Accel’s Botteri: How Startups Can Compete Against Behemoths

Philippe Botteri, a partner at Accel, along with his firm’s comprehensive 2025 Globalscape report, meticulously charted the new AI power map, revealing how a select group of "Super Six" companies – Nvidia, Microsoft, Apple, Alphabet, Amazon, and Meta – are leveraging their hundreds of billions in operating cash flow to aggressively invest in AI infrastructure. This concentration of capital creates a formidable competitive landscape. Accel, notably one of the three most active investors on The Crunchbase Unicorn Board, which itself surged in value amid the 2025 AI boom, has strategically backed a diverse wave of native AI startups spanning both the foundational model and application layers. Their portfolio includes Anthropic, a leading developer of advanced AI models; H Co., a specialist in small model development; Nebius Group, a publicly traded AI infrastructure provider; and application-layer innovators like Anysphere (creator of Cursor), Perplexity, Synthesia, and cybersecurity startup Cyera. Botteri’s analysis underscores that while incumbents are indeed capturing enormous market share, there remains ample room for agile, focused, and fast-growing AI-native companies to carve out new categories or radically reinvent existing ones. These startups succeed by identifying specific pain points and delivering order-of-magnitude improvements. Speaking at Web Summit in Lisbon, Botteri passionately articulated the broader economic impact, stating, "If you don’t think that GenAI is going to generate a 1%-2% increase in the global GDP, then I’m not sure why we’re doing all this." His conviction highlights the firm’s belief in AI’s profound and pervasive economic transformation.

Where Foundation Capital Sees Opportunities in Physical Tech

The staggering acceleration of AI and its insatiable demand for computing power have precipitated an existential question for the industry: Can physical infrastructure keep pace? This year, many investors we engaged with consistently pointed to AI’s increasingly physical bottlenecks – namely power generation, specialized chips, and data centers – as the fertile ground for some of the most compelling startup opportunities. Botteri’s analysis, for instance, projects an estimated 117-gigawatt energy shortfall over the next five years to meet anticipated AI demand, an amount roughly equivalent to powering three large European economies combined. Steve Vassallo, general partner at Foundation Capital, approaches this critical problem from a venture perspective. His firm exhibited remarkable foresight by incubating AI chipmaker Cerebras Systems back in 2016, long before the current widespread enthusiasm for AI infrastructure. At the time, betting on semiconductors was generally considered a risky proposition, "a recipe for losing a lot of money," as Vassallo recalled in our interview. However, Foundation Capital’s team believed that AI workloads were growing at such an exponential rate that traditional chip architectures would eventually hit a wall, necessitating specialized hardware. This thesis has proven prescient, with Cerebras signaling plans for an IPO and Nvidia’s market capitalization soaring past $4 trillion.

Vassallo further argued that the most impactful companies in this AI cycle will be those that not only harness AI’s power but also deeply understand how humans can be "hacked" – essentially, developing products that honor human psychology as much as the laws of physics. He particularly highlighted the importance of reinforcement learning with human feedback, where human operators actively participate in refining AI behavior, thereby becoming more adept collaborators with the systems they help train. Foundation Capital’s AI investments reflect this philosophy, including seed rounds in Tennr, a company automating complex and often paper-based healthcare authorization workflows, and Jasper, an application that assists with writing, built on OpenAI’s GPT-3. They also led the Series A for PlayerZero, a product that predicts and debugs software failures in AI-written code pre-deployment, embodying the symbiotic relationship between AI and human oversight. "We love working with founders who are living right at that edge," Vassallo affirmed, emphasizing the firm’s appetite for innovation at the intersection of technology and human interaction.

Why Dell Technologies Capital Invests at the Silicon Level

At Dell Technologies Capital (DTC), managing director Daniel Docter and partner Elana Lian operate at the crucial nexus of infrastructure and enterprise demand. Dell anticipates an impressive $20 billion in AI server shipments by fiscal 2026, and its venture arm has demonstrated remarkable agility, logging six exits since June – one IPO and five acquisitions – in a venture landscape where exits have otherwise been scarce. As we explored in our discussion, Dell’s prominent position as a leading GPU server provider grants its venture arm unparalleled visibility into nearly every serious enterprise AI buyer and builder. This unique vantage point allows DTC to identify foundational shifts and emerging needs. Like other investors we interviewed, Docter and Lian noted the unprecedented pace of investment in hot AI startups. "We’ll meet with a company on a Tuesday for the first time and sometimes by Thursday, they have a term sheet that they’ve already signed," Docter revealed, illustrating the intense competitive environment.

DTC’s infrastructure-level investments underscore their belief in foundational disruption. This includes AI chipmaker Rivos, which Meta plans to acquire for an undisclosed sum (pending regulatory approval); SiMa.ai, a company developing chips for embedded edge use cases in areas like automobiles, drones, and robotics; and Runpod, an AI developer software layer offering on-demand access to GPUs. Docter explained their focus on the silicon level, stating that such investments "can be incredibly disruptive to the ecosystem," laying the groundwork for future innovation. At the application level, DTC’s portfolio features Maven AGI, which provides sophisticated customer support for complex, high-compliance enterprise scenarios, and Series Entertainment, a GenAI platform designed to drastically shorten game development timelines. Elana Lian succinctly encapsulated a pervasive challenge in the sector, stating, "AI is almost a data problem." She argued that for models to continuously improve and achieve true differentiation, they require access to high-quality, domain-specific data, rather than merely an increase in parameters. This insight highlights the critical role of data as a strategic asset in the AI era.

Sierra Ventures’ Layer-Cake Approach

If compute power represents a critical bottleneck, then data unequivocally stands as the ultimate differentiator in the AI landscape. Elana Lian’s observation that "AI is almost a data problem" resonates deeply with Tim Guleri, managing partner at Sierra Ventures, who is also intensely focused on the strategic importance of data. In our interview, Guleri articulated his firm’s consistent pattern for identifying promising startups: they target significant, painful workflows, promise order-of-magnitude productivity improvements, and crucially, are built upon rich, proprietary datasets. Sierra Ventures employs a sophisticated "layered cake" framework to dissect AI investing into five distinct levels: the foundational infrastructure layer; applied infrastructure built atop foundational models; horizontal applications that serve broad business functions; vertical applications tailored to specific industries; and entirely novel innovations that could only exist thanks to AI.

Guleri clarified that Sierra Ventures deliberately avoids competing in the most capital-intensive infrastructure layer. Instead, the firm strategically leans into applied infrastructure and various application layers where proprietary data and clever distribution strategies can establish robust, durable moats. He highlighted the immense economic opportunity, noting that global GDP is approximately $110 trillion, with only about $6 trillion in agriculture. This leaves over $100 trillion in services and other industries where Guleri anticipates substantial AI-driven efficiency gains to accrue. "AI is the wave that’s lifting everything on top of it," he declared, emphasizing the pervasive and transformative power of the technology. "There’s going to be a tremendous amount of value creation in the coming decades." Sierra Ventures’ approach underscores the belief that while the foundational layers are essential, true enduring value often resides in the intelligent application of AI, leveraging unique data to solve real-world problems across a vast economic spectrum.

How a Google Brain Co-founder Builds and Backs AI Startups

Andrew Ng, a co-founder of both Google Brain and Coursera, has adopted a uniquely hands-on approach to cultivating AI innovation through his venture studio, AI Fund, launched in 2018. Ng’s model emphasizes generating proprietary data and deep industry insights through strategic corporate partnerships. With corporate limited partners including AES, HP, Mitsui & Co., and Mitsubishi, AI Fund gains unparalleled access to highly specialized sectors—such as renewable energy, large-scale industrial operations, and insurance—where internal data is not only difficult to access but also absolutely critical for building defensible AI products. Ng explained that a significant fraction of AI Fund’s startup ideas originate directly from these corporate partners, who pinpoint market gaps within massive but often under-digitized economic sectors. "It turns out a meaningful fraction of our startup ideas come from corporate partners that have spotted a market need, often in some sector of the economy, which is very large, very important but completely foreign to the typical consumer, or completely foreign to the typical AI engineer," Ng elaborated in our interview. "I find that it’s been interesting how often we get to play in these spaces. We think it’s wildly exciting, while no one else cares."

While traditional venture firms like Sequoia Capital and New Enterprise Associates also invest in AI Fund, Ng asserted that his fund’s strategy fundamentally diverges from the conventional venture capital approach. "Unlike a traditional VC, our primary business activity is not to compete for deal flow," he stated. "Our primary business activity is to identify promising startup ideas, validate the market need and the customer need. Then we recruit a CEO to work alongside us to build a company." This venture studio model allows AI Fund to proactively create companies tailored to specific, validated opportunities, rather than reactively investing in existing startups. Ng remains optimistic about future opportunities, particularly in specific verticals like visual and voice AI. He views AI not as a monolithic entity but as a multifaceted technology, asserting, "It feels like AI is not one thing; it is many different things that are creating new opportunities." His strategy is a testament to the power of targeted, data-rich innovation in overlooked but high-value markets.

GV on Being Willing to Invest at AI’s Premium Valuations

GV, formerly Google Ventures, has quietly emerged as one of the most active and notably flexible corporate investors in the AI space. Despite having Alphabet as its sole limited partner, GV operates with full independence in its investment decisions. This unique structure allows the firm to confidently back startups that directly compete with Google’s own products, a strategy famously demonstrated with Slack and now applied to AI companies challenging Alphabet’s internal initiatives. Managing partners Dave Munichiello and Tom Hulme highlighted in our interview their firm’s comprehensive investment strategy, writing checks across the entire AI stack—from foundational chips and compilers to diverse applications—and engaging at both early and late stages. Crucially, GV demonstrates a distinct willingness to accept premium AI valuations when they believe the underlying opportunity truly warrants it.

Munichiello articulated the rationale behind this stance, observing, "When we look at companies that are coming in to raise, the revenue run rate is insane. These companies are growing incredibly fast, faster than ever before." He underscored the difficulty of returning to evaluate non-AI companies once immersed in the rapid growth of the AI sector: "And it’s very hard to spend a lot of time looking at AI applications companies, and then go back to looking at other companies." This perspective reflects a profound belief in the unparalleled growth trajectories and market potential of leading AI startups, justifying what might appear to be elevated valuations in other contexts. Hulme further emphasized GV’s commitment to supporting disruptive innovation, regardless of its competitive implications for its parent company. Their approach signals a recognition that the current AI wave is not just another tech cycle but a fundamental reordering of industries, demanding a bold and unconstrained investment philosophy to capture its immense value.

In summary, 2025 marked a pivotal year for AI investment, characterized by rapid acceleration and a fundamental rethinking of venture strategies. The insights from these six active investors—Accel, Dell Technologies Capital, Foundation Capital, GV, AI Fund, and Sierra Ventures—collectively paint a vivid picture of an industry in dynamic flux. Key takeaways include the critical importance of physical infrastructure, particularly compute power, as a foundational bottleneck and an emerging area of opportunity; the indispensable role of high-quality, domain-specific data as the primary differentiator for AI models and applications; the strategic imperative for startups to carve out niche categories and deliver exponential productivity gains to compete with tech giants; and the evolving investment models, from traditional venture capital to venture studios, all striving to capitalize on AI’s transformative potential. Whether investing at the silicon level, building companies from the ground up through corporate partnerships, or accepting premium valuations for hyper-growth applications, these investors underscore that AI is not a singular phenomenon but a multifaceted wave lifting and reshaping every sector of the global economy. The journey through 2025 has clearly demonstrated that the pursuit of AI innovation demands foresight, adaptability, and an unwavering belief in its capacity to generate unprecedented value in the decades to come.