Vasanth Ganesan

Vivek Raghavan, co-founder of Sarvam AI, the company that just shipped India's first foundation model trained from scratch, warned yesterday that India risks becoming a "digital colony" if it does not build foundational AI itself. At the same summit, JioStar, India's largest media conglomerate, partnered with OpenAI. I recently wrote on Substack about why this keeps happening: "Indians Rule Big Tech. Why Can’t India Build?" (https://lnkd.in/ewQBMCYf). The structural forces are thirty years deep. Source: The Economic Times (https://lnkd.in/eGEgSb95)

4

We refresh our views on the 6 Ps driving AI/data center power demand growth -- Pervasiveness, Productivity, Price, Policy, Parts and People. Pervasiveness: Compute and power intensity per server for inference likely to rise. Based on our TMT team's outlook, our inference forecasts reflect an increase in power intensity, though overall we are still largely assuming lower-power servers are deployed. As hyperscaler AI investment accelerates, there is a more intense focus on AI use cases and their value relative to invested capital. Productivity: While integrating increased deployment of latest Vera Rubin generation of servers improves energy efficiency via greater compute per unit of power, this is offset by higher industry demand outlook as compute/token demand remains pent-up. Price: Hyperscaler commentary, budget increases and contracting suggest willingness to pay Green Reliability Premiums to ensure diverse sourcing of long-term power supply. Policy: Increased popular concerns that data centers are driving Affordability/Reliability risks are likely to remain elevated, prompting increased push among data center power consumers, policymakers and power producers towards contracts that attempt to ring-fence data center buildout from impacting broader customer Affordability and Reliability. Parts: Constraints in timing/availability of power generation equipment continues to drive how data center power will be sourced as part of an overall all-in strategy -- renewables/battery storage/simple cycle natural gas in the shorter term, natural gas combined cycle in the medium term, nuclear in the longer term. People: We continue to see risks that skilled labor availability constraints -- particularly electricians -- represents a key risk to execution. This is also, in our view, helping to drive greater consideration for behind-the-meter power solutions that require less electricity transmission. As mentioned in our data center power demand note, the increase in our Utilities team's US outlook is driven by BTM solutions.

1

Xora Innovation has led Hang Ten Systems’ additional $53M Seed-2 funding round, bringing the company’s total funding to $85M. We’re proud to invest alongside Navin Chaddha at Mayfield, Aramco Ventures, Lip-Bu Tan, Sanjay Mehrotra, and others — and to welcome Jerry Yang to the board. We believe the gap between enterprises that use AI to real advantage and those that do not will increasingly define entire industries. As access to leading models becomes more widespread, the hard part is making AI work in the systems and workflows that actually run an enterprise: connecting models and agents to proprietary data and systems of record, rebuilding applications and workflows around their capabilities, and putting the security, controls, evaluation, and reliability required for production use in place. This is difficult because enterprise technology rarely operates in a greenfield environment. Most organizations have deeply interconnected estates of packaged software, custom applications, legacy systems, integrations, and business processes built up over decades. Capturing the full value of AI therefore requires changing not only that software, but also how it is built, modified, and operated. But AI is also changing the economics of transforming that estate. Hang Ten is built around this shift, with a delivery model fundamentally different from traditional, headcount-heavy enterprise services. The company combines agentic code generation and reusable AI skills with small teams of engineers who bring deep domain expertise — enabling them to move faster and deliver 10× more value on existing software development and transformation programs. The early traction is strong. Within weeks of founding, Hang Ten signed multiple engagements with global enterprises, with projects already in production and several successfully completed. We’re backing Vishal Sikka because he is one of the rare leaders who combines deep technical understanding of enterprise software and AI with firsthand experience navigating the organizational and operating complexity of transformation at scale. He has spent his career doing exactly that — from serving as SAP’s first CTO and leading the development of SAP HANA to serving as CEO of Infosys. Vishal and the Hang Ten team are now applying that experience to build a fundamentally different model for the global technology services industry. We’re excited to deepen our partnership with Vishal, Navin Budhiraja, Sanjay Rajagopalan, Tao Liu, Frank Yu, Yusuf Safdari, Pradeep K Panicker and the entire Hang Ten Systems team as they build what we believe can become a defining company in enterprise technology services. Check out what the team is building in this TechCrunch exclusive (link in first comment).

413

18 Comments

This article, co-authored by Vardaan Ahluwalia and me lays out India's significant new Research Development and Innovation Fund initiative in the context of other successful interventions: when governments and the private sector work together, they can make a transformative difference to deep technology entrepreneurship. Equally, initiatives to help manage financial risk should be followed by structural reforms in areas ranging from intellectual property regulation, ease of clearances for start-ups, tax rationalization in capital gains and options that helps start-ups attract talent, and a visa regime that attracts leading global researchers. https://lnkd.in/gk92YSZa

60

Sometimes the investors who know your space best are the hardest to convince. In a recent conversation from nextNYC's “How I Raised My Seed” series, Manvitha Mallela of Flock AI, Yuvraj Singh Shergill of ARUSTO and I talked about something founders often find surprising during fundraising. Specialist investors have usually seen every version of the idea before. They’ve watched similar companies try and fail, sometimes multiple times. So when you show up convinced that now is the right moment, they may be more skeptical than someone seeing the opportunity for the first time. That’s why generalist investors sometimes move faster. They’re encountering the problem with fresh eyes. We also talked about outreach. Cold emails usually fail when they’re generic. But when a founder references something specific, an investor’s podcast, thesis, or past investment, the response rate changes dramatically. Fundraising is a numbers game. But it’s also a relevance game. Thanks to Qapita for sponsoring this series! #venturecapital #fundraising #startups #founders #seedround #investors #entrepreneurship #howiraisedmyseed

5

2 Comments