The AI Sector’s $600B Revenue Gap Bubble
Goldman Sachs, Sequoia reports of AI bubble,$ 600b p/a AI hardware spend, Meta, xAI spend big on Nvidia GPUs, Groq, Etched, ElevenLabs, Crypto VC, China FSD
Dear Tech Buzz Reader,
Last call for the Real Intent membership 40% discount. Grow your sales pipeline with consumer intent leads direct in your inbox at <$0.05 per lead.
∆ Tech in the News
Equity Fund Pile-on - Global investors piled ~$15.41b into equity funds last week and ~$21.08b in the prior week, on expectations of U.S. rate cuts, .
Amazon High-end Play - Amazon is taking a minority stake in the merged entity created by Saks's acquisition of Neiman Marcus, to get a foot in the luxury goods market it has failed to crack so far.
Crypto VC Crackdown - The SEC is investigating crypto VC firms for alleged ‘pump and dump’ violations of securities laws, per BlockTower CIO. .
China FSD - Beijing plans to support use of self-driving cars in online ride-hailing services, cites municipal draft rules for autonomous vehicles.
James Dean Lives - ElevenLabs’ Reader app uses AI to resurrect deceased actors' voices to read audiobooks with their estates’ permission.
The Hottest Thing in Tech this Week
The AI Sector’s $600B Revenue Gap Bubble
Two significant reports just stoked fears of an AI Bubble. One was a Goldman Sachs investor note questioning gen AI’s ROI and value. The other was a Sequoia Report highlighting the massive $600 billion hardware spend of the global AI sector. This data, and other factors including AI’s environmental footprint, is giving rise to very real concerns about the sustainability of AI startup valuations and expectations. So how realistic are the fears, and what are some possible solutions to the problem?
Why it Matters
A new Sequoia Report on the AI industry finds that the sector needs to earn ~$600 billion per year to pay for the massive hardware spend required to run its tech. Goldman Sachs has until now been a vocal cheerleader for AI’s potential as a world changing technology and investment opportunity. The report Gen AI: Too much spend, too little benefit? dialled back some of its prior enthusiasm with a dose of realism. It pointed out that outside of a few leading companies like Nvidia, OpenAI, Microsoft, and Google, Generative AI isn’t generating ROI and value.
It is unclear if it ever will. Sequoia Analyst David Cahn believes we could be witnessing the growth of an AI-fueled financial bubble. He points out even optimistic projections for major tech companies' AI revenues fall well short of breaking even. Cahn believes the industry must temper expectations of quick profits or we face a bust and a possible global economic crisis.
Costs of Doing (AI) Business
Nvidia alone earned nearly $47.5b in datacenter hardware revenue last year from AI and HPC. Companies like AWS, Google, Meta, Microsoft, and xAI are investing heavily in their AI infrastructure. Meta "scorched earth" strategy involves acquiring ~350k Nvidia H100 GPUs in 2024, worth ~$30b (larger than the Apollo Moon Mission). xAI’s planned GPU supercomputer clusters in Tesla's gigafactory will also use 350k Nvidia GPUs at similar cost.
On top of that there are significant power grid and environmental costs involved in generating the energy to run the data centers needed.
More Efficient Solutions
One solution is making AI more efficient. Big tech is investing heavily in this. Google DeepMind just published new research introducing JEST - a method that accelerates model training while reducing computing requirements. Microsoft just unveiled its MInference, tech to accelerate its LM processing by up to 90%, reducing costs and energy use. Startups like Groq and Etched are innovating in AI chip design for efficiency and Baseten is using multiple AI models to offer high efficiency and AI scalability.
Another solution is State-Space Models which are effective for sequence modeling to summarize past information to predict future states or outputs. Stanford & CMU’s Cartesia is pioneering this innovative approach.
On the energy front, Intel just lodged a patent to improve AI energy resource management via a ML model that predicts likely energy emissions in processing tasks and prioritizing renewable energy sources.
“Speculative frenzies are part of tech, and not something to be afraid of. Those who remain level-headed through this moment have the chance to build extremely important companies…”
— David Cahn, Sequoia Analyst
Go Premium to get the Perplexity AI DEEP DIVE featuring:
Full AI search market analysis
Full Business case and competition comparison
How companies are using Perplexity to add value
Job Candidates (For Employers)
Request an intro to candidates - jobs@thetech.buzz
Lockheed Martin Software Engineer, ex-Rutgers Junior Full Stack Developer. Secret clearance. Expertise in C++, PHP, SQL. Rutgers BA in Computer Science.
Meta Technical Recruiter for ML/AI roles, 98% candidate response rate, 70% hire rate. Led training for 135 sourcers.
Deals We’re Watching
Founders to Watch and Opportunities to Catch
Open Deals:
Request an intro to founders - invest@thetech.buzz
Cannabis/Hemp E-comm - New York - The first AI-Powered end-to-end online platform for the hemp industry - 2,000% growth in 2 years. (Seed+)
AI Agent Employees - New York - Sophisticated workflow automation for businesses, by a previously-exited founder. Building towards fully autonomous organizations. Backed by Tribe & Kenetic, rolling out to extensive waitlist. (Seed)
Outbound B2B Campaign AI Co-Pilot - Lexington, MA - Turn any CRM into an AI optimized sales agent for personalized customer campaigns. Founder previous exit to PE. Notable Angel on board. Top Hubspot App. Post revenue. (Seed)
Trending Tools and Buzzy Tech
Tools
xAI - Grok to integrate on X to get insights and access via an in-app pop-up.
Apple - New home accessory to get A18 chip to power for Apple Intelligence.
Intel - First fully integrated optical compute interconnect chip to scale up AI.
Kyutai - Moshi ‘real-time’ AI voice assistant beats OpenAI to market, positive reports.
Tech
Tesla/xAI - Tesla 'Gigafactory' expanding to include liquid-cooled AI data centers for supercomputers.
Radboud U. - AI system reconstructs accurate images from brain activity.
Fast-Charging Batteries - World 1st anode-free, sodium, solid-state battery.
Closed Deals:
Document Search - New York startup Hebbia raised a ~$130M Series B led by Andreessen Horowitz to offer AI-powered Document Search for asset management, legal, and professional services firms.
Deals to Watch:
Paramount Global - The film and TV giant looks set to merge with David Ellison’s Skydance concluding a months-long auction for the company’s controlling stake.
Anduril - California defense AI/tech startup with multiple defense contracts in FYE 2024. $500k revenue at 150% growth rate in 2023, strong projections. Raising $1.5B Series F at $12.5B pre-money led by Peter Thiel’s Founders Fund and Sands Capital.
SUBSCRIBE TO PREMIUM TO GET MORE INSIGHTS
Paying members get access to the rest of this post and other subscriber-only content and more. Subscribers get:
Premium deep dives and market reports
Access to exclusive content
Unlimited access to ‘Open Deals’ & ‘Startup Directory’
Unlimited warm introductions to founders and job candidates
Up to 50% OFF advertising in The Tech Buzz
0% management fee on Buzz Capital syndicate investments