In partnership with

🗒️ StepSF Panel Recap: AI Infra, Energy, and U.S. Competitiveness

At StepSF (Aug21, San Francisco) we had the privilege of moderating one of the most pressing conversations in AI today: how infrastructure and energy will shape U.S. competitiveness in the AI era. I led the discussion, joined by a panel of leaders at the intersection of venture capital, private equity, consulting, and infrastructure:

• Ibrahim Alsuwaidi Partner, DCM Ventures

• Syed Natif Nawaz  Managing Partner, Aurum Equity Partners

• Lari Hämäläinen Senior Partner, McKinsey & Company

• Val Bercovici Chief AI Officer, WEKA

Together, they unpacked why AI infrastructure is no longer just a technology challenge it’s a national priority.

From Innovation to National Security

Alsuwaidi set the tone with a sharp observation: “Investment in AI is a down payment on national security.” In his view, the first nation to achieve artificial general intelligence will secure lasting economic and geopolitical advantages.

Nawaz extended that point, comparing AI infrastructure to roads and railways. This, he argued, requires government involvement from day one something that wasn’t true in the early days of the internet or cloud.

The Energy Bottleneck

Panelists agreed: the real limiting factor is energy. Only about 10% of today’s data centers can handle modern AI workloads, and the U.S. has added little net new power generation capacity in the past 30 years. Meanwhile, AI could demand 12–14% of total grid power by 2030.

When asked where future supply might come from, perspectives diverged:

• Alsuwaidi placed his bet on nuclear as the most stable baseload.

• Nawaz pointed to natural gas as the near-term solution, with small modular reactors still years away.

• Hämäläinen highlighted solar as an underutilized option.

• Bercovici argued for “virtual energy” radically improving GPU efficiency to create capacity without new plants.

Token Economics and Access

Bercovici introduced another provocative concept: economies may soon be measured by gross token production rather than GDP. More tokens mean more reasoning cycles and higher-quality outputs. But access is uneven. Developers spend thousands per month to unlock the best tools, while most users are limited to constrained free tiers.

His solution: universal basic tokens a public-private initiative to ensure broad access to AI’s benefits.

The Coming Surge of Agentic Workloads

The panel also explored what comes after chatbots and copilots: agentic workloads. These multi-step, enterprise-scale agents will run hundreds of tasks in parallel, driving massive demand for compute and tokens.

Hämäläinen predicted a more heterogeneous landscape ahead: distilled models running on-prem for routine operations, with frontier models reserved for high-value reasoning. Alsuwaidi added that efficiency in architectures will matter as much as raw infrastructure buildout.

Public-Private Partnerships and Global Competition

Unlike cloud adoption, where private industry led, AI requires government and industry to work together. Nawaz pointed to Latin American projects as proof that policy, capital, and community can align from the outset.

Hämäläinen sounded a note of urgency: the U.S. risks falling behind China not in innovation, but in renewable deployment and the skilled labor needed to scale data center construction.

Closing Thoughts

• Alsuwaidi warned that the new digital divide will be the ability to command intelligence.

• Nawaz argued AI will unlock human potential at scale, transforming industries beyond tech.

• Hämäläinen urged focus on re-employment, capital distribution, and social stability.

• Bercovici called for universal basic tokens to make AI benefits broadly accessible.

Big takeaway: The AI race isn’t just about faster chips or bigger models. It’s about energy, efficiency, and equitable access and the ability of the U.S. to align public and private efforts at unprecedented speed.

AI INFRA SUMMIT / NOV7 / SAN FRANCISCO

Our argument is clear: building together is non-negotiable.

Only the most agile, ecosystem-scale partnerships between founders, hyper-scalers, energy innovators, investors, and policymakers will win in this new race. The AI INFRA SUMMIT 4, taking place November 7 in San Francisco, is positioned as the only forum capable of convening this coalition and igniting the operational alliances required to build at speed and at scale.

Blitzscaling Investor Summit // Sep11, Palo Alto

Reid Hoffman Discusses AI Investment Strategy at Blitzscaling Summit

Reid Hoffman shared his investment philosophy and AI market insights at the recent Blitzscaling Investor Summit, emphasizing the importance of long-term thinking in equity investments with 5-10 year horizons rather than short-term gains. He highlighted his focus on identifying Silicon Valley blind spots that could yield tens or hundreds of billions in returns, citing Manus as an example—an AI drug discovery company that accelerates existing scientific expertise rather than attempting to simulate all biology or create pure AI agents.

Looking toward enterprise adoption, Hoffman stressed that current AI utilization falls far short of available capabilities, with most people not leveraging existing tools effectively. He advised Fortune 500 executives to begin immediate experimentation with AI applications, sharing learnings across organizations rather than treating them as isolated pilot projects. However, he cautioned that despite the acceleration in AI capabilities, investors must maintain discipline as not all AI investments will succeed, and standard business metrics around go-to-market strategy, enterprise integration, and network effects remain crucial.

The discussion also addressed emerging challenges including the growing gap between human readiness and technology advance ment, the content creation versus consumption imbalance, and the critical need for truth determination systems as AI speeds up decision-making processes. Hoffman emphasized that while AI will transform many aspects of business, fundamental human nature—including tribal connections and competition—will persist even as we gain access to enhanced creativity and longer lives.

Llama Lounge 19 // Sep11, Palo Alto

Upcoming Events

BA Startups Collectively Secured $17.6B in September

Bay Area startups closed on $3.4B in the second week of September, including seven megadeals that totaled $2.96B this week. Those megadeals made up 87% of this week's venture funding to startups. While that percentage may seem high, more than 80% of YTD funding for startups has been in the form of megadeals. $2B of this week's total came in just two rounds – Databricks and PSIquantum.

Companies are choosing to stay private longer; the average age of companies at IPO continues to increase, now up to fourteen years according to Professor Jay Ritter at the University of Florida. Databricks is a prime example of companies choosing to stay private (much) longer. The Databrick's round this week is the first Series K we've seen here. We're seeing Series E, F and even Gs more frequently as companies elect to raise additional late stage capital rather than venturing into the public market. Multiple secondary sales by the company that enable employees and early investors to get some liquidity removes the pressure to IPO that those groups would otherwise apply. A burgeoning secondary market (doubling in size since 2022) offers alternatives to VCs pressed by LPs for distributions. And the availability of late stage funding - or venture debt funding – is the remaining element needed for these companies to stay private indefinitely.

For startups raising capital: The Pulse of the Valley weekday newsletter keeps you current with the startups that are getting funded and who's investing in them – and not just the big, high profile deals. We surface the 50% of early stage fundings – and their investors - that close quietly and never hit the press. Details include investor and executive connections and contact information on 25+ years of fundings and business activity. Founders get the full $50/month package for just $7/month with LinkSV's Entrepreneur membership, details and signup here.

Follow us on LinkedIn to stay on top of what's happening in 2025 in startup fundings, M&A and IPOs, VC fundraising plus new executive hires & investor moves.

Early Stage:

• Higgsfield AI closed a $50M Series A, using our AI-native, browser-based platform to condense ideation, editing, and post-production into a single workflow and produce cinematic-quality short-form video in minutes instead of weeks.

• Brain Co. closed a $30M Series A, our mission is to help governments and corporations deploy artificial intelligence.

• Ridge Biotechnologies closed a $25M Seed, pioneering enzyme and targeted drug design to enable the next generation of precision medicines.

• Throxy closed a $6.2M Seed, designs and executes complete, done-for-you outbound systems that handle everything from deliverability and list building to decision-maker identification and expert outreach.

• Candid Intelligence closed a $5.5M Seed, compresses months of pre-construction engineering into hours of compute.

Growth Stage:

• Databricks closed a $1000M Series J, the Data and AI company with the Databricks Data Intelligence Platform that takes control of data and puts it to work with AI.

• PSIquantum closed a $1000M Series E, building the world’s first useful, fault-tolerant quantum computing systems using a photonic approach to leverage high-volume semiconductor manufacturing and existing cryogenic infrastructure to rapidly scale its systems.

• Cognition AI closed a $400M Series C, the maker of Devin, the first AI software engineer.

• Replit closed a $250M Series C, the agentic software creation platform that enables anyone to build applications using natural language.

• Aven closed a $110M Series E, transforming access to low-cost capital by helping consumers unlock the value of their existing assets.

Supermicro DCBBS | The Blueprint for Modern AI Data Centers

Supermicro’s Data Center Building Block Solutions® (DCBBS) are designed to deliver complete, modular infrastructure for modern AI data centers. From GPUs and servers to full racks and facility-side liquid cooling, DCBBS enables rapid deployment of scalable, high-performance environments optimized for AI and mission-critical workloads.

Rapid Deployment
Gain first access to the latest GPUs, CPUs, storage, networking, and liquid-cooling technologies. Supermicro integrates and validates systems at the rack and cluster level, reducing time to production with plug-and-play deployment.

Complete Lifecycle Management
Supermicro serves as a one-stop partner, covering design, assembly, onsite deployment, and support. With capacity to deliver 5,000+ racks per month (including 2,000+ liquid-cooled), global production facilities ensure availability at scale.

Customizable to Any Workload
The modular “building block” approach allows customization at the server, rack, cluster, cooling, and power level. Each solution is tuned to meet specific application requirements and power budgets.

System-Level Building Blocks

• Servers & Components: Flexible configuration of CPUs, GPUs, NICs, DIMMs, storage, and chassis for optimal performance.

• Advanced Cooling: Direct-to-chip liquid cooling for CPUs, GPUs, PCIe switches, VRMs, and power supplies, improving compute density and cutting power costs.

• Fast Production: Standardized components reduce supply chain bottlenecks and speed delivery.

Rack & Cluster Solutions

• Pre-validated scalable units allow up to 256-node AI clusters.

• Optimized cable layouts for airflow and performance.

• Non-blocking network topologies for high-speed node-to-node communication.

Facility-Level Infrastructure

• Liquid Cooling: End-to-end DLC deployment, from cold plates to coolant distribution units and data-center-scale cooling towers.

• Efficiency Gains: Up to 40% power and water savings, 98% heat capture, and up to 60% space savings with 2.5× compute density compared to air-cooled systems.

• Quiet Operations: Data centers operate at library-level noise (~50 dB).

📍 Learn More: Supermicro DCBBS

Your Feedback Matters!

Our mission is to provide an insider's view of Silicon Valley's undercurrents – insights often overlooked by mainstream sources. While many newsletters offer broad market overviews, we focus on delivering a unique, in-depth understanding of the local ecosystem. We share behind-the-scenes conversations, introduce key players we meet at events, and offer exclusive insights.

Your feedback is crucial in helping us refine our content and maintain the newsletter's value for you and your fellow readers. We welcome your suggestions on how we can improve our offering. [email protected] 

Logan Lemery
Head of Content // Team Ignite

Keep This Stock Ticker on Your Watchlist

They’re a private company, but Pacaso just reserved the Nasdaq ticker “$PCSO.”

No surprise the same firms that backed Uber, eBay, and Venmo already invested in Pacaso. What is unique is Pacaso is giving the same opportunity to everyday investors. And 10,000+ people have already joined them.

Created a former Zillow exec who sold his first venture for $120M, Pacaso brings co-ownership to the $1.3T vacation home industry.

They’ve generated $1B+ worth of luxury home transactions across 2,000+ owners. That’s good for more than $110M in gross profit since inception, including 41% YoY growth last year alone.

And you can join them today for just $2.90/share. But don’t wait too long. Invest in Pacaso before the opportunity ends September 18.

Invest While You Still Can

_{Paid advertisement for Pacaso’s Regulation A offering. Read the offering circular at invest.pacaso.com. Reserving a ticker symbol is not a guarantee that the company will go public. Listing on the NASDAQ is subject to approvals.}