $13B Check

Google's dominant contribution ($13B of $16B) sends two messages:

Message 1: We're not letting this fail. Waymo is strategic to Alphabet's long-term positioning in autonomous systems, logistics, and mobility.

Message 2: We're willing to share the capital burden. By bringing in outside investors at a step-up valuation, Alphabet is effectively:

De-risking via external validation
Sharing dilution with partners who bring operational expertise
Creating exit optionality (eventual spin-out, IPO, or strategic sale)

The Competitive Pressure

Waymo's round creates a problem for everyone else in autonomous vehicles:

For Tesla: You can't just talk about robotaxis anymore. You need to show operational metrics.

For Cruise (GM): Your safety incident put you 2+ years behind. How do you catch up when Waymo has $16B and a 20M trip lead?

For Zoox (Amazon): Your tech might be great, but can you scale operations without a Waymo-sized war chest?

The autonomy market is bifurcating: leaders with proof points can raise at massive valuations. Everyone else is stuck in the "promising tech, unclear path" limbo.

Waabi: The $1B "Software-First" Bet

$1B Series C+ (Canada)
Investors: Khosla, G2, Uber, NVIDIA, Volvo, BlackRock, ADIA

Waabi's round is the counter-narrative to Waymo: what if you could build autonomous systems with less capital by focusing on generative AI for simulation?

The Thesis

Instead of driving billions of real-world miles (expensive, slow, risky), Waabi is betting on:

Simulated training using generative AI to create realistic edge cases
Sparse data requirements by learning transferable concepts
Faster iteration via software vs. hardware-dependent testing

The investor mix is telling: Uber (wants cheaper autonomous tech), Volvo (needs a software partner), NVIDIA (selling simulation tools), and BlackRock/ADIA (long-duration infrastructure investors).

The Risk

Simulation is great until it's not. The edge cases that kill autonomous systems are often the ones you can't simulate because you didn't know they existed.

Waymo's 20M trips aren't just marketing—they're data on every weird thing that happens in the real world. That's harder to replicate in simulation.

But if Waabi can prove their approach works at 10% of Waymo's capital intensity, they rewrite the economics of autonomous development.

Skild AI: The $1.4B "Foundation Model for Robotics"

$1.4B Series C+
Investors: SoftBank, NVIDIA Ventures, Bezos Expeditions, Samsung, LG, Tencent

Skild is betting that the future of robotics isn't task-specific models—it's general-purpose foundation models that can adapt to any physical task.

Think: GPT-4, but for robotic manipulation.

Why This Matters

Current robotics is like software before APIs: every task requires custom code, custom training, and custom deployment. Skild is trying to build the "API layer" for robots.

The use cases:

Warehouse automation (picking, packing, sorting)
Manufacturing (assembly, inspection, quality control)
Healthcare (patient assistance, medication delivery)
Agriculture (harvesting, monitoring, maintenance)

The Strategic Investors

Look at who's investing: Samsung, LG, Tencent. These are hardware companies that need software differentiation.

If Skild succeeds, they become the "Android of robotics"—the software platform that every hardware manufacturer licenses. The investor group is essentially pre-buying that future.

Part 3: The Quiet Infrastructure Deals

These didn't make headlines, but they matter:

Jupiter Power: $500M Debt for Energy Storage

$500M debt facility from Barclays, HSBC, ING, Société Générale, SMBC

Battery energy storage systems (BESS) are critical for AI infrastructure because:

Power smoothing: Data centers need stable power; grid power is variable
Cost arbitrage: Charge during cheap (renewable) hours, discharge during expensive hours
Grid services: Provide frequency regulation and demand response

Jupiter's debt financing signals that BESS is now considered predictable infrastructure by banks—not risky venture tech.

Redwood Materials: $425M for Battery Recycling

$425M Series C+ from Google, Capricorn, Goldman Sachs, Eclipse, NVIDIA

Redwood is betting on a closed-loop battery economy: mine batteries, not lithium.

Why this matters for AI: data centers and EVs both need massive battery capacity. As those batteries age out, recycling them is cheaper and faster than mining new materials.

NVIDIA's investment is strategic: they need battery capacity for their AI hardware roadmap. Securing supply via recycling is a hedge against lithium price volatility.

The Big Picture: Infrastructure Over Innovation

What Changed in January 2026

The market used to pay premiums for innovation:

Better algorithms
Novel architectures
Breakthrough research

Now the market pays premiums for access:

Compute capacity
Power contracts
Regulatory approvals
Operational proof points

This is a phase transition. And it has implications:

For AI Startups

If you're building a model: You need to show why your model will be cheaper to train and serve than incumbents. "Slightly better accuracy" isn't enough anymore.

If you're building tooling: Focus on making existing infrastructure more efficient. Cost optimization, observability, and security are where the money is.

If you're building applications: Distribution is everything. Owning a channel (industry vertical, geographic market, user base) is more valuable than owning IP.

For Infrastructure Companies

If you own physical assets: You have pricing power. But you need to move fast—this window won't last forever.

If you're a service provider: Long-term contracts are worth more than spot revenue. Lock in customers now while they're desperate for capacity.

If you're solving bottlenecks: Energy, cooling, networking—these are the new venture-scale opportunities.

For Investors

Early-stage: Be ruthless about capital efficiency. The companies that win will be the ones that can build with 10% of what the leaders raise.

Late-stage: Operational metrics matter more than vision. Show me revenue, show me utilization, show me unit economics.

Growth equity: Infrastructure plays are the new software SaaS. Predictable, high-margin, defensible.

Three Predictions for Q1-Q2 2026

1. Data Center M&A Accelerates

Hyperscalers (AWS, Azure, GCP) will start acquiring data center operators to secure capacity. Expect $5B+ acquisitions.

2. Autonomy Consolidation Begins

Companies that can't show Waymo-level metrics will sell to OEMs or strategic buyers. Expect 3-5 acquisitions in autonomous vehicles by mid-2026.

3. Energy Storage Becomes a Venture Category

As data centers hit power constraints, battery systems will become fundable on their own merits. Expect a wave of $100M+ BESS rounds.

The Bottom Line

January 2026 marked a shift from "who has the best technology" to "who has access to the resources that make technology possible."

Compute is the new collateral. Proof is the new vision. Infrastructure is the new innovation.

The companies that understand this are raising at premiums. The companies that don't are stuck pitching last year's story.

Which side of that divide are you on?

Track Infrastructure Deals Before They're Announced

Fundup AI monitors data center, energy, and infrastructure deals alongside traditional VC rounds. Get alerts when companies in your sectors raise capital.

Start tracking infrastructure deals → Start Free

Subscribe to The Funding Pulse for weekly deep dives into funding trends across AI, infrastructure, and emerging tech.

About Fundup AI: We deliver real-time funding intelligence across 40+ countries, with AI-powered ICP matching and CRM integration. Learn more →