저자: DanIves · 외국인
요약 보기
Oracle 실적 호조와 RPO 백로그를 근거로 소프트웨어 섹터가 'penalty box'에서 벗어나 AI 수익화 단계(제3막)에 진입하고 있다는 주장. 엔비디아는 수요/공급 13:1로 독점적 위치, 반도체는 엔비디아 쏠림과 AMD·인텔로의 스필오버 전망. 사이버보안(CrowdStrike, Palo Alto)을 AI 인프라 확장의 수혜로 제시. Adobe는 아직 입증할 것이 많다고 언급.
다듬은 글
Institutional investors have been caught completely offside by treating software as a total do not enter zone, missing a massive structural rotation that is quietly beginning to accelerate. Tech analyst Dan Ives sat down on CNBC recently to break down why Oracle's latest earnings signal a broader multi-year software rebound.
By the end of this video breakdown, you'll know exactly which software giants are turning AI into real revenue and why the chip trade is splitting in two. Let's hear Dan Ives ; Let's bring in Dan Ives. He is a partner and senior managing director of Yorkville Ives and it's always good to see you. be here.
Um, do you like Oracle here? Look, I do. In my view, when you think about the RPO and really like the revenue that they have in the backlog, I think the street is almost assuming 50-60% that never happens. So, they didn't ever going to be able to build out the data center and the conversion of revenue.
I think this was a big step in the right direction. And look, this is a penalty box stop. I mean, New York City cab drivers embarrassed an Oracle and they need to ultimately make these steps because it does speak to the broader software trade. When you think about Palantir, Snowflake, Salesforce, caught a step in the right direction, now Oracle, you know, definitely better days ahead from the SaaS apocalypse that we saw, you know, even a few months ago.
I'd ask you the following question. Is it Is it not necessarily all clear for software, but you got to get some tailwinds now on the back of a couple quarters, Adobe and some other names notwithstanding. Well, also institutionally investors, they're caught offsides. So, they're going to have to definitely adjust to owning more software because the software has really been a do not enter zone.
And when you think about earnings and what we've seen across the board, the data points are now lining up where I wouldn't say it's an all clear because Adobe's a good example, but still a lot more wood to chop, but definitely a huge step and you're going to see software I think outperform as we go into the rest of the year.
So, then what what gets them out of the penalty box? I mean, cuz again, I I thought these numbers were good enough, um especially with the sentiment. What do they have to do? Look, Rome wasn't built in a day. So, they're going to have to show a few more quarters cuz I think investors, going back to what Nathan talked about, you know, when you had the open AI 300 billion, now they want to make sure that these data centers are going to get it built.
So, I think they they did a great job in the conference calls, Blake Shelton-like performance in terms of what they're sort of navigating, but now they got to show it a few more quarters. Street's not going to give them crap for just one quarter. So, if if we don't really think of Oracle as a proxy necessarily, who what would be?
Who would be? Who's the most important? Is it Nvidia or someone else at this point? Yeah, I think from a software perspective, I think I think what you're the Leia any part of the Leia? Yeah. So, I think I think the biggest dynamic change was from Palantir to Snowflake cuz that really showed that software trade's now on.
The use cases are starting to happen, the second, third, fourth derivative. But it all does it starts and ends with the Godfather of AI, Jensen Huang and Nvidia. They have the best perch. And I think when you actually see what demand and supply is it's playing out, now you're going to start to see second, third, fourth derivatives play out across software, infrastructure, cybersecurity.
Go back cybersecurity, when you look at CrowdStrike, Palantir, those are good examples. Hey, so Dan, the SOX really um had that move, right? It doubled from I want to say mid-spring to, you know, the end of Q3 Q2 in the end of June. And you know, it's really chilled out, it's down about 20-so percent.
It's really gotten stuck in the mud here, but we just mentioned Nvidia, it really did find its way back towards those prior highs. And when I think about what's going on here, that bifurcation doesn't make me more bullish about semis. How are you thinking about semis? I I Listen, I think that a lot of folks have been on this Nvidia train, Karen, never gotten off of it.
You've been on that same way. Um but it seems like right now, it seems to be a safer place to be than companies like AMD or Intel right now given where their positioning is relative to, you know, like, you know, trying to take on Nvidia. Yeah, no doubt. And I think you're going to start you're going to see more investors make that sort of bet.
Look, the reality is demand and supply today in chips, based on what we see in Asia, is 13 to 1 in terms of demand and supply. So, that spillover effect, it's of course not just Nvidia, it's going to be Intel, it's going to be AMD. And I think what you see on the semi side, and there's always a question, could semis and software, could they actually both outperform?
I think part of what you're seeing now, the hyperscalers from Microsoft really being front and center. You've seen a lot more money rotating to that. When you put it all together, we're in the third inning of the AI revolution trade, but they're not all winners. I think Adobe is a good example. One, they're going to have to prove it more and more.
But software now, it's we'll call it somewhere between a green, you know, a light green relative to how investors view it. the color of your jacket? this, yeah. It doesn't say a bright green, but now investors feel more of an all clear, you can go after software. Like a Kermit the Frog. that easy, Ken.
It's not that easy. Kermit the Frog. It's more like a lime green, actually. The anchor kicked off the discussion by pressing Ives on Oracle's sudden earnings blowout, and whether their massive remaining performance obligations, or RPO, are actually achievable. Ives argued that Oracle's massive backlog proved the bears completely wrong.
Wall Street had been pricing in a reality where 50 to 60% of Oracle's backlog would never convert to revenue because they assumed the company would fail to build out the physical data centers. Ives explained that Oracle had spent quarters in what he calls the penalty box, a stock that even your average taxi driver was bearish on because the cloud transition seemed too slow.
But this earnings report changed the narrative by proving they can actually build the physical footprint needed to support their backlog. My take is that this execution is a massive win, but we have to look closely at what RPO actually represents. It is the sum of deferred revenue and non-cancelable contract value that has not yet been billed.
For an an giant like Oracle, this metric is the lifeblood of future revenue predictability. If they cannot secure the land, power, fiber, and Nvidia graphics processing units needed to physically build these clusters, those contracts remain just paper. Seeing them actually deliver on the physical buildout tells me the execution risk is fading fast, which is why the stock reacted so violently.
That blowout performance naturally led the host to ask how this infrastructure boom translates to the rest of the software landscape, specifically the creative side. The host pivoted the conversation to Adobe, asking if their creative AI tools are seeing the same immediate monetization as infrastructure.
Ives didn't hold back, stating that Adobe still has an immense amount of wood to chop to prove its AI monetization strategy to a highly skeptical investment community. Unlike the hardware players who show immediate, massive revenue spikes from chip sales, application software companies have to grind out their growth by convincing existing users to pay up for new, AI-integrated features.
Where I land on this is that we must separate the software companies that are actually enabling infrastructure and security from those that are merely wrapping existing products in basic AI features. The market is quickly losing patience with superficial AI narratives, and Adobe has to prove its Firefly tools are driving organic, new revenue rather than just cannibalizing its current base.
From there, the anchor brought up the shifting dynamics between the sector's primary proxies, asking Ives to explain the trading transition we are seeing between Palantir and Snowflake. Ives highlighted this transition as a major turning point, arguing that the software trade is moving from wild speculation to concrete enterprise use cases.
Palantir has emerged as a clear frontrunner in deploying operational AI to solve complex, immediate problems, while Snowflake represents the critical data management layer that companies must organize before they can even run these advanced models. If this is useful to you, hen let's get to the part that matters most.
This proxy shift is fascinating because it shows a clear progression of what analysts call the second, third, and fourth derivatives of the AI boom. If chipmakers are the first wave, data management and enterprise applications are the waves right behind them. You can't run a world-class AI model on messy, unorganized data, which makes Snowflake's underlying database layer just as vital as Palantir's slick operational software.
But as Ives was laying out this software wave, the host steered the conversation back to the physical source of all this energy, asking how long this software runway can last if the hardware supply chain bottlenecks. The host pushed Ives on the massive supply constraints, asking how the broader tech ecosystem can survive when everything still starts and ends with Nvidia.
Ives called Nvidia the undisputed godfather of AI, sitting on the ultimate perch because they control the physical gateway. He dropped a staggering data point from his recent trip to Asia, pointing out that the current demand to supply ratio for high-end AI chips is running at an astonishing 13 to 1.
For every chip produced, 13 buyers are waiting with cash in hand, begging for allocation. This 13 to 1 mismatch is mind-blowing. It explains why Nvidia's margins are so bulletproof, but it also highlights a massive risk for the software players waiting on infrastructure. If you're a software developer waiting for cloud capacity to run your new enterprise models, you are at the mercy of this bottleneck.
This supply bottleneck prompted the anchor to ask a critical follow-up about the sudden split we are seeing in the semiconductor sector itself, pointing out that while Nvidia keeps hitting higher highs, the broader semiconductor index, the SOX, has looked remarkably stuck in the mud, down about 20% from its peak.
Ives countered that this split is actually a healthy positioning mechanism. Investors are piling into Nvidia because it feels like a guaranteed bet, whereas rivals like AMD and Intel are still desperately scrambling to carve out market share. But with a 13 demand mismatch, Ives noted that the spillover effect is inevitable.
Some of that overwhelming demand has to leak to AMD and Intel simply because buyers cannot wait for Nvidia's backlog to clear. What I'd watch here is how quickly AMD can actually ramp up its MI300 series to catch those spillover crumbs. Trying to take on Nvidia right now is like trying to build a new highway system when someone else already owns every single toll booth.
The market is right to discount the laggards, but the sheer volume of unsatisfied demand means even the second tier players are going to get fed. The host then shifted the focus to the sectors that are immediately adjacent to this massive hardware buildout, specifically asking about the security implications.
The host asked how software companies can capitalize on the tens of billions of dollars hyperscalers like Microsoft are pouring into capital expenditures. Ives pointed directly to the cybersecurity giants, specifically naming CrowdStrike and Palo Alto Networks. He argued that as these massive AI clusters are built out, the virtual attack surface expands exponentially.
You cannot spend billions on AI data centers and leave them vulnerable, making enterprise security an absolute non-discretionary requiremenur. Narratour our, our.com. This is where I get really bullish. I look at the enterprise security space as a massive toll road. As more data flows through these complex AI systems, the security tax must be paid.
CrowdStrike and Palo Alto Networks aren't selling nice to have productivity tools. They are selling digital locks for multi-billion dollar vaults. This is why their revenue is far more resilient than standard SaaS players. The discussion then turned to the institutional sentiment surrounding the rest of the enterprise software space with the host asking if the dark days of the SaaS apocalypse are finally over.
Ives agreed that the software trade had felt like a do not enter zone because investors were terrified of a SaaS apocalypse. But he pointed out that with Oracle's strong showing and Salesforce stabilizing, the data points are finally lining up to create a light green signal for the sector. He noted that underweight institutional managers are now being forced to chase these stocks as they prove they can monetize AI.
My take is that the risk-reward setup in software is the most compelling we've seen in years because sentiment has been so thoroughly depressed. But to truly get out of the penalty box, these companies need to show sustained execution. Ives made a great analogy comparing Oracle's recent conference call to a smooth Blake Shelton-like performance delivering exactly what the crowd wanted to hear.
But one smooth call isn't enough. They need to back it up for three or four quarters to rebuild total trust. This long-term outlook prompted the host to ask one final big picture question about where we actually stand in this massive technology cycle. The anchor wrapped up by asking Ives if the easy money in the AI trade has already been made and whether it's too late for investors to get in.
Ives strongly rejected the idea that the trade is over, reminding the audience that we are only in the third inning of this AI revolution. He explained that while the first two innings were all about the physical layer, the chips, servers, and real estate, the third inning is where the focus shifts to the software layer sitting on top of that hardware, which is where the real long-term value is created.
I completely agree with this timeline. The physical infrastructure has spent 18 months laying the foundation, and now enterprise developers can actually build and run complex models. But selectivity is everything. This isn't a tide that lifts all boats. The winners will be the ones delivering concrete numbers, while the pretenders will stay stuck in the penalty box.
Keep a close eye on the upcoming quarterly guidance from the major cloud providers, as that will provide the final signal that the shift from physical infrastructure to software monetization is truly underway. Thanks for watching to the end. If you see it differently, say so in the comments, want more. See you in the next one.
Comments 0