The AI Infrastructure Arms Race Just Got a Credit Line

Alphabet sold $20 billion in bonds, including a 100-year note, to fund $185 billion in AI capex. OpenAI started selling ads. Stripe hit $140 billion without going public. Three different bets on how to pay for the future of tech, three completely different playbooks. The common thread: everyone's scrambling to fund infrastructure, and nobody agrees on the business model.

OpenAI starts testing ads in ChatGPT. The company that spent two years insisting subscriptions would fund the AI revolution just put sponsored content at the bottom of free-tier conversations. Ads are rolling out to U.S. users on the Free and $8/month Go plans: Plus, Pro, and Enterprise stay clean. The timing is telling: Anthropic ran a Super Bowl ad last week explicitly promising no ads in Claude. Sam Altman called it "clearly dishonest." Meanwhile, OpenAI expects ads to make up less than half of revenue long-term. Translation: they need the money now, and they'll figure out the ratio later. Read More…

Alphabet raised $20 billion in its largest-ever bond sale, upsized from $15 billion after drawing $100 billion in orders, to help fund up to $185 billion in AI capex this year, more than double its 2025 spend. The deal includes a rare 100-year sterling bond, the first from a tech company since the dotcom era. When a $2 trillion company borrows on a century-long timeline to build data centers, it's not hedging, it's telling you AI infrastructure is the bet. Combined hyperscaler capex is now forecast at $650 billion for 2026. Morgan Stanley expects $400 billion in hyperscaler bond issuance this year alone. Read More…

Stripe is arranging a tender offer at $140 billion, Bloomberg and Axios reported, up from $107 billion last fall and nearly 3x its 2023 low of $50 billion. The Collison brothers continue to perfect the art of staying private while giving employees liquidity every six months. Stripe processed $1.4 trillion in payments in 2024, posted ~$19.4 billion in 2025 gross revenue, and has been profitable since 2024. At $140 billion, Stripe is worth more than Adyen, Square, and most public fintech companies. Why IPO when the private market pays more and asks fewer questions? Read More…

Waymo raised $16 billion at a $126 billion valuation, the largest private funding round ever for an autonomous vehicle company. Fifteen months ago, the Alphabet subsidiary was worth $45 billion. It just nearly tripled.

Dragoneer, DST Global, and Sequoia led the round, joined by a16z, Tiger Global, Fidelity, T. Rowe Price, Silver Lake, and Mubadala. Alphabet itself remains the majority investor, putting up the bulk of the capital. When your parent company writes most of the checks, and Sequoia still shows up, the signal is clear: smart money thinks this works.

The numbers are starting to back up the thesis. Waymo provided 15 million rides in 2025 (triple the prior year) and now runs 400,000 paid rides per week across six U.S. metros. The company reportedly crossed $350 million in ARR. That puts the valuation at roughly 360x revenue, insane by any traditional metric, but investors aren't paying for today's rides. They're paying for the 20+ cities Waymo plans to launch in 2026, including Tokyo and London, its first international markets.

What's changed since autonomous driving's first hype cycle is simple: Waymo actually works. 127 million miles logged, 90% fewer serious injury crashes than human drivers. Competitors pivoted (Uber), stalled (Cruise), or died (Argo AI). Every mile Waymo drives generates data that makes the system better, a flywheel that's nearly impossible to replicate.

But the bull case has cracks. Waymo robotaxis illegally passed school buses 19 times in Texas. A vehicle struck a child near a California school in January. During a San Francisco blackout, robotaxis stalled at intersections. Not showstoppers, but exactly the kind of incidents that give regulators pause, and Waymo needs regulatory approval in every new city it enters.

The investor syndicate tells a story worth watching. Dragoneer and DST specialize in pre-IPO positioning. Fidelity and T. Rowe Price write checks when they see a clear path to public markets. This round has the look of a pre-IPO staging deal, park capital at a premium valuation, ride it to a listing. At $126 billion, Waymo would be one of the largest tech IPOs ever. The question is whether public markets will pay a premium on top of an already astronomical private valuation, or whether this round's investors are the last ones buying at these prices.

For the broader autonomous vehicle market, the signal is unmistakable: one company won, and everyone else is fighting for second. The $16 billion raise makes that gap even harder to close.

Cerebras Systems raised $1B Series H at ~$23B valuation, led by Tiger Global, with Benchmark, Fidelity, AMD, Coatue, and Altimeter participating. That's nearly 3x the $8.1B valuation from its Series G just five months ago. Cerebras makes wafer-scale AI processors, an entire silicon wafer as a single chip, and recently signed a reported $10B+ deal with OpenAI for 750 megawatts of compute. Tiger leading is notable: the firm has been quiet for months, and this is a big re-entry into AI hardware. Read More…

ElevenLabs closed a $500M Series D at $11B valuation, led by Sequoia. a16z quadrupled its investment; ICONIQ tripled. The voice AI company hit $330M ARR, more than tripling its valuation in a year. ElevenLabs is expanding beyond voice into agents and video, essentially betting that audio is the interface layer for AI interactions, not just a feature. Founded in 2022 by two Polish childhood friends, the company now has 400 employees across 13 cities. IPO talk is already circulating. Read More…

Positron AI raised $230M Series B at $1B+ valuation, co-led by Arena Private Wealth, Jump Trading, and Unless, with strategic backing from Qatar Investment Authority and Arm. The Reno-based startup builds energy-efficient AI inference hardware, its Atlas system claims to match Nvidia H100 performance at one-third the power. The tell: Jump Trading co-led after deploying Atlas as a customer first. A trading firm investing in its own inference hardware vendor is the strongest signal that the product actually works. Positron's next-gen Asimov chip targets 5x Nvidia's upcoming Rubin GPU on tokens-per-watt. Read More…

Turnstile launched with $29M in funding for a quote-to-cash platform aimed at sales-led startups. The pitch: replace the Frankenstein stack of billing tools that early-stage companies cobble together. Not flashy, but the pain point is real: broken billing infrastructure quietly kills startups.

Daytona raised $24M Series A for infrastructure powering AI agents. The NYC-based company builds programmatic, composable computers that agents can spin up and operate autonomously. As agents move from demos to production, the plumbing underneath them is becoming its own investment category.

Accrual launched with $75M in funding for AI-native accounting automation. The company targets preparation and review workflows at accounting firms—the kind of tedious, rules-heavy work that AI is genuinely good at. With accounting firms facing talent shortages, the timing is right. The funding size says investors see this as infrastructure, not a feature.

Andreessen Horowitz closed $15 billion across multiple strategies—its largest fundraise ever. The headline number: $1.7B earmarked for AI infrastructure, led by GP Jennifer Li, whose portfolio includes OpenAI, ElevenLabs, Cursor, and Black Forest Labs. Another $1.7B goes to apps, $1.176B to American Dynamism (defense/security), and $700M to bio/health. Li's thesis: 2026 is the "AI super cycle," agents are "finally real," and the talent crunch, not capital, is the bottleneck for AI-native startups. What a16z is deliberately avoiding: commodity GPU clouds and capital-intensive data centers. They want the software control layers, not the physical infrastructure. Read More…

Cerebras is widely expected to file for IPO in the coming months—the $23B valuation and Tiger-led round look like pre-IPO positioning. If Cerebras goes public in H1 2026, it would be the first major AI chip IPO since the boom began and a critical test of whether public markets will pay private-market multiples for Nvidia alternatives.

The a16z $15B close may trigger competitive responses from other mega-funds. Nobody wants to be seen as under-resourced in an AI arms race.

Alphabet's 100-year sterling bond is still being priced—watch for final terms. If it's oversubscribed, expect Oracle, Microsoft, and Amazon to follow with their own long-dated issuances.

ElevenLabs hinted at additional strategic investors to be disclosed later in February. Nvidia invested in September 2025; don't be surprised if other hyperscalers follow, looking to lock in voice AI capabilities.

The Waymo investor list reads like a pre-IPO cap table. If Waymo files confidentially this quarter, it sets up one of the largest tech IPOs in years.