Sep 19, 2025 · 50m · neon-show
GPT-5 is NOT Your Competitor It's GPT-7 |Ashu Garg, Foundation Capital | Databricks,Turing, Cohesity
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
Foundation Capital General Partner Ashu Garg joins host Siddharth Ahluwalia to discuss the technological shift toward autonomous reasoning agents and the strategic imperatives for building defensible, decacorn-scale AI enterprises against frontier model competition.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. How this is scored →
Siddhartha as informed peer 4.0 Guest teaching 4.9 Guest disagreement 1.7 Siddhartha pushing back 2.1
speaking balance: gold is Siddhartha, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 21:23 Perplexity is not in the same zip code as model giants
Ashu firmly dismisses Siddharth's suggestion that Perplexity is competing at the same tier as OpenAI, Google, and Anthropic.
Hardest push from Siddhartha ▶ 27:46 Pushback on startup survival as models subsume workflowsSiddharth directly challenges Ashu's optimistic startup thesis by citing how model providers copy downstream startup workflows like Windsurf.
Biggest teaching moment ▶ 37:09 Explaining the antitrust and CapEx mechanics behind talent lift-outsAshu systematically educates Siddharth on why regulatory delays and multi-billion-dollar GPU budgets make massive talent buyouts mathematically rational for Big Tech.
Siddhartha holds their own ▶ 47:47 Synthesizing greenfield revenue models versus legacy SaaS displacementSiddharth demonstrates sharp domain knowledge by contrasting prosumer ARR ramps like Lovable against traditional SaaS displacement budgets.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Siddhartha as informed peer | Guest teaching | Guest disagreement | Siddhartha pushing back | Why |
|---|---|---|---|---|---|---|
| The Technological Evolution: From Simple Wrappers to Autonomous Agents | 4 | 5 | 1 | 1 | Siddharth introduces Ashu's thesis on reasoning and reinforcement learning replacing simple RAG workflows. Ashu provides a masterclass on the evolution from conversational wrappers to autonomous multi-step reasoning agents. | |
| The Agentic Web: Redesigning UX, Payment Rails, and Governance | 3 | 6 | 1 | 1 | Siddharth asks about the implications of an agentic web. Ashu thoroughly explains why current browser design, payment rails, and fraud systems fail when transactions are executed by autonomous agents via micropayments. | |
| Parallels to the 1990s Internet and Synchronous Global Adoption | 4 | 4 | 1 | 2 | Ashu compares the current AI revolution to the 1997 internet boom, highlighting the unprecedented speed of synchronous global adoption in places like India. Siddharth contributes context regarding high token consumption in India. | |
| Founder Mindset: The Value of Curiosity, Intensity, and Geographic Hubs | 3 | 5 | 2 | 1 | Ashu highlights why repeat founders often carry baggage and explains why Silicon Valley remains an unmatched ecosystem despite global talent distribution. Siddharth prompts with questions on founder advantages and geographic hubs. | |
| The Big Tech Battle for the Consumer and Prosumer Interface | 4 | 5 | 3 | 3 | Siddharth presses Ashu on whether Perplexity should be considered a top-tier contender against OpenAI, Google, and Anthropic. Ashu directly clarifies that Perplexity is not in the same tier at the foundational model level. | |
| Startup Defensibility: Vertical Workflows, Post-Training, and Model Independence | 5 | 5 | 2 | 4 | Siddharth challenges Ashu by raising the risk that model providers will copy vertical workflows (citing Windsurf). Ashu acknowledges the dilemma, detailing defensive architectural strategies like local code graphs and customer-specific post-training. | |
| Enterprise Distribution, Relationship Capital, and Agility Under Disruption | 4 | 4 | 2 | 2 | Siddharth asks whether legacy SaaS distribution and enterprise relationship capital remain defensible moats. Ashu illustrates with Salesforce's struggles that distribution alone fails without rapid product innovation. | |
| Preparing for GPT-7, Structural Acqui-Hires, and the Frontier Talent Scarcity | 5 | 6 | 2 | 3 | Ashu explains the necessity of building for GPT-7 rather than current models and breaks down why Big Tech pays massive sums for talent lift-outs due to antitrust scrutiny and massive GPU CapEx allocations. Siddharth probes on why frontier skills remain unteachable. | |
| Evaluating Decacorn Ambition, Revenue Growth, and Wedge Strategies | 4 | 4 | 1 | 2 | Siddharth questions whether AI startups should focus on greenfield prosumer tools like Lovable or enterprise replacement wedges. Ashu evaluates the tradeoffs between high-velocity consumer wedges and large enterprise contract sizes. |
Statements from this episode (26)
Garg: AI is transitioning from assistive co-pilots to human-replacing autonomous agents
“If you step away from the technology, ultimately what is going on is we're seeing the shift from co-pilots which assist a human
To an autonomous agent that for a certain class of tasks can replace a human.”
Garg: Every person will eventually have one or multiple personal AI agents
“First and foremost, the world we're headed to is each one of us will have one or multiple personal agents.”
Garg: Stripe's fraud analytics are designed solely to detect human fraud
“If you look at current payment systems, if you look at Stripe, and all of Stripe's fraud analytics and fraud systems, they're designed to test for a human being, and is there human fraud happening?”
Garg: Zero-cost AI agent activity will force a reinvention of the internet
“In a world where the cost of the activity comes to close to zero, because an agent is doing the activity, the infrastructure to enable the activity will all have to change. So I think that's what's underway today. And that is going to cause us to reinvent the …”
Garg: AI will see a decade of internet progress in five years
“I anticipate that over the next five years, we will see the same level of progress that we saw over a decade in the internet.”
Garg: India's ChatGPT consumption matches or exceeds US usage
“Today, you're seeing that chat GPT consumption in India is actually simultaneously with the U.S. I think there is as much consumption, perhaps more.”
Garg: Founder intensity and curiosity matter more than prior startup experience
“Certainly I think the value of experience Has diminished in the value of intensity and curiosity has exploded.”
Garg: Bangalore has a shot at matching Silicon Valley in 10–20 years
“Maybe in a decade, maybe in two, I think Bangalore is an interesting ecosystem. I think in a decade or two, I think Bangalore, you know, has a shot at being there.”
Garg: Perplexity is not in the same league to own consumer AI
“I think Perplexity has a shot at that. So I wouldn't rule it out. I think Arvind has done a remarkable job. I think, ah, I am blown away by what he has accomplished, and I think he has, you know, he has put a stake in the ground. I think he has a shot. But to …”
Garg: Anthropic, Google, and OpenAI lead the consumer AI battle
“Ah, but really when you think about the players today, I think the three contenders today are really Anthropic, Google, And OpenAI, and I think Microsoft and Meta are sort of fast followers, ah, and Apple is the dark horse.”
Garg: Open-source models trail proprietary SOTA models by only 6-12 months
“I also think that open source has demonstrated that while it can't compete with the SOTA models, it's only six to 12 months away.”
Garg: OpenAI has pivoted into a consumer and prosumer applications company
“So, OpenAI has pivoted from being a developer-focused company to being a consumer and prosumer-focused applications company.”
Garg: Six-month-old portfolio startup is about to sign $30M first deal
“I have a startup that's, you know, barely six months off the ground. They're about to sign what might be a thirty million dollar deal.”
Garg: AI application startups must realize their model vendors are competitors
“Every startup in the AI application space, every, you know, service as software startup has to first and foremost internalize that their vendor is also their competitor. The model provider is both vendor and competitor. And that's a hard place to be. And they …”
Garg: Successful AI founders are more technical than SaaS-era founders
“So, in some ways, the founders today that I see being successful are more technical than the founders in the SaaS way. Which also means that repeat entrepreneurs who grew up in a less technical world are struggling with the technical complexity. They're strugg…”
Garg: Dozens of startups are poised to eat Salesforce's lunch
“I think there are dozens of companies that are poised to eat Salesforce's lunch. But those companies, despite the distribution disadvantage, are positioning themselves to do so.”
Garg: Relationships matter more in AI because enterprise automation is high-risk
“If anything, relationships matter more today because people are putting more at risk. If you're going to a company and say I'm going to outsource this process for you, that's pretty scary for them too. So if they're going to do something that's scary for them,…”
Garg: Cohesity founder Mohit Aron is launching sales tech startup Siphon
“Mohit Aron, who's starting a new company, ah, going after, you know, sales tech, Siphon, and this is his third company. And Mohit's had two Decker coins, and, you know, this will be his third.”
Garg: AI startups must build to beat GPT-7, not GPT-5
“If you cannot have a product that can compete with GPT-Five, you're dead on arrival. Because you want to spend the next five years losing money. To the point that you can have a profitable business five years from now, and you'll be competing with GPT-Sev.”
Garg: Big Tech AI team lift-outs bypass 18-month antitrust delays
“So, because of the antitrust issues, companies are choosing to, you know, lift out teams, including the IP. But the bet they're making is the IP is changing so rapidly, you take a licensed copy of the IP, you leave the, you leave a copy behind, And that team, …”
Garg: Meta will spend $50B to $70B on AI GPUs
“Look there, Meta will spend well north of fifty billion, I think some number between 50 and seventy billion dollars on the underlying data set, on GPUs, essentially.”
Garg: Only 1,000 people globally are truly innovating at frontier model layer
“When I say there's a thousand people, those are the people that are really innovating at the model layer.”
Garg: Databricks will be a trillion-dollar company one day
“You know, I think Databricks will be a trillion dollar company one day.”
Garg: More decacorns will be created in next decade than prior decade
“I think many more Decacorns will get created over the next decade than over the last decade.”
Garg: Code generation tools have the fastest zero to $100M ARR ramps
“If you look at, ah, code generation tools, where there has been the most zero to a hundred, you know, the fastest zero to a hundred ramp have been code generation tools.”
Garg: Selling training data to frontier AI labs is extremely difficult
“That said, I think providing data to the soda labs, Is a very hard business. It's a great business and scale proved to be very successful. Turing has been very successful. There are a couple of other players, but it is not one for the faint of heart.”