Latest / Tech IPO Conversations with Fexingo: Public Software Companies and Capital Markets / How Post-IPO Companies Use AI Data Centers as Growth Anchors
Transcript
- Lucas: So Meta just signed its first AI data center deal in India with Reliance — that's a gargantuan move, and it got me thinking about how post ipo companies are increasingly using infrastructure like this as a growth anchor. Luna: A growth anchor — I like that phrase. Because it's not just about building compute for today's models, it's about signaling to the market that you're investing for the next five years. Lucas: Exactly. And this is a shift I'm seeing across the board with recently public companies. Instead of plowing all their free cash flow into buybacks — which we talked about a few episodes ago — they're committing billions to data centers, power procurement, and even battery storage. Luna: GM jumping into the battery race for AI data centers caught my eye too. That's an automaker positioning itself as an energy infrastructure play. Lucas: Right — that headline from a few days ago, GM joining the race to build batteries for AI data centers and the grid. It shows how far the tentacles of AI infrastructure reach. But let's focus on the post ipo angle. Take Reddit, for example — they went public in March 2024, and by early 2025, they'd announced a $200 million annual commitment to Google Cloud for AI training and inference. Luna: And at the time, people questioned whether a social media company needed that kind of compute. But now, looking at their ad revenue growth tied to ai powered recommendations, it's starting to make sense. Lucas: That's the key — these infrastructure investments aren't just cost centers, they're revenue drivers. Reddit's ai driven ad targeting helped them grow ad revenue by over 40% year-over-year in their most recent quarter. The data center spend is directly linked to that top-line acceleration. Luna: So the market is rewarding companies that can tell a convincing story about how their AI infrastructure will generate returns. But what about companies that over-invest and don't see the payback? That's a risk. Lucas: Absolutely. And that's where the discipline comes in. Look at Meta — they're spending an estimated $800 million on that single Reliance deal. But Meta has a proven track record of monetizing engagement through ads. For a smaller post ipo company, that kind of bet could be existential. Luna: Right, so the bar for credibility is higher. You need to show investors a path to ROI, not just a big check. Lucas: And that's why we're seeing more post ipo companies use a phased approach. Instead of building a massive data center upfront, they lease capacity from cloud providers like AWS, Azure, or Google Cloud. That's what Reddit did — it's a variable cost model that scales with revenue. Luna: But eventually, if you're successful, the variable cost becomes so large that building your own infrastructure starts to make economic sense. It's the classic build vs. buy decision, now applied to AI compute. Lucas: Exactly. And some companies are skipping the cloud phase entirely. I was looking at the S-1 for a recent AI startup that went public — they're planning to spend $300 million on their own data center in the first year post ipo. That's a bold statement. Luna: That's a huge chunk of their IPO proceeds. It's almost like they're saying, 'Trust us, the demand is real, and we need to own the hardware to capture the margin.' Lucas: And that ties back to the Meta deal — Meta is essentially doing the same thing, just at a much larger scale. They're building data centers in India to serve the next billion users, and they're using AI to personalize content and ads in dozens of Indian languages. Luna: So the data center becomes a moat, not just a cost. If you can train models on local data with low latency, competitors can't easily replicate that. Lucas: Right. And this is why post ipo tech companies are increasingly using AI infrastructure as a narrative driver for their stock. When they announce a new data center deal, it's a signal to investors: we are serious about AI, we are investing for growth, and we have the capital to do it. Luna: But there's a counterpoint — some investors prefer companies that return capital via buybacks rather than spending on unproven infrastructure. Especially in a high-interest-rate environment. Lucas: That tension is real. We saw it with Reddit — after their IPO, they had a lot of cash from the offering, and some analysts wanted them to do a buyback. Instead, they committed to the Google Cloud deal. And so far, the market has rewarded them — their stock is up over 50% from the IPO price. Luna: So the key is communication. Companies need to explain how the infrastructure spend will drive measurable outcomes — user growth, ad revenue, margin expansion. Lucas: Precisely. And that's where we see a split. The ones that do it well — like Meta and Reddit — get a premium valuation. The ones that just say 'we're spending on AI' without a clear thesis get punished. Luna: One interesting case is Lucid Motors — they're more of an EV play, but they had a top executive depart this week as part of a leadership shakeup. Lucid has talked about using AI in manufacturing, but they're not a pure AI infrastructure story. Their capital allocation is under a different kind of scrutiny. Lucas: Right, and that's the thing — not every company can or should follow the Meta playbook. If your core business isn't digital, pouring money into AI data centers might be a distraction. Lucid needs to focus on vehicle production and cash burn first. Luna: So for post ipo companies, the question is: does AI infrastructure fit your business model and your stage of maturity? If you're a high-growth software company, probably yes. If you're a hardware manufacturer, maybe not. Lucas: And that's the nuance we don't hear enough in the coverage. Every company is different. But the trend is clear — for tech IPOs in 2025 and 2026, a credible AI infrastructure plan is becoming table stakes for a successful public debut. Luna: Before we wrap, I want to circle back to something you mentioned earlier about the market rewarding these investments. It really does come down to storytelling with numbers. Lucas: Absolutely. And that kind of honest, numbers-driven conversation is exactly what this show is about. We keep it ad-free and focused because a small group of listeners chip in monthly through buy me a coffee dot com slash fexingo. That support literally funds making this many episodes — so thank you if you're one of them. Luna: Yeah, it's a small community, but it makes a difference. And it keeps us independent — no sponsor pressure to hype things we don't believe in. Lucas: Exactly. So back to infrastructure — I think one more thing worth noting is the role of government incentives. In the US, the CHIPS Act and IRA are providing tax credits for domestic data center construction, which lowers the cost for post ipo companies. Luna: That's a good point. It makes the build decision more attractive, especially for companies that want to control their supply chain. Lucas: And globally, we're seeing countries like India offer incentives to attract AI infrastructure. That Meta-Reliance deal probably got some favorable tax treatment. So the decision isn't just about technology — it's about geopolitics and economics. Luna: So the takeaway for this episode might be: post ipo companies that can credibly tie AI infrastructure investment to revenue growth are positioning themselves as long-term winners. But it's not a one-size-fits-all strategy. Lucas: Right. And as we watch the next wave of tech IPOs — especially in AI and SaaS — we'll see which companies get the infrastructure story right. That's going to be a key differentiator in the coming years.