Owning the Inputs Isn’t the Same as Owning a Moat

Today, I’m digging into Yandex N.V. (YNDX).

Yandex used to be Russia’s Google, Uber, Amazon, and DoorDash rolled into one company. After selling its Russia-based businesses, the company re-emerged as Nebius Group, an AI infrastructure business focused on GPU-powered cloud computing.

At a high level, they appear more like a capital allocator instead of a hard tech business (I'm probably shortchanging them here, but I wanted to get the point across).

Right now, their biggest advantages look like:

• Access to capital to build data centers

• Access to energy to power them

• Access to GPUs to fill them

Those matter today, but they’re not great long-term moats on their own.

Why? Capital, power, and GPUs tend to get more available over time. Currently, the vibe in the market is that "this time is different" due to nearly all of the large hyperscalers talking a lot about how compute-constrained they are.

I tend to lean very bullish when it comes to the impact that AI is going to have on the economy. But I also sprinkled with a dose of reality. That the crazy high gross margins of companies like Nvidia are not going to remain in perpetuity.

Prices fluctuate, supply chains adjust, and competitors with enough money can catch up. That could makes these advantages temporary than what it feels like today.

I tried to think about how a data center company could widen its moat in the future.

I am far from an expert in this domain, so I poked around via some AI-fueled deep research, and here's what I came up with. They could:

• Build deeper switching costs by tightly integrating AI workflows, tooling, and deployment into customer operations

• Develop proprietary infrastructure software that materially lowers customer costs versus generic cloud options

• Turn scale into better economics through smarter workload routing and power optimization

That all makes sense to me!

There are also downsides business as it sits today:

• The business is still losing a lot of money

• CapEx is huge

• If utilization slips, returns can fall fast

The main takeaway here is this: owning the inputs is not the same as owning the advantage. Capital can buy speed, but distribution, process, product, and customer lock-in what are needed to build a true moat?

If you’re building something capital-intensive, this is a good reminder to ask: what actually keeps customers with me once others can afford the same inputs?

With that, I'll see you tomorrow!

Nick

TL;DR

• Yandex N.V. no longer operates as a Russian internet conglomerate. In 2024, it divested its Russia-based assets and rebranded as Nebius Group N.V., pivoting hard into AI infrastructure.

• Today, Nebius sells GPU-powered cloud infrastructure and AI-native services, plus maintains smaller bets in autonomy and education.

• Financially, this is a CapEx-heavy, loss-making business with improving gross margins but massive upfront investment.

• The entrepreneur takeaway is sharp: forced pivots can work, but only if you fully reset the product, capital allocation, and narrative at the same time.

The 30,000-Foot View

Nebius Group is an AI infrastructure company. Its core product is cloud computing optimized for training and running AI models, including GPU compute, storage, and managed services. Think less "general-purpose cloud" and more "AI-first data centers with software wrapped around them."

Revenue mix is still evolving, but FY2024 shows a clear shift toward infrastructure:

• Nebius AI Cloud: ~55% of revenue

• TripleTen (edtech bootcamps): ~23%

• Toloka (AI training data, now deconsolidated): ~21%

• Avride (autonomous driving): <1%

Key stats (most recent disclosed):

• Market cap: ~$19.6B

• TTM revenue: ~$363M

• TTM gross margin: ~59%

• TTM operating profit: ~$539M loss

• Employees: ~1,370

• Industry: AI infrastructure and cloud computing

For operators, this is not a software-margin business. It is a capital deployment business where returns depend on utilization, pricing power, and access to GPUs and energy.

Company History

• 1997: Yandex launches as a Russian search engine.

• 2011: Yandex N.V. IPOs on Nasdaq under ticker YNDX.

• 2022: Sanctions and geopolitical pressure fracture the company’s global structure.

• February 2024: Binding agreement announced to divest Russia-based assets.

• May to July 2024: Divestment closes in stages. Yandex exits Russia entirely.

• August 2024: Company rebrands as Nebius Group N.V.

• 2025: Nebius accelerates AI infrastructure build-out and announces hyperscaler partnerships.

Show Me the Money

Stand-out financial features:

• Revenue growth is explosive off a small base, with TTM revenue tripling FY2024.

• Gross margin inflected positive as cloud utilization improved.

• Operating losses remain massive due to aggressive capacity build-out.

• CapEx is the story, this is infrastructure-first economics.

• R&D and product development spend remains high relative to revenue.

Financial Data

The N.O.O.B. Nine — Competitive Powers

The Nerd Out on Business Nine is made up of Hamliton Helmer's famous "7 Powers" of competitive advantage (Scale Economies, Network Economies, Counter-Positioning, Switching Costs, Branding, Cornered Resource, and Process Power) combined with two of my own (Data Flywheel and Distribution Advantage).

Average Score: 2.6/5 - Nebius has raw inputs for a moat, but defensibility is still up in the air.

Memorable Marketing

Nebius markets through credibility, not creativity. The strategy is trust transfer through partners, milestones, and scale signals.

Notable tactics:

• Nebius Rebrand Reset (2024)

  • Hook: Clean break from Russian operating history.

  • Channels: PR, investor comms, enterprise sales.

  • Why it worked: Removed buyer friction in Western markets.

• Partner-Led Credibility Stack (2024)

  • Hook: Signal legitimacy via top-tier partners.

  • Channels: Press releases, sales decks.

  • Why it worked: Buyers shortcut diligence through association.

• Hyperscaler Proof Points (2025)

  • Hook: Land massive enterprise customers to validate reliability.

  • Channels: Media, analyst coverage.

  • Why it worked: One whale unlocks many smaller deals.

Tactical takeaways:

• Borrow trust aggressively when entering skeptical markets.

• Turn internal milestones into outward-facing marketing.

• Rebrands should remove friction, not add polish.

• Proof beats promises in enterprise sales.

AI Uses & Opportunities

Current uses:

• AI-optimized cloud infrastructure for training and inference.

• AI tooling for deployment, monitoring, and optimization.

• Applied AI in autonomy and education platforms.

Future opportunities:

• AI-driven workload routing to minimize compute and energy costs.

• AI-assisted sales engineering to shorten enterprise sales cycles.

• Predictive churn detection based on workload behavior.

• Vertical-specific AI clusters for regulated industries.

Bumps in the Road

• Forced divestment reshaped the entire company.

• Material weaknesses disclosed in internal controls.

• Extreme capital intensity with delayed payback.

• Share dilution risk from equity programs.

• Reporting complexity due to discontinued operations.

• Ongoing shareholder litigation tied to the 2024 transaction.

Your Swipe File

• If your business is CapEx-first, capital access and balance sheet management are critical. When the money is flowing, credit can be easy to come by. Eager lenders/investors often disappear when markets tighten.

• Gross margin improvement does not equal a healthy business. Long-term free cash flow is the North Star I suggest as the ultimate KPI (along with the KPIs that flow into the factors that impact that metric, obviously.)

• Big customers validate you but also concentrate risk.

• When you pivot, pivot completely.