Three years have passed since OpenAI introduced ChatGPT, sparking a wave of innovation and heightened focus on artificial intelligence. Since that time, optimists have frequently argued that AI will become essential to enterprise software, leading to a rapid rise in enterprise AI startups fueled by substantial investment. Yet many businesses continue to find it challenging to recognize tangible benefits from adopting these new AI tools. An MIT survey from August revealed that 95% of enterprises were not achieving a meaningful return on their AI investments. This raises the question: when will companies begin to see real value from using and integrating AI? Venture capitalists focused on the enterprise have been asking this for three years. Could 2026 finally mark a turning point? Here is what they have to say.

**What enterprise-related trends do you expect to take off in 2026?**

Kirby Winfield, founding general partner, Ascend: Companies are recognizing that large language models are not a universal solution for most challenges. Just because a company like Starbucks could use Claude to develop its own CRM software doesn't mean it should. Attention will shift toward custom models, fine-tuning, evaluation, observability, orchestration, and data sovereignty.

Molly Alter, partner, Northzone: A segment of enterprise AI companies will transition from product-based businesses to AI consulting. These firms might begin with a dedicated product, such as AI-driven customer support or coding agents. However, once they have sufficient customer workflows operating on their platform, they can adopt a forward-deployed engineering model, using their own teams to build additional use cases for clients. Essentially, numerous specialized AI product companies will evolve into generalist AI implementers.

Marcie Vu, partner, Greycroft: We are particularly enthusiastic about the potential in voice AI. Voice offers a far more natural, efficient, and expressive method for people to communicate with each other and with machines. While we have spent decades typing on computers and looking at screens, speech is how we interact in the real world. I look forward to seeing how creators reinvent products, experiences, and interfaces using voice as the primary mode of engaging with intelligence.

Alexa von Tobel, founder and managing partner, Inspired Capital: 2026 will be the year AI transforms the physical world—particularly in infrastructure, manufacturing, and climate monitoring. We are shifting from a reactive to a predictive world, where physical systems can detect issues before they lead to failures.

Lonne Jaffe, managing director, Insight Partners: We are observing how frontier AI labs approach the application layer. Many assumed these labs would simply train models and allow others to build on them, but that does not appear to be their strategy. We may see frontier labs deploying more ready-to-use applications directly into production in fields such as finance, law, healthcare, and education than anticipated.

Tom Henriksson, general partner at OpenOcean: If I had to choose one word for quantum computing in 2026, it would be momentum. Confidence in quantum advantage is growing rapidly, with companies publishing roadmaps to clarify the technology. However, do not expect major software breakthroughs just yet; we still require greater hardware performance to reach that threshold.

**Which areas are you looking to invest in?**

Emily Zhao, principal, Salesforce Ventures: We are focusing on two distinct frontiers: AI entering the physical world and the next evolution of model research.

Michael Stewart, managing partner, M12: Future data center technology. Over the past year or so, we have made several new investments that reflect our interest in future "token factory" technology, with a focus on what can genuinely improve efficiency and sustainability. This trend will continue in 2026 and beyond, encompassing everything within data center walls: cooling, compute, memory, and networking within and between sites.

Jonathan Lehr, co-founder and general partner, Work-Bench: Vertical enterprise software where proprietary workflows and data create defensibility, especially in regulated industries, supply chain, retail, and other complex operational environments.

Aaron Jacobson, partner, NEA: We are reaching the limit of humanity's capacity to generate sufficient energy for power-hungry GPUs. As an investor, I am seeking software and hardware that can drive breakthroughs in performance per watt. This could involve better GPU management, more efficient AI chips, next-generation networking approaches like optical technology, or rethinking thermal management within AI systems and data centers.

**When it comes to AI startups, how do you determine that a company has a moat?**

Rob Biederman, managing partner, Asymmetric Capital Partners: A moat in AI is less about the model itself and more about economics and integration. We look for companies that are deeply embedded in enterprise workflows, have access to proprietary or continuously improving data, and demonstrate defensibility through switching costs, cost advantages, or outcomes that are difficult to replicate.

Jake Flomenberg, partner, Wing Venture Capital: I am skeptical of moats based purely on model performance or prompting—those advantages can erode within months. The question I ask is: if OpenAI or Anthropic releases a model tomorrow that is ten times better, does this company still have a reason to exist?

Molly Alter, partner, Northzone: It is much easier today to build a moat in a vertical category rather than a horizontal one. The strongest moats are data moats, where each additional customer, data point, or interaction improves the product. These are somewhat easier to establish in specialized fields like manufacturing, construction, health, or legal, where data is more consistent across customers. There are also interesting "workflow moats," where defensibility comes from understanding how a task or project moves from start to finish within an industry.

Harsha Kapre, director, Snowflake Ventures: For AI startups, the strongest moat comes from how effectively they transform an enterprise's existing data into better decisions, workflows, and customer experiences. Enterprises already possess incredibly rich data; what they lack is the ability to analyze it in a targeted, trustworthy manner. We look for startups that combine technical expertise with deep industry knowledge and can deliver domain-specific solutions directly to a customer's governed data—without creating new silos—to provide insights or automation that were previously unattainable.

**Will 2026 be the year when enterprises start to gain value from AI investments?**

Kirby Winfield, founding general partner, Ascend: Enterprises are realizing that random experiments with dozens of solutions create chaos. They will concentrate on fewer solutions with more thoughtful engagement.

Antonia Dean, partner, Black Operator Ventures: The complexity is that many enterprises, regardless of their readiness to successfully use AI solutions, will claim they are increasing AI investments to justify cutting spending in other areas or reducing workforces. In reality, AI will become a scapegoat for executives seeking to cover past mistakes.

Scott Beechuk, partner, Norwest Venture Partners: We are definitely getting closer. If last year was about laying the AI infrastructure, 2026 is when we start to see whether the application layer can turn that investment into real value. As specialized models mature and oversight improves, AI systems are becoming more reliable in daily workflows.

Marell Evans, founder and managing partner, Exceptional Capital: Yes, but progress will still be incremental. There remains considerable iteration, and AI is still improving to the point of demonstrating solutions for enterprise pain points across various industries. I believe solving simulation-to-reality training will likely create many opportunities for a range of industries, both established and emerging.

Jennifer Li, general partner, Andreessen Horowitz: There have been sensational headlines this year about enterprises not seeing returns on their AI investments. Ask any software engineer if they would want to return to the era before AI coding tools. Unlikely. My point is that enterprises are already gaining value this year, and it will multiply across organizations next year.

**Do you think enterprises will increase their AI budgets in 2026?**

Rajeev Dham, managing director, Sapphire: Yes, I believe they will, though it's nuanced. Rather than simply increasing AI budgets, organizations will shift portions of their labor spending toward AI technologies or generate such strong top-line ROI from AI capabilities that the investment effectively pays for itself three to five times over.

Rob Biederman, managing partner, Asymmetric Capital Partners: Budgets will increase for a narrow set of AI products that clearly deliver results and will decline sharply for everything else. Overall spending may grow, but it will be significantly more concentrated. We anticipate a bifurcation, where a small number of vendors capture a disproportionate share of enterprise AI budgets while many others see revenue flatten or contract.

Gordon Ritter, founder and general partner, Emergence Capital: Yes, but spending will concentrate. Enterprises will increase budgets where AI enhances institutional advantages and pull back from tools that merely automate workflows without capturing (and securing) proprietary intelligence.

Andrew Ferguson, vice president, Databricks Ventures: 2026 will be the year that CIOs push back on AI vendor sprawl. Currently, enterprises are testing multiple tools for a single use case—monthly spending and switching costs are often low, encouraging experimentation—and there is an explosion of startups focused on certain buying centers like go-to-market, where differentiation is extremely hard to discern even during proof of concepts. As enterprises see real proof points from AI, they will cut some experimentation budgets, rationalize overlapping tools, and redirect those savings into AI technologies that have delivered results.

Ryan Isono, managing director, Maverick Ventures: In aggregate, yes, and there will be some shifting from pilot and experimental budgets to budgeted line items. A boon for AI startups in 2026 will be enterprises that attempted to build in-house solutions and have now recognized the difficulty and complexity of production at scale.

**What does it take to raise a Series A as an enterprise-focused AI startup in 2026?**

Jake Flomenberg, partner, Wing Venture Capital: The strongest companies currently combine two elements: a compelling "why now" narrative—typically linked to generative AI creating new attack surfaces, infrastructure needs, or workflow opportunities—and concrete proof of enterprise adoption. $1 million to $2 million in annual recurring revenue is the baseline, but what matters more is whether customers view your product as mission-critical to their business rather than just a nice-to-have. Revenue without a narrative is merely a feature; a narrative without traction is vaporware. You need both.

Lonne Jaffe, managing director, Insight Partners: You should aim to demonstrate that you are building in a space where the total addressable market expands rather than evaporates as AI reduces costs. Some markets have high elasticity of demand—a 90% price decline leads to a tenfold increase in market size. Others have low elasticity, where dropping the price can vaporize the market, allowing customers to retain all the value being created.

Jonathan Lehr, co-founder and general partner, Work-Bench: Customers are using the product in real, day-to-day operations and are willing to take reference calls and discuss impact, reliability, and the buying process honestly. Companies should be able to clearly show how the product saves time, reduces costs, or increases output in a way that holds up through security, legal, and procurement reviews.

Michael Stewart, managing partner, M12: Until recently, investors were skeptical of estimated annual recurring revenue or pilot revenue. Now, it is not seen as much of an asterisk; rather, the customer's interest and willingness to evaluate a solution amid many options is key. Securing those engagements and customer buy-in for running an evaluation is not just about forward-deployed engineers making it easier for the customer. It requires quality and a winning marketing message in 2026. Investors expect to see conversions become the leading part of the story after six months of pilot use.

Marell Evans, founder and managing partner, Exceptional Capital: Execution and traction. The strongest signal is users who are genuinely delighted with the product and the technical sophistication of the business. We look for significant North Star metrics like real contractual agreements lasting twelve months or more. Additionally, was the founder able to attract top-tier talent to join their startup over competitors or traditional hyperscalers?

**What role will AI agents play at enterprises by the end of 2026?**

Nnamdi Okike, managing partner and co-founder, 645 Ventures: Agents will still be in their initial adoption phase by the end of 2026. Numerous technical and compliance hurdles must be overcome for enterprises to truly benefit from AI agents. Standards also need to be established for agent-to-agent communication.

Rajeev Dham, managing director, Sapphire: One universal agent will emerge. Currently, each agent is siloed in its role—for example, inbound sales development representative, outbound SDR, customer support, product discovery, etc. But by late next year, we will start to see these roles converge into a single agent with shared context and memory, breaking down long-standing organizational silos and enabling a more unified, contextual conversation between companies and their users.

Antonia Dean, partner, Black Operator Ventures: The winners will be organizations that quickly determine the right balance of autonomy and oversight and recognize agent deployment as collaborative augmentation rather than a clean division of labor. Instead of agents handling all routine work while humans do all the thinking, we will see more sophisticated collaboration between humans and agents on complex tasks, with the boundary between their roles continuously evolving.

Aaron Jacobson, partner, NEA: The majority of knowledge workers will have at least one agentic co-worker they know by name.

Eric Bahn, co-founder, general partner, Hustle Fund: I believe that AI agents will likely constitute a larger portion of the workforce than any humans in enterprises. Proliferating AI agents is essentially free and has zero marginal cost. So why not grow through bots?

**What kinds of companies in your portfolio are seeing the strongest growth?**

Jake Flomenberg, partner, Wing Venture Capital: The fastest-growing companies are those that identified a workflow or security gap created by generative AI adoption and then executed relentlessly on product-market fit. In cybersecurity, these are tools addressing data security so LLMs can interact with sensitive data safely, and agent governance ensuring autonomous systems have appropriate controls. In marketing, it's new areas like Answer Engine Optimization—getting discovered in AI responses, not just search results. The common thread is that these categories did not exist two years ago but are now essential for enterprises deploying AI at scale.

Andrew Ferguson, vice president, Databricks Ventures: We are seeing growth tied to a few common themes. One is companies that land with focused use cases—firms that start with a narrower wedge (whether a specific target persona or use case), truly excel at it, become sticky, and earn the right to expand from that initial wedge.

Jennifer Li, general partner, Andreessen Horowitz: Companies that help enterprises