This Early Groq Investor Expects Half Her Bets To Fail

Sandhya Venkatachalam spent the initial phase of her career developing technology firms. She directed product at an early-stage data center hardware company acquired by Cisco, then served as a product executive at Skype prior to its acquisition by Microsoft. Those positions immersed her in data and machine learning long before artificial intelligence became the dominant focus of venture capital.

Subsequently, she became a general partner at Social Capital, where she spearheaded early institutional funding rounds in AI chip manufacturer Groq, before moving on to invest at Khosla Ventures. At present, she operates as the founder and managing partner of Axiom Partners, a $52 million fund that finances startups leveraging AI to handle tasks within sectors like construction, industrials, and insurance.

During an interview with Crunchbase News, Venkatachalam explores why she looks beyond traditional founder profiles, the elements that create a durable AI enterprise, and how her early backing of Groq influenced her investment philosophy.

This conversation has been adjusted for clarity and length.

You invested at Khosla Ventures prior to launching Axiom. What lessons did you carry over from that experience, and what did you hope to execute differently?

Venkatachalam: One key takeaway was Vinod Khosla’s expansive perspective regarding the origins of exceptional founders. Silicon Valley has historically favored a rather narrow archetype for who can forge the next major AI enterprise: someone possessing a Stanford computer science or machine learning background, or prior tenure at OpenAI. Instead, we actively seek nonobvious founders, especially within nonobvious industries.

Another lesson involved our approach to risk assessment. Rather than inundating a company with every imaginable diligence question, we zero in on the specific risks critical to achieving its upcoming milestones. Can this team execute on its promises? And if successful, could the ultimate impact be monumental?

This strategy requires accepting that numerous wagers will fail while keeping our sights set on outliers. I believe my investors back me to pinpoint future categories rather than follow the crowded paths everyone already recognizes.

The primary distinction at Axiom is that we structured the organization around individuals actively working directly with AI. Without regular involvement in building, productizing, pricing, or bringing AI to market, maintaining relevance is exceedingly difficult. Our personnel includes practitioners performing those exact functions in other capacities. They keep our investment perspective sharp, and founders appreciate collaborating with them because they have navigated identical operational hurdles.

Additionally, we utilize AI across the firm. We developed an internal system named the Axiom Brain to analyze market patterns, discover compelling people and enterprises, and accelerate diligence procedures alongside other administrative tasks. For me, the true advantage lies in the capacity for rapid action.

You mentioned that several of those AI practitioners hold outside employment. How does their involvement with Axiom function?

Venkatachalam: They dedicate specific hours to Axiom on a part-time basis and receive carried interest in the fund. They operate as true partners in the enterprise rather than holding symbolic positions on an advisory roster.

Their external roles are vital to our framework. Some of the finest angel investors are individuals actively operating and building in the market. I do not require them full-time; in fact, they would lose value for Axiom if they disconnected from the very work keeping them close to industry realities.

Axiom states that it invests in “AI for the real world.” What does that criteria entail when assessing a startup?

Venkatachalam: Our perspective dictates that AI should serve a significantly wider demographic than just the early adopters already utilizing it. We examine markets underserved by technology where AI delivers a tangible outcome instead of merely acting as an additional software tool.

That focus frequently directs us toward construction, industrials, or insurance. Certain portfolio ventures incorporate hardware, sensors, or robotics, while others remain strictly software-driven. The unifying thread is their execution of labor that matters to customers operating in physical industries.

Typically, we steer clear of products resembling standard enterprise software tools. We look for AI that clearly delivers a finished result.

You have highlighted a transition from software utilized by humans to digital workers executing entire jobs. Are clients genuinely paying for AI out of standard labor budgets?

Venkatachalam: Yes, and that dynamic serves as one of our core investment filters. Even when a portfolio company operates at an alpha or design-partner phase, we perform due diligence to verify whether customers are willing to purchase the product under those terms. We frequently observe contract values reaching the hundreds of thousands of dollars, far exceeding the modest contract sizes typical of midmarket software tools.

We have watched this purchasing behavior manifest across the vast majority of our portfolio companies.

AI products are accelerating in development speed and ease of replication. What attributes grant a company enough durability to scale into a major enterprise?

Venkatachalam: Engaging in critical operations within a customer’s business—where that work holds high financial value—makes a company exceptionally difficult to replace. You are managing what we define as the final mile of the task.

Within industrial environments, for example, delivering a concrete outcome demands deep integration with the client’s existing systems. You must comprehend their data, train models upon it, master essential workflows, and stand firmly behind the final output. Accomplishing that requires far more than simply layering an interface over an existing model.

Such relationships and technical capabilities prove challenging for competing startups to duplicate. Furthermore, they involve labor that major AI model developers may lack the inclination to pursue independently.

Prior to Axiom, you backed Groq at a time when AI inference was far from an obvious investment target. What motivated that decision?

Venkatachalam: My background spans both hardware and software. At one point, I grew curious about why Google manufactured its own networking switches instead of purchasing them from third-party vendors. Investigating that query revealed that the company was designing its own proprietary chips as well.

That inquiry led me to Jonathan Ross, who had participated in those initiatives and departed to establish Groq. I began studying why major technology enterprises were building chips to train models. Subsequently, Jonathan argued that inference would represent a vastly larger future market.

To be entirely transparent: in 2016, I barely grasped inference. However, assuming those models would proliferate, it stood to reason that third parties would build applications above them and require the underlying infrastructure to support that expansion. That realization catalyzed my investment.

How did that background influence your current evaluation criteria?

Venkatachalam: It demonstrated the value of arriving slightly ahead of the curve and exercising patience. You do not need to be radically contrarian, but you must identify an opportunity before consensus forms around it.

In many ways, our underlying thesis remains unaltered. We continue to ask what will be constructed on top of AI infrastructure and core models. Our goal is to invest while the answer is still emerging, prior to widespread agreement.

What occurs when one of those early-stage bets fails to pan out?

Venkatachalam: We factor that into our strategy. Operating a $52 million fund, we plan to execute roughly 35 investments, and we fully anticipate that approximately half of them will fail—whether that involves a company closing operations or simply missing the growth trajectory we targeted.

Our financial model relies on discovering an exceptional outlier. We require one standout investment to return the entire fund. Engaging early enough so that a company scales massively can easily offset numerous wagers that did not succeed.

That willingness to embrace losses is a necessary component of funding opportunities before they become obvious. It is fundamentally embedded in our fund’s approach.

?Frequently Asked Questions

01What is Axiom Partners?

Axiom Partners is a $52 million venture fund founded by Sandhya Venkatachalam that invests in startups leveraging AI to perform work in real-world industries like construction, industrials, and insurance.

02What does “AI for the real world” mean to Axiom?

It refers to AI applications that deliver tangible outcomes and handle operational tasks in underserved industries, rather than merely acting as conventional software tools.

03How does Axiom view failure in its investment portfolio?

The firm expects about half of its 35 investments to fail, relying on exceptional outlier outcomes to return the fund.

Related Crunchbase query:

• Global Venture Funding To AI Startups In 2026

Illustration: Dom Guzman