SB Energy IPO: strong order book but delivery challenges

Bottom line: SB Energy is worth adding to your AI infrastructure tracking pool, but beginners shouldn't treat it as a "guaranteed profit" stock. It's suitable for those who understand data center construction cycles, power permits, and contract delivery; not for those who just see "$400 billion backlog" and want to jump in.

On Day 1, when I saw the report that Nvidia plans to invest up to $3 billion in SB Energy, I ran my own data center health check process. Not using their product, but treating them as a target, I fed public info into a local Python script to organize a table: orders, power assets, customers, funding sources, delivery risks. The interface was just CSVs and some Markdown; I mainly used Claude to compress long reports.

SB Energy is expected to file for its IPO as early as this week, launching its initial public offering in September with a goal to raise $5-7 billion.

Let me explain two terms. An IPO is when a company first offers its shares to the public; a backlog refers to signed contracts that haven't been fully delivered or recognized as revenue yet.

By Day 3, the most striking numbers in the table were these: on one side, over $400 billion in signed but unexecuted orders for the data center business; on the other, revenue of around $140 million, up 66% year-over-year. This isn't a contradiction—it shows huge orders, but cash comes in slowly. Data centers don't make money immediately upon signing; you need land acquisition, grid connection, construction, equipment installation, and acceptance testing. My sticking points are insufficient details on customer concentration, Ohio project delivery milestones, and natural gas generation approval progress, so I have to put question marks there for now.

A week later, looking at the IPO news again today, my judgment is clearer. The surprise appeared in the funding source column: SB Energy has expanded from solar storage to natural gas generation and data center construction, operating and under construction about 5 GW of energy assets, and secured over $18 billion in project financing. Of Nvidia's $3 billion investment, $1.5 billion might be tied to the IPO. This combination of compute, power, and capital has more potential than just selling racks.

The downsides are also direct. The bigger the orders, the more they test delivery capability. What AI data centers fear most isn't lack of customers, but power supply issues, public backlash, equipment delays, and cost overruns. I've written before that evaluating data center stocks requires looking at local opposition to site selection and the proportion of electricity costs. Now I'll add one more point: Order value does not equal cash flow; delivery capability is key to realizing valuation.

So my positioning is "wait and see." If you follow AI infrastructure long-term and want to observe SoftBank's path of connecting OpenAI, Nvidia, and power assets, the prospectus materials are worth waiting for. If you're just chasing short-term hot topics, heavy positions are not recommended.

Actionable advice: Once the IPO filing is out, don't look at the fundraising amount first. Look at how much deposit has been paid in the backlog, how many contracts have cancellation rights, and how many projects have clear grid connection timelines. Only if you can understand these three items should you decide whether to add it to your tracking pool.