/ai

Eric Vishria Says Investors Undersize AI Market Again · Digg

Eric Vishria Says Investors Undersize AI Market Again

Podcast covers AI infrastructure buildout and investing lessons from hardware and software.

TLDR

Patrick O'Shaughnessy hosted a conversation with Benchmark partner Eric Vishria. Vishria has invested for a decade in software and hardware companies including Fireworks, Sierra, Sunday Robotics, and Cerebras. Posts describe the episode as drawing lessons from cloud history for the current AI moment and sharing details from inside the AI buildout. Visible replies praise the discussion on scaling, energy limits, robotics, and the potential size of AI markets. The episode runs about 78 minutes according to one post.

Featured Source

Combined views

1.1M

30 Sources, first seen 44d ago

2.1K likes

TagsPatrick O'Shaughnessy

44d ago

first seen 44d ago

Eric Vishria Says Investors Undersize AI Market Again

Podcast covers AI infrastructure buildout and investing lessons from hardware and software.

TagsPatrick O'Shaughnessy

30 Sources, 44d ago, first seen 44d ago

TLDR

Patrick O'Shaughnessy hosted a conversation with Benchmark partner Eric Vishria. Vishria has invested for a decade in software and hardware companies including Fireworks, Sierra, Sunday Robotics, and Cerebras. Posts describe the episode as drawing lessons from cloud history for the current AI moment and sharing details from inside the AI buildout. Visible replies praise the discussion on scaling, energy limits, robotics, and the potential size of AI markets. The episode runs about 78 minutes according to one post.

Featured Source

Combined views

1.1M

30 Sources, first seen 44d ago

2.1K likes132 comments1.6K saves280 reposts

Sentiment

Positive——Negative

Summary

Not enough discussion yet.

No sentiment analysis available yet.

132 comments

1.6K saves

280 reposts

Sentiment

Positive——Negative

Summary

Not enough discussion yet.

No sentiment analysis available yet.

Related

O'Shaughnessy Holds Second Conversation With Sarah Guo

Investor shares details of second talk with Conviction founder on AI progress.

Ben Thompson Calls US AI Win Problematic

Discussing AI race dynamics with China on Patrick O'Shaughnessy's podcast.

Investors Discuss AI Stock Selloff Versus Valley Activity

Patrick O'Shaughnessy talks with Gavin Baker about market prices versus on-ground AI demand.

Investors Discuss AI Stock Selloff Versus Valley Activity

Today's Rank

—

Not ranked yet

Today's Rank

—

Not ranked yet

160012345678901234567890123456789

160012345678901234567890123456789

160012345678901234567890123456789

160012345678901234567890123456789

30 Sources

@patrick_oshagMy conversation with @ericvishria of Benchmark. Eric has spent a decade investing across software and hardware, backing companies like Fireworks, Sierra, Sunday Robotics, and Cerebras. This one is about what history teaches us about the current moment, and a dispatch from inside the AI buildout through his companies. We discuss: - What AWS tells us about how big AI can get - How the goalposts have moved for every software company - Lessons from a decade with Cerebras - China and the energy bottleneck - Benchmark's return to growth investing - Robotics Enjoy! TIMESTAMPS 0:00 Intro 3:36 What Cloud Teaches Us About AI 12:52 Sierra, Sandcastles, and AI Products 17:35 The New SaaS Competitive Frontier 28:28 AI Demand and the Energy Bottleneck 31:28 The Cerebras Story and AI Chips 45:10 The Future of Robotics 52:53 Eric’s Venture Investing Philosophy 1:07:30 Going Public, AI Value, and Jobs

@ericvishriaApparently when I force myself not to rock (iykyk), I wave my hands an overwhelmingly distracting amount.

@jaltmaWorking with Eric over the past six months has set a totally new bar for me, he's truly as good as it gets.

@dadiomovMust watch - esp the last bit about radiology. Jensen talks about new jobs created wherever "there exists a backlog of aspiration or ambition." Selling magic into segments with such backlogs is a whole new motion. This is such a good breakdown of our moment in tech history.

@blakeirEric approaches the craft of venture with a level of rigor, curiosity, and seriousness that is incredibly rare.

@ShanuMathew93One of the best pods of the year. @patrick_oshag doing god's work cranking out all this good content. Parts that stood out to me from @ericvishria >Core AI thesis: investors are still undersizing the market by treating it as zero sum. CSPs, neoclouds, inference platforms, semis, edge + apps can all work. That doesn't mean every company works. Most won't. >His SaaS framing is sharper than “AI kills software”: AI changes the competitive frontier. If agents make migrations cheap, historical moats built on implementation complexity (e.g., database migrations) + switching costs become much less valuable. >Provocative line: a legacy SaaS company can hit its operating plan every day and still destroy equity value if that plan optimizes against yesterday's competitive environment. Liked @btaylor's framing: modern AI teams are building sandcastles that they expect the next model wave to wash away. Optimizing every component for long-term permanence can become counterproductive. >Fireworks is his evidence that inference isn't automatically a commodity: same open-source models, same Nvidia hardware, but he claims ~5x speed + multiple-x throughput versus conventional cloud implementations. Execution/software still matter enormously. >Vishria's biggest AI infrastructure concern isn't capital. It's energy. Models translate compute into intelligence; if intelligence demand keeps scaling, power availability ultimately determines token supply and token cost. >Robotics framing: don't obsess over whether laundry [or whatever menial task] is the killer app. The asset is the data flywheel: high-value embodied data → pre-training → targeted post-training → increasingly general capabilities. >Best labor-market point: correctly predicting AI capability does not mean you've correctly predicted jobs. Workflow, liability, reimbursement, regulation and demand elasticity sit between technological capability and economic outcome.

@chetanpAmazing podcast with Patrick and Eric. Eric expresses our deeply held belief that it is an incredible time to start companies across multiple areas. One company will not do everything and the markets will end up bigger than most imagine.

@sonyatweetybird@patrick_oshag @ericvishria 👑

@SchornackI enjoyed this podcast. Listening to this after a great one with @GavinSBaker sets a pretty good stage for how to think about the next leg of AI. We are just getting started and don't get caught thinking zero sum.

@martin_casado@jaltma He really is