Key Points
-
OpenAI’s growing compute needs have led it to form new partnerships beyond Microsoft (NASDAQ: MSFT), including with Google (NASDAQ: GOOGL), signaling unexpectedly strong infrastructure demand.
-
Oracle (NYSE: ORCL) reported a surge in AI-related revenues, with infrastructure growth forecasted to accelerate from 50% to 70% and backlog (RPO) expected to double, reinforcing signs of industry-wide AI momentum.
-
Despite earlier fears of a pullback in AI spending post-Deep Seek, high model usage retention and ongoing training needs suggest accelerating demand into 2026, contrary to prior Wall Street expectations.
- Are you ahead, or behind on retirement? SmartAsset’s free tool can match you with a financial advisor in minutes to help you answer that today. Each advisor has been carefully vetted, and must act in your best interests. Don’t waste another minute; get started by clicking here.(Sponsor)
Watch the Segment on AI Stocks Soaring in June
Notes on This Segment
- In the early months of the year, many AI stocks sold off on fears that Microsoft (Nasdaq: MSFT) was backing out of data center contracts. Bears pointed to this being evidence that investment in AI was unsustainable and would soon crater.
- We repeatedly poured cold water on this story on the AI Investor Podcast.
- First: Microsoft had publicly broken up with OpenAI – its main AI customer. Demand they had previously booked was simply moving to Oracle – the company behind OpenAI’s ‘Stargate’ Project.
- Second: Despite all the talk of pulling back on spending, Microsoft itself was still projecting growth in its capital spending next year.
- Fast forward to today:
- Oracle reported earnings recently and saw its stock soar.
- The headline: The company projects that growth rates in cloud infrastructure will grow from 50% this year to 70% next year.
- And the company’s RPO (think of this as backlog) is increasing at over a 100% rate.
- And beyond Oracle, news has recently broken that OpenAI will also be partnering with Google for additional demand.
- When it comes to AI, there are two primary costs for a company like OpenAI:
- Training models
- ‘Inference’ – or running them
- Training: After DeepSeek, many believed that training costs would plummet.
- Yet, the opposite is actually happening.
- For example, OpenAI can’t move forward with its next-generation GPT5 model until it gets more compute capacity.
- And Inference demand has been absolutely exploding this year.
- OpenAI itself is now at a run rate of more than $10 billion in annual revenue and continues to grow fast.
- The company is seeing some of the highest retention among any consumer applications. The reason for that incredible retention: people are using the product!
- The big question for 2026 that will determine how most AI stocks perform…
- If OpenAI is racing to find more computing capacity and might be stalled until Stargate comes online
- That means all the other companies building frontier models are likely facing similar constraints.
- If that’s the case, 2026 will see unexpectedly high growth. Stocks across the AI space are are north as projections for AI spending in 2026 and beyond continue to increase.
Don’t Miss an Episode of the AI Investor Podcast
This segment was from our AI Investor Podcast, where we discuss the biggest news in artificial intelligence and what it means for the stocks in your portfolio. You can watch recent episodes on our YouTube channel or on any major podcast service.
Transcript:
[00:00:05] Austin Smith: Open AI has an insatiable appetite for compute. Uh, they famously paired up with Microsoft in their early days. Now they’re pairing up with anyone and everyone who will give them compute, including Google (
[00:00:20] Eric Bleeker: Yeah. One of the key storylines we covered at the beginning of the year was there is a major sell off in AI stocks because of DeepSeek. There’s geopolitical reasons, and a third major reason was reports about Microsoft basically scaling back their ambitions.
[00:00:34] Eric Bleeker: Now, we had said that we weren’t as concerned with this because number one, Microsoft very publicly had broken up with Open AI. So if you’re looking at Microsoft decreasing spending. Well, if they had a breakup with their biggest partner, and you need to watch for where demand was headed with, you know, its new partnerships.
[00:00:52] Eric Bleeker: And in spite of all that, despite Microsoft losing its biggest and best customer, they were still projecting growth next year on their capital expenditures. So let’s fast forward today. Oracle reported last night, Wednesday night. The stocks up 8%. The headline is they’re predicting a massive acceleration.
[00:01:13] Eric Bleeker: AI revenues next year, their growth rate for uh, infrastructure is supposed to go from 50 to 70%. And their RPO, which you could think of it like a backlog that’s projected to grow another a hundred percent. And then you have the news that Google. Is going to work with open AI, you know, giving them one more, uh, source for this compute demand.
[00:01:37] Eric Bleeker: So, you know, Austin, I think the big picture is there are two main costs when we’re looking at all these companies in the data center chain, whether it’s Nvidia (NASDAQ: NVDA) or up to all these optic plays that we’ve talked about. One cost is training, the other cost is actually running the models themselves. So the training costs.
[00:02:00] Eric Bleeker: The fact is a lot of the belief post deep seek was that training simply too expensive. And we’re going to see companies pull back on this. We’re seeing the opposite right now, which is open ai. They cannot move forward with their next major generation, GPT5 until Stargate is online. And then they have a second wave, which is inference Austin, right now, open ai.
[00:02:28] Eric Bleeker: They are seeing 90% retention after one month, 80% retention after six months. It’s not just the fastest grown consumer product ever. It’s also the most sticky. You know, that’s the best in breed companies in terms of retention would kill for 90% after a month would kill for 80%. So there’s huge usage. And then there’s the side that this additional training is also needed.
[00:02:55] Eric Bleeker: So, you know, I would say Austin. The big bullet point here is just we’re beginning to see a storyline of acceleration into 2026, far beyond what Wall Street’s been estimating months ago, let alone today. Because if Open AI’s waiting for more capacity to do this pre-training, people thought was going to moderate.
[00:03:16] Eric Bleeker: All the other big companies are going to have to follow that same game plan. And number two, the consumer usage of AI is exploding, which is causing storylines like open AI, going to Google. So we’ll cover this more and more, but this is, this is the main theme of today’s show that. When we started the show back in 2024, we said, well, we’re not gonna see the slowdown people expect in 2025, but that’s gotten pushed into 2026, and that’s what you worry about.
[00:03:48] Eric Bleeker: And now we’re starting to see 2026, not just be a period of moderation or fears of worst case scenario. There’s a lot of talk of acceleration, which might be crazy to people watching because. You know what? Everything is wrapped in fear. We had this deep seek moment. What if AI stops and all of a sudden that narrative, it’s really changing when you see what people are saying.
[00:04:13] Eric Bleeker: You see the actions of these companies.
The post The #1 Reason AI Stocks Are Soaring Again in June appeared first on 24/7 Wall St..
