 From theCUBE Studios in Palo Alto in Boston, bringing you data-driven insights from theCUBE and ETR. This is Breaking Analysis with Dave Vellante. The data from enterprise customers is clear, but at the same time it's conflicted. While 94% of customers say they're spending more on AI this year, they're doing so with budget constraints that will steal from other initiatives. As well, where customers are running AI generally and generative AI specifically is split almost exactly down the middle in terms of public cloud or on-premises slash at the edge. In addition, developers report the attractiveness of feature richness and velocity found in the public cloud. At the same time, organizations express concerns about IP leakage, compliance, illegal risks, and costs that will limit their use of the public cloud. Hello and welcome to this week's Wikibon Cube Insights powered by ETR. In this Breaking Analysis, we'll share the most recent data and thinking around the adoption of large language models and address the factors to consider when thinking about how the market will evolve. First, let's look at the overall spending climate to get a read on enterprise technology budgets. Senior IT decision makers exited 2022 with an expectation that their budgets would increase between four and 5%. By January, that figure was down to 4.1% and despite small sequential incremental increases throughout the year, that currently stands at 2.9% well below initial expectations. Budget constraints combined with the dash to generative AI has forced organizations to re-prioritize. As we shared in Breaking Analysis and one that we did with Andy Terai late last year, the return on AI investments was elusive at the time and somewhat still remains so. The chat GPT craze has forced a top-down mandate from boardrooms and as such has shifted the spending priorities in enterprise tech. This chart shows the sectors that ETR tracks. Net score or spending momentum is on the vertical axis and pervasiveness in the survey is on the horizontal axis. It's a measure of presence. Well, all sectors felt the pinch of budget constraints in 2022. AI, which was leading all segments was suppressed to the point where by October, 2022 it fell below the 40% red dotted line. That's a high watermark for spending velocity. Chat GPT was introduced to the market in November. You see, we've highlighted October at the bottom and then one month later, Chat GPT was introduced and since then AI spending has accelerated. However, budgets really haven't changed dramatically. As we said earlier, as such, we're seeing compression in other sectors suggesting that in the near term anyway, funding for Gen AI is going to be somewhat dilutive to other segments of the market. Now, as we mentioned at the top, the data shows that for those customers spending actively on generative AI, a huge majority of customers, 94% report accelerating their AI spend in 2023. And while most customers are saying it's a modest spending increase of 10% or less, 36% say their spending will increase by double digits. So top-down pressures from the corner office to figure out generative AI is an urgent matter but the actual doing is much more challenging and complicated. This chart shows what customers are doing with Gen AI in production environments and it's not really shocking and it's hard to track what's driving revenue. It's certainly helping potentially with productivity. While 34% say they're not evaluating, that number is way down from last quarter. And while you may think that 34% is very high or one of these people talking about, we believe there is a difference in the minds of respondents between playing with generative AI and actively evaluating. So that's why the number is potentially so high. Regardless, when you look at what's actually happening in production environments, two things stand out. One, most people are still in eval mode and two, the use cases are pretty straightforward with chatbots at the top of the list followed by code generation, summarizing text and writing marketing copy is the main areas of interest today. Not really that surprising. We believe it's critical for organizations to truly understand the business case and identify ROI. The big ROI driver is going to come down to minimizing labor costs. You can put this in the productivity bucket, but at the end of the day, it's going to be about lessening the need for humans. Now, this doesn't necessarily mean that unemployment will rise. It simply means that the number one driver of value is going to be reducing headcount requirements. And that will most certainly change the skills required for employment. Now, another key challenge facing organizations while top-down momentum is high and has accelerated, deployment opens a can of risky worms. Here's a slide from a recently released study by Technalysis, an independent analyst firm run by Bob O'Donnell. It shares results from 1,000 IT decision makers on their top concerns about gen AI. Compliance, IP leakage, legal concerns, such as copyright infringement, bias, data and tools quality, et cetera. These are legitimate reasons for being careful with generative AI and how it's being used. And much of that concern leads folks to say that they're going to do gen AI on-prem. Here's some data from ETR that shows organizations report an identical mix of private and public infrastructure, i.e. public cloud or on-prem slash edge deployments. The allure of cloud is it has the best tooling, but for the reasons that we mentioned in the Technalysis survey, private infrastructure is going to be in demand. At the same time, there's now lots of data in the cloud. We think it's closer to 40 to 45% of workloads are running in the cloud today, perhaps even as high as 50%. And as we've reported, the cloud and on-prem are coming somewhat more into balance, a little bit of an equilibrium. I mean, cloud is still growing much faster, but cloud migration is not as robust as the business case for moving workloads that haven't moved isn't as attractive. So we think that much of the cloud growth is new apps or features on top of existing cloud workloads. But the fact is in speaking with developers, the cloud is exceedingly capable when it comes to AI. Here are eight points that we've highlighted that devs tell us the public cloud is delivering on today. The pace of innovation in AI, building on previous tooling like Amazon SageMaker, for example, the simplicity of integration and the productivity that it's driving for developers, allowing those developers to get to an outcome very quickly, go to thecubeai.com, thecubeai.com and you'll see an example and sign up for our private beta. Our team built this very quickly in a matter of weeks. Now we're taking more time to train based on the queries that we're getting on the inbounds, but the time to MVP was one-tenth of a normal software product development cycle. Number four here is really, really important, model optionality and diversity, not only from the cloud vendor, but third parties. Number five and number six are also critical. The ability to fence off inference requests which that the LLM vendor can't access any customer IP and security, things like ensuring data stays in the region and encryption for data in flight and the cloud tools of first rate from silicon all the way through AI tool chains, maximum database optionality, governance choices, identity access, availability of open source tools, et cetera. So one has to ask the likes of HPE and Dell with GreenLake and Apex, sort of on-prem incumbents, even though you're talking about having LLMs or in the case of GreenLake, LLMs as a service, how capable are they and how truly integrated are they into a seamless as a service offering? This is something to watch closely. While doing work on-prem can reduce risks and makes a lot of sense, perhaps on a cost basis as well, a lot of work needs to be done for incumbent firms to build out their offerings and full stack of ecosystem partners. As well, the cloud players, they have business momentum, despite all the talk of cloud optimization and repatriation and slowing growth, the numbers still dramatically favor the cloud players. Here is ETR data showing the net score breakdown for several aspiring LLM leaders. Net score is a measure of spending velocity. It measures the percent of customers that are new logos, that's the lime green. The forest green represents customers spending 6% or more relative to last year. The gray is flat spending, the pink is spending 6% less or worse, and the bright red is churn. You subtract the reds from the greens and you get net score as shown in the column to the right of the bars. And then to the right of that net score, we show a column with the number of responses, the ends in the survey, which is a proxy for market presence. So you can see AWS, Microsoft and Google have net scores of 51%, 49% and 34% respectively and ends near or in the case of Microsoft, well over a thousand. Compare this to Dell and HPE with net scores of 18% and 9% respectively. Now, Dell has a large market presence with an end of over 800 and HPEs got a respectable 483, but the cloud still has a meaningfully higher momentum from a spending standpoint and a large presence. Now, not to leave out some of the other players, chart two of two here, here's the same data for Databricks, Snowflake, IBM and Oracle, some of the key data players. Databricks with a very solid net score of 60% has taken over the top spot from Snowflake, which is at 47% net score. Although Snowflake has a larger market presence, bigger ends, but clearly Databricks is converging in. IBM and Oracle, as you can see, have much lower net scores of 10% and minus 1% respectively, both with large ends and the dataset, of course, Oracle as we know, focuses more on revenue than on number of customers. So using AWS as a proxy, this chart shows AWS's revenue growth going back to growth rates, going back to Q1 2022, this is our forecast. And we think the deceleration will moderate and we're looking for flat growth in Q3. And our current forecast calls for a re-acceleration in Q4 with a modest 14% growth due to AI and Gen AI as a tailwind. There are many risks to this scenario, not the least of which is the macro environment and the law of large numbers kicking in as well as competition, but our current thinking is we're at the tail end of cloud optimization, we're going to shift to new network enablement. All right, we're going to close by looking at how we see a kind of modified power law distribution of large language models. A power law distribution is a statistical relationship between two quantities. The simple way to think of a power law distribution is the 80-20 rule. For example, 80% of the sales come from 20% of the products in our portfolio. And on this chart, we're taking liberties with the concept of saying 80% of the large models will be really built by 20% of the companies and that long tail on the X axis will be very specific to industry and smaller models. So the number of points that we'd like to make here. First, we believe that enterprise tech innovation continues to be driven by consumer volumes, PC chips, data prowess from search and social media, flash storage and more. And recently tech gaming with Nvidia and driving AI, these all found their way into the enterprise via the consumer channel. The big cloud and consumer brands we believe will dominate the large model space and the sustained running of models, whereas inference will happen on-prem and at the edge, increasingly. What's different here from, for example, the web, which is that sort of orange curve, is the power law curve was like a wall straight down with no torso, the large internet giants dominated and yet a long tail. The LLM space we believe will be different pulled up to the right as shown by that red dotted area where open source and third party tools and the likes of Snowflake and Databricks will fill the gap along with cloud partners in addition to Snowflake and Databricks. There are others as well maybe the security business, et cetera. The on-prem incumbents like Dell, HPE and IBM will succeed to the extent that they're able to leverage LLMs, LLM diversity and build it into an ecosystem, deploy it in their go-to-market models in a manner that is as simple as the cloud and more cost effective. Importantly, we believe that enterprise AI will demand a clear ROI and economic value or it will die in the vine. And as we said earlier, that value will largely come from headcount reductions or limits to headcount increases. Enterprise AI will succeed if it can reduce headcount growth. Meanwhile, we believe that inference at the edge will be dominated by architectures built on low cost, low power silicon, high performance systems. Very often these are gonna be arm-based designs that have massive volume. Think Apple, think Tesla and we believe that the economics at the edge will eventually find their way into the enterprise and be a major disruptive force. Now it may take the better part of a decade but the economics of enterprise IT since the PC disrupted the mainframe have been driven by consumer volumes and we think this wave will be no different. AI plus data plus volume economics will determine the fundamental structure of the industry in the coming years. That's a bet we think is worth making and whatever industry you're in, applying it however will require careful thought and deep thinking, not AI washing. Okay, we'll leave it there. Thank you to Alex Meyerson who's on production and he manages the podcast. Ken Schiffman as well. Kristen Martin and Cheryl Knight that helped get the word out on social media and in our newsletters and Rob Hoef is our EIC over at siliconangle.com. He does some great editing. Thank you all. Remember all these podcasts are available. All these episodes are available as podcasts. Wherever you listen, just search Breaking Analysis Podcasts. We publish each week on wikiman.com and siliconangle.com. You want to reach me or pitch me david.volante at siliconangle.com or DM me at dvolante. You can comment on our LinkedIn post and please do check out etr.ai. They've got great survey data in the business. They're current, they're doing tons of stuff on gen AI. Check that out. This is Dave Vellante for theCUBE Insights Powered by ETR. Thanks for watching everybody and we'll see you next time on Breaking Analysis.