CIOs will need to navigate the nuance for enterprise AI vendors | Constellation Research

CIOs will need to navigate the nuance for enterprise AI vendors Published October 25, 2026 When it comes to your enterprise AI vendors you'll need a nuanced view and remember that two things can be true at once. The operative word of 2027 for enterprises may be nuance. We're entering a phase where enterprise AI isn't going to have a lot of black and white answers. It’s not frontier LLMs vs. open LLMs because the answer is you need both. On-premises AI vs. cloud? You’ll need both. There will be a lot of nuances that will apply to your vendors since we're entering the tale of two AI markets. I've never been more bearish and bullish at the same time. Bearish Larry says: This AI infrastructure boom is really a real estate bubble financed by debt. Toss in circular financing and the reality that the math doesn't work and the only debate is when this whole thing blows up spectacularly. Most of the AI bubble debate is about timing. Bullish Larry says: Enterprise AI is reaching its potential and reports in the field among IT buyers are positive. Enterprise AI projects have matured , CxOs have the deterministic vs. probabilistic dance down and techniques such as build over buy, on-premises and open models to control your destiny are popular . Two things can be true at once. The AI headlines could be dire and still be the best of times for enterprise AI. Keep this in mind as you scan the AI headlines and evaluate your vendors. Here are a few examples of how nuance matters. CoreWeave What a difference a year makes for CoreWeave. From the enterprise perspective, CoreWeave was merely a GPU landlord standardized on Nvidia's stack. Today, CoreWeave has put together a comprehensive platform called Forge. In many respects, CoreWeave is way more software and AI model management focused than not . At its recent Fully Connected conference, CoreWeave had some blue chip enterprises--led by companies like Caterpillar and Capital One --presenting. CoreWeave's bills are still paid by AI labs, AI startups and a few hyperscalers that'll offload workloads. But the enterprise roster is growing. The big question is whether the enterprise revenue can offset what's likely to be an AI infrastructure disaster ahead. CoreWeave is staying focused. At Nvidia GTC Berlin, CoreWeave launched the CoreWeave Zero Trust AI Framework so customers can control access to models, data and runtime environments. The company also announced CoreWeave Physical AI Launchables, which are tested configurations designed to scale physical AI and autonomous vehicles faster. These offerings are likely to be well received by the customer base. There is also a lot of noise. After all, CoreWeave is a real estate business. AI infrastructure is one big leveraged bet that's only getting more leveraged. Balance sheet concerns whacked all the neocloud providers and interest rates on CoreWeave debt spiked to 15% for a few tranches. Nearly every quarter CoreWeave raises capital. As a result, CoreWeave's latest 10-Q mentions debt 152 times and liabilities 122 times. CoreWeave's bet is that it can build out AI infrastructure, drive returns and work off debt. Some people will even call debt "growth capital" and that's why you should run away from them. Here's the reality. CoreWeave is a fine option for your AI workloads. Wall Street is freaked out by the balance sheet and neoclouds overall. Nevertheless, CoreWeave is among the top neoclouds. Two things can be true at once. Meta Enterprise Platform With great fanfare, Meta poached then MongoDB CEO CJ Desai and announced plans to turn its stack into products and services for enterprises . Today, Meta Enterprise Platform has just a few slides. But we know Meta CEO Mark Zuckerberg will happily throw money at something if he thinks the idea is a winner. In addition, Meta's new large language models and AI harness show a lot of potential for enterprises and developers. Meta even made up with Apple so it can chase the cool kids who want a supercomputer in the basement without paying obscene sums for tokens. At some point, Muse Agent, Meta Business Agent, Muse Code and Muse API could be running your business (at least in theory). All of that AI capex will allegedly turn Meta into a cloud services juggernaut. The storyline that Meta will be a huge enterprise vendor may stick with Wall Street analysts and traders, but could be delightfully comical to CxOs. Meta will need a decade to be a real enterprise option and it's unclear whether Zuckerberg has the attention span for it. Now if you're an enterprise buyer, you'll kick the tires on Meta Enterprise Platform and even SpaceX at some point because you'll need leverage against AWS, Google Cloud and Microsoft Azure. Will enterprises truly buy into Meta's enterprise ambitions? Not for a while. Meta has already had a few enterprise false starts. Anyone remember Meta Workplace and how it awesome it was for frontline workers? How about those metaverse meetings we have daily? During Meta's third quarter earnings call , CEO Mark Zuckerberg was optimistic about the company's higher margin enterprise efforts. "The main enterprise offering in the near term is going to be the model API and Muse Code. We have business agents that we've been scaling for a while, and each of those 3 things is growing quite quickly, so we're excited about that. Over time, we would like to make it so that we have a whole kind of business network of Muse agents as well," said Zuckerberg. "What we see is a lot of people are actually using Muse today to help run small businesses." Zuckerberg said the company also plans on "making a version of Muse that people can bring into different corporate environments and work there, we think is going to be a big deal." Meta CFO Susan Li also touted the enterprise ambitions. "AI is creating huge amounts of disruption in enterprise today,” said Li. “We have differentiated assets to serve this market, bringing together models, infrastructure, global distribution and long-standing relationships with businesses. And it's our hope that this will be able to be a new diversified recurring revenue stream for us." CxOs will look at Meta Enterprise Platform with a lot of skepticism. Traders will price in world domination. Two things can be true at once. Anthropic and OpenAI Now let's address the two enterprise vendors that serve as the elephants in the room for the AI infrastructure bubble. Both frontier AI labs have plenty of enterprise customers. Most companies have used Anthropic and/or OpenAI for coding and software development. Anthropic and OpenAI are also used by companies via cloud giants AWS, Google Cloud and Microsoft Azure, which are shareholders too. There was a time when announcing an enterprise AI deal with Anthropic and OpenAI was a cool thing to do. Tokenmaxxing changed that equation and enterprises instantly started looking at open models via Nvidia Nemotron. Security issues with OpenAI and Anthropic AI agents won't help enterprise perception either. Open models gain traction in enterprises, tech CEOs say There is no moat for LLM giants Nvidia’s Nemotron hit enterprise AI inflection point Apollo Chief Economist Torsten Slok said in a report : "There are still no signs of AI boosting productivity growth, see the third chart below. With no impact on the bottom line, mainstream corporate America is pulling back and starting to view token spending not as a source of growth but as a cost center that weighs on margins. VC-backed companies are spending more and more on tokens, but mainstream corporate America does not need the latest model and is unwilling to pay for it." Nevertheless, it's clear Anthropic and OpenAI will be in the enterprise stack somewhere. However, enterprises already smell lock-in though as well as a potential security headache. On the AI infrastructure side, the big question is whether Anthropic and OpenAI can meet their capital spending commitments. Those two AI labs are the epicenter of the AI infrastructure boom. Here's an interesting thought exercise. If Anthropic and OpenAI blew up and disappeared, it would blow up the AI infrastructure food chain. However, new models would emerge and enterprise AI's potential would still be realized. Even the compute would be used by other companies over time. It’s fairly obvious that Anthropic and OpenAI don’t have much of a moat. IBM CEO Arvind Krishna, SAP CEO Christian Klein and Meta CFO Susan Li all talked about using open models because they deliver performance for much less. “Would you like something that's 10% of the cost if it's 90% as good as the best?” asked Krishna. “I think what we are seeing is that at least larger enterprises are becoming a lot more cognizant and a lot more aware of what is the true ROI of all of their AI spend. They're also getting worried if some of the AI spending completely out of control with no real return,” said Krishna. Klein said SAP is routing models more . Klein said: "We are now checking every agent use case, we are shipping also on what is the best model and not only from a performance, from an outcome perspective, but also from a price perspective. We are switching the models now more often. And we see also that we have an increasing share of open-weight models.” We won't know how the Anthropic and OpenAI story turns out until they become public companies and report real numbers. What is clear that OpenAI and Anthropic benefited greatly from runaway AI budgets. Those days are over already. Two (maybe three or four) things will be true at once with both OpenAI and Anthropic. Copy link Link Copied Copied to clipboard Larry Dignan Editor in Chief of Constellation Insights Constellation Research Larry Dignan is Editor in Chief of Constellation Insights at Constellation Research, where he leads editorial coverage focused on enterprise technology, digital transformation, and emerging trends shaping the future of business. He oversees research-driven news, analysis, interviews, and event coverage designed to help technology buyers and vendors navigate complex markets with clarity and context. ... Read more Results Insights News October 23, 2026 How Google Public Sector’s RISE unit works: Show don’t tell Innovation & Product-led Growth Google Public Sector's Rapid Innovation Specialist Engineering (RISE) group is a bit of a vibe coding jazz band that aims to show customers what they can do via prototypes and the... Larry Dignan Insights News October 22, 2026 Open models gain traction in enterprises, tech CEOs say Data to Decisions Technology leaders from IBM, SAP and Meta all said that usage of open models internally and among customers is gaining momentum. The comments are just the latest indicator that fro... Larry Dignan Insights News October 21, 2026 Meta optimistic about Muse, Meta Enterprise Platform returns Next-Generation Customer Experience Meta said it will continue to invest heavily in AI infrastructure, but can't predict how efforts like Muse, Muse Wonder and Meta Enterprise Platform will juice growth beyond its co... Larry Dignan Insights News October 21, 2026 IBM delivers strong Q3, sees 4% to 5% revenue growth in 2026 Data to Decisions IBM delivered third quarter revenue growth of 5% as the company recovered from a rocky second quarter outing. ... Larry Dignan Insights News October 21, 2026 SAP posts strong Q3 on cloud growth Data to Decisions SAP posted third quarter revenue growth of 13%, cloud revenue growth of 26% and crossed the €10 billion quarterly revenue mark for the first time. ... Larry Dignan Insights News October 21, 2026 Microsoft launches Contact Center Intelligence in public preview Next-Generation Customer Experience Microsoft launched Microsoft Contact Center Intelligence in public preview in a move that keeps existing systems in place, but adds an agentic AI layer that can add context to cust... Larry Dignan