Three years have passed since ChatGPT’s launch by OpenAI ignited a wave of AI innovation and interest. Throughout this period, proponents consistently asserted AI’s vital role in enterprise software, leading to a proliferation of enterprise AI startups fueled by substantial investment.
However, businesses continue to find it challenging to realize the advantages of these nascent AI solutions. An August MIT survey indicated that 95% of companies were not achieving significant returns from their AI investments.
When can businesses anticipate genuine benefits from AI implementation and integration? TechCrunch questioned 24 VCs specializing in enterprise, who largely believe 2026 will mark the year when enterprises truly embrace AI, recognize its worth, and allocate more funds towards the technology.
Enterprise VCs have been making this prediction for the past three years. Will 2026 truly bring a different outcome?
Let’s explore their perspectives:
Which enterprise trends do you foresee gaining traction in 2026?
Kirby Winfield, founding general partner, Ascend: Enterprises are increasingly understanding that LLMs aren’t universal solutions for most challenges. The fact that a company like Starbucks could leverage Claude to develop custom CRM software doesn’t imply it’s the optimal approach. Our attention will shift to custom models, fine-tuning, evaluations, observability, orchestration, and data sovereignty.
Molly Alter, partner, Northzone: Some enterprise AI firms will transition from offering products to providing AI consulting services. Initially, these businesses might launch with a focused product, like AI-driven customer support or coding agents. However, after accumulating sufficient customer workflows on their platform, they can adopt a forward-deployed engineer model with their teams to develop more customer-specific use cases. Essentially, numerous specialized AI product companies will evolve into broader AI implementation providers.
Marcie Vu, partner, Greycroft: We are extremely enthusiastic about the potential of voice AI. Voice offers a significantly more intuitive, effective, and expressive means for human-to-human and human-to-machine communication. While we’ve spent decades interacting via keyboards and screens, speech is our natural mode of engagement. I keenly await how innovators will rethink products, experiences, and interfaces, prioritizing voice as the main method for interacting with intelligence.
Alexa von Tobel, founder and managing partner, Inspired Capital: In 2026, AI is expected to transform the physical realm, particularly in areas like infrastructure, manufacturing, and climate monitoring. Our progression is from a reactive state to a predictive one, enabling physical systems to detect issues before they escalate into failures.
Lonne Jaffe, managing director, Insight Partners: We are observing how advanced research labs are tackling the application layer. Many anticipated that these labs would simply train models for others to build upon, but their current strategy appears different. It’s possible that frontier labs will deploy more ready-to-use applications directly into production across sectors such as finance, law, healthcare, and education, more than commonly expected.
Tom Henriksson, general partner at OpenOcean: If I were to describe quantum computing in 2026 with a single word, it would be ‘momentum’. Confidence in quantum advantage is rapidly growing, with companies releasing roadmaps to clarify the technology. However, significant software innovations shouldn’t be anticipated just yet, as greater hardware performance is still needed to reach that benchmark.
In what sectors are you seeking investment opportunities?
Emily Zhao, principal, Salesforce Ventures: Our focus is on two separate domains: the integration of AI into the physical environment and the subsequent advancement of model research.
Michael Stewart, managing partner, M12: We are keen on future data center technology. Over the past year, we’ve initiated several new investments indicating our interest in forthcoming ‘token factory’ technologies, aiming to enhance their operational efficiency and cleanliness. This focus will persist into 2026 and beyond, covering all aspects within data centers: cooling, computation, memory, and internal and inter-site networking.
Jonathan Lehr, co-founder and general partner, Work-Bench: Specialized enterprise software where unique workflows and data establish a competitive advantage, especially within regulated sectors, supply chain, retail, and other intricate operational settings.
Aaron Jacobson, partner, NEA: Humanity is approaching the limits of energy generation needed to power energy-intensive GPUs. As an investor, I seek software and hardware capable of achieving significant advancements in performance per watt. This includes improved GPU management, more energy-efficient AI chips, advanced networking solutions such as optical technology, or innovative approaches to thermal management in AI systems and data centers.
Regarding AI startups, what criteria do you use to identify a company’s competitive advantage?
Rob Biederman, managing partner, Asymmetric Capital Partners: An AI company’s competitive edge is less about the model itself and more about its economic factors and integration. We seek companies deeply integrated into enterprise operations, possessing access to proprietary or continually refined data, and exhibiting resilience through high switching costs, cost efficiencies, or unique, hard-to-imitate results.
Jake Flomenberg, partner, Wing Venture Capital: I am wary of competitive advantages solely based on model performance or prompting, as these benefits quickly diminish. My key question is: if OpenAI or Anthropic were to release a model that is ten times superior tomorrow, would this company still maintain its relevance?
Molly Alter, partner, Northzone: Currently, establishing a competitive advantage is considerably simpler within a vertical category compared to a horizontal one. The most effective advantages are data-driven, where every additional customer, data input, or interaction enhances the product. Such advantages are relatively easier to cultivate in niche sectors like manufacturing, construction, healthcare, or legal, where data uniformity among customers is higher. Additionally, compelling “workflow moats” exist, deriving their strength from a deep comprehension of how tasks or projects progress within a specific industry.
Harsha Kapre, director, Snowflake Ventures: The most robust competitive edge for AI startups stems from their proficiency in converting an enterprise’s current data into superior decisions, processes, and customer interactions. While businesses possess vast amounts of data, they often lack the capability to analyze it in a focused and reliable manner. We seek startups that combine technical prowess with profound industry insight, capable of applying domain-specific solutions directly to a client’s governed data, thereby avoiding new data silos, and delivering unprecedented insights or automation.
Do you anticipate enterprises will begin realizing value from AI investments in 2026?
Kirby Winfield, founding general partner, Ascend: Enterprises are recognizing that indiscriminate experimentation with numerous solutions leads to disarray. Their focus will shift towards fewer solutions, accompanied by more deliberate engagement.
Antonia Dean, partner, Black Operator Ventures: The intricacy lies in the fact that many enterprises, irrespective of their preparedness to effectively deploy AI solutions, will assert an increase in AI investments as a rationale for reducing expenditures elsewhere or cutting staff. In truth, AI may serve as a convenient excuse for executives seeking to conceal previous errors.
Scott Beechuk, partner, Norwest Venture Partners: We are certainly approaching that point. While the previous year focused on establishing AI infrastructure, 2026 will reveal if the application layer can convert that investment into tangible value. With the maturation of specialized models and enhanced oversight, AI systems are proving increasingly dependable in routine operations.
Marell Evans, founder and managing partner, Exceptional Capital: Yes, though the progress will remain gradual. AI is still undergoing significant refinement and is continuously improving to the extent of demonstrating solutions for enterprise pain points across diverse sectors. I anticipate that resolving the challenge of simulation-to-reality training will unlock numerous prospects for various industries, both established and emerging.
Jennifer Li, general partner, Andreessen Horowitz: This year, there have been striking headlines about businesses failing to see returns on their AI investments. Yet, inquire of any software engineer if they would revert to the pre-AI coding tool era. The answer is likely no. My contention is that enterprises are already benefiting this year, and these advantages will significantly expand across organizations next year.
Will enterprises expand their AI budgets in 2026?
Rajeev Dham, managing director, Sapphire: Yes, I anticipate they will, though the situation is complex. Instead of merely raising AI budgets, organizations will redirect parts of their labor spending towards AI technologies, or achieve such substantial top-line ROI from AI functionalities that the investment effectively yields a three to five-fold return.
Rob Biederman, managing partner, Asymmetric Capital Partners: Budgets will expand for a select group of AI products that demonstrably provide results, while sharply decreasing for all other solutions. While total expenditure might rise, it will be considerably more focused. We foresee a division, where a limited number of providers will secure a disproportionate share of enterprise AI funding, concurrently with many others experiencing stagnant or diminishing revenues.
Gordon Ritter, founder and general partner, Emergence Capital: Yes, but spending will become more focused. Enterprises will boost budgets for AI solutions that amplify existing institutional strengths, and reduce investment in tools that merely automate processes without capturing (and protecting!) proprietary intelligence.
Andrew Ferguson, vice president, Databricks Ventures: In 2026, CIOs will likely resist the proliferation of AI vendors. Currently, businesses often pilot numerous tools for a single purpose – with low monthly costs and switching hurdles, experimentation is encouraged. There’s also a surge of startups targeting specific buying centers, such as go-to-market, making differentiation incredibly challenging even during proofs of concept. As enterprises observe concrete benefits from AI, they will reduce experimental spending, consolidate redundant tools, and reinvest those savings into proven AI technologies.
Ryan Isono, managing director, Maverick Ventures: Overall, yes, and we’ll see a reallocation from pilot/experimental funds to established budget line items. A significant advantage for AI startups in 2026 will be the shift from enterprises that initially attempted in-house solutions but have since recognized the challenges and complexities of scaling them in production.
What are the requirements for an enterprise AI startup to secure Series A funding in 2026?
Jake Flomenberg, partner, Wing Venture Capital: Currently, the most successful companies integrate two key elements: a persuasive “why now” argument—typically linked to generative AI creating novel vulnerabilities, infrastructure demands, or workflow possibilities—and demonstrable evidence of enterprise acceptance. While $1 million to $2 million in annual recurring revenue serves as a minimum, more crucial is whether clients consider your product indispensable to their operations, rather than merely a desirable addition. Revenue without a compelling story is merely a feature; a story without real-world uptake is insubstantial. Both are essential.
Lonne Jaffe, managing director, Insight Partners: Your objective should be to demonstrate development within a market where the total addressable market grows, rather than diminishes, as AI reduces costs. Certain markets exhibit high demand elasticity—a 90% price drop can result in a tenfold market expansion. Conversely, others have low elasticity, where price reductions might cause the market to disappear, with customers retaining all generated value.
Jonathan Lehr, co-founder and general partner, Work-Bench: Customers are utilizing the product in their daily operations and are prepared to serve as references, openly discussing its impact, dependability, and the purchasing process. Companies must clearly illustrate how the product delivers time savings, cost reductions, or increased output in a manner that withstands security, legal, and procurement scrutiny.
Michael Stewart, managing partner, M12: Until recently, investors viewed estimated annual recurring revenue or pilot revenue with skepticism. Now, it’s less of a secondary consideration and more indicative of a customer’s genuine interest and readiness to assess a solution amidst numerous choices. Securing these engagements and customer commitment for evaluations isn’t merely about having forward-deployed engineers simplify things; it demands a quality product and a compelling marketing strategy in 2026. Investors anticipate conversions becoming the primary success indicator following six months of pilot deployment.
Marell Evans, founder and managing partner, Exceptional Capital: Execution and market traction are crucial. The strongest indicator is truly satisfied users and the business’s technical sophistication. We prioritize significant contractual agreements lasting 12 months or more. Furthermore, we assess whether the founder successfully attracted elite talent to their startup, choosing it over competitors or established hyperscalers?
By late 2026, what function will AI agents serve within enterprises?
Nnamdi Okike, managing partner and co-founder, 645 Ventures: By the close of 2026, agents will likely remain in their early adoption stages. Numerous technical and regulatory obstacles must be addressed before enterprises can fully leverage AI agents. Furthermore, standards for agent-to-agent communication are necessary.
Rajeev Dham, managing director, Sapphire: A singular, comprehensive agent will materialize. Presently, each agent operates within a distinct silo—such as inbound sales development representatives, outbound SDRs, customer support, or product discovery. However, by late next year, we anticipate these functions consolidating into one agent possessing shared context and memory, thereby dismantling traditional organizational barriers and facilitating a more cohesive, context-aware dialogue between businesses and their clientele.
Antonia Dean, partner, Black Operator Ventures: Successful organizations will be those that swiftly establish the optimal equilibrium between autonomy and supervision, acknowledging agent deployment as collaborative enhancement rather than a rigid division of tasks. Instead of agents managing all mundane duties while humans perform all cognitive functions, we will observe more intricate partnerships between humans and agents on complex assignments, with their role boundaries progressively shifting.
Aaron Jacobson, partner, NEA: Most knowledge workers will have at least one intelligent agent as a named colleague!
Eric Bahn, co-founder, general partner, Hustle Fund: I believe AI agents will likely constitute a larger segment of the enterprise workforce than human employees. The propagation of AI agents is virtually cost-free and entails zero marginal expense. Therefore, why not pursue growth through automated bots?
Which types of companies within your portfolio are experiencing the most robust growth?
Jake Flomenberg, partner, Wing Venture Capital: The companies exhibiting the fastest growth are those that pinpointed a workflow or security vulnerability emerging from GenAI adoption, then rigorously pursued product-market fit. In cybersecurity, this includes tools for securing data, enabling LLMs to safely handle sensitive information, and agent governance to ensure autonomous systems have proper oversight. In marketing, new domains like Answer Engine Optimization (AEO)—achieving visibility in AI responses rather than just search results—are key. The recurring theme is that these categories, non-existent two years ago, are now essential for enterprises implementing AI broadly.
Andrew Ferguson, vice president, Databricks Ventures: We observe growth connected to several consistent patterns. One involves companies that establish themselves with specific use cases—firms commencing with a concentrated niche (perhaps a particular target persona or application), excelling in it, achieving strong customer loyalty, and thereby gaining the opportunity to broaden their scope from that initial foundation.
Jennifer Li, general partner, Andreessen Horowitz: Companies assisting enterprises in deploying AI into production are performing strongly. This includes sectors such as data extraction and organization, enhancing developer productivity for AI systems, providing infrastructure for generative media, and integrating voice and audio for media and applications like customer support or call centers.
Which categories of companies are experiencing the highest customer retention?
Jake Flomenberg, partner, Wing Venture Capital: Companies demonstrating strong retention and expansion exhibit a common trait: they address issues that become more pronounced as clients integrate more AI. Robust retention is driven by three factors: their mission-critical nature (removing them disrupts production workflows), the accumulation of unique, hard-to-replicate proprietary context, and solving problems that evolve with AI adoption instead of being single, resolved issues.
Tom Henriksson, general partner at OpenOcean: Measuring retention is more challenging for newer companies, but we observe the strongest retention among established enterprise software providers, particularly those augmented with AI. A prime illustration is Operations1, which fully digitizes employee-driven production processes. These firms deeply integrate into a customer’s operations, revolutionize their functioning, and amass proprietary data and expertise, making them exceptionally difficult to replace.
Michael Stewart, managing partner, M12: Startups catering to enterprises with data tools and specialized AI applications, supported by dedicated teams that enhance customer satisfaction, quality, and product refinement. This appears to be the successful approach embraced by all prominent startups in these sectors. In the long run, these embedded teams might become less prominent as customers increasingly integrate AI into their organizational structures and daily operations.
Jonathan Lehr, co-founder and general partner, Work-Bench: Retention peaks when software functions as fundamental infrastructure rather than a standalone solution. Authzed demonstrates robust retention because its authorization and policy capabilities are central to contemporary systems, becoming prohibitively expensive to remove once integrated. Courier Health and GovWell serve as primary record-keeping and orchestration platforms for comprehensive workflows, such as patient care pathways in healthcare and permitting processes in government, ensuring deep integration upon deployment.