Scribbler Logo

    You have 4 summaries left

    Odd Lots

    Josh Wolfe on Where Investors Will Make Money in AI

    Mon Jul 17 2023
    Bank of AmericaNvidiaMetaAI investmentVenture capitalBig companies

    Description

    Bank of America's expansion in Paris, Nvidia's role in self-driving cars, emerging programming languages, Meta's open source models, corporate VC impact, investing in AI, future of communication, venture capital landscape, big companies' revenue generation, AI accessibility

    Insights

    Investing in AI

    Investing in AI is highly competitive, but great founders want to work with top firms. There is a five-year psychological bias in investing where people want to invest where others were five years ago. Evaluating credibility of founders and their academic publications is important when assessing AI investments.

    AI Models and Communication

    Communications will increasingly be generated by AI, leading to a flood of emails, texts, and tweets that are not written by humans. This chat phishing phenomenon will make people yearn for in-person communications and private clubs where human interaction is guaranteed. The influx of money and talent in AI may ironically bring us closer to valuing human communication more.

    Big Companies' Revenue Generation

    The core business for big companies like Alphabet (Google) will continue to generate revenue and income. Inference may be costlier than a typical search query, but big companies can find ways to sell something profitably from AI technology. Good strong leaders are crucial for big companies to succeed in producing revenue and income.

    Chapters

    1. Bank of America's Expansion and AI Investment
    2. Nvidia's Role in Self-Driving Cars and AI Opportunities
    3. Emerging Programming Languages and Specialized Models
    4. Meta's Open Source Models and Data Advantage
    5. Corporate VC Impact and Domestic AI Industry
    6. Investing in AI and Hiring Challenges
    7. AI Models and the Future of Communication
    8. Venture Capital Landscape and Investment Opportunities
    9. Big Companies' Revenue Generation and Future Focus
    10. AI Accessibility and the Future of Tech Industry
    Summary
    Transcript

    Bank of America's Expansion and AI Investment

    00:00 - 06:50

    • Bank of America expanded operations in Paris in 2019 following the Brexit vote to ensure seamless service for European clients.
    • Bank of America aimed to be a one-stop shop for clients and bring the full spectrum of services.
    • There is significant interest and investment in AI technology, with companies investing internally and investors also showing interest.
    • AI is being incorporated into various industries, including grocery stores like Kroger.
    • The challenge lies in smartly investing in AI amidst marketing hype and separating it from reality.
    • Big tech incumbents have been the winners so far, but there may be new players who find better ways to monetize AI.
    • Josh Wolf, founding partner at Lux Capital, has been investing in AI for a long time and will discuss making money in AI and identifying investment opportunities.

    Nvidia's Role in Self-Driving Cars and AI Opportunities

    06:35 - 13:07

    • Nvidia's success in the tech industry is attributed to the demand for higher video resolution and competition among gaming companies.
    • Nvidia's chips played a crucial role in the development of self-driving cars by enabling simulations and training.
    • Investors like Intel recognized the potential of AI and acquired companies like Nirvana Systems and Mosaic.
    • Evaluating AI opportunities involves considering both hardware (chips) and software (models and IP design).
    • Hype in the chip sector can lead to inflated valuations and subsequent failures of many new companies.
    • Excess GPUs from the crypto market crash are now being sold to AI researchers at lower prices.
    • The cost of building semiconductor fabs for making chips keeps increasing, impacting the cost of training foundation models like GPT-3 and GPT-4.
    • Nvidia's dominance with its CUDA programming language is vulnerable due to Facebook's Py-- language.

    Emerging Programming Languages and Specialized Models

    12:50 - 19:41

    • NVIDIA's dominant programming language, CUDA, is vulnerable due to the emergence of PyTorch and Triton.
    • PyTorch is open source and hardware agnostic, allowing developers to use different chips.
    • Triton is from OpenAI but may be less trusted compared to PyTorch.
    • Generalized models like GPT-4 have limitations due to training on public internet data that includes misinformation.
    • Specialized models in financial and healthcare sectors are expected to be smaller and more accurate.
    • Bloomberg's proprietary data set makes it a potential leader in specialized financial models.

    Meta's Open Source Models and Data Advantage

    19:20 - 26:01

    • Meta is releasing new open source models to compete with OpenAI and forming partnerships with interesting companies.
    • Access to reliable and exclusive sources of big data is crucial in AI technology.
    • Companies with large data sets, such as banks and insurers, have an advantage.
    • Hugging Face started as a chatbot but became an open source repository for models.
    • Compute and algorithms are becoming more abundant, making reliable and proprietary data the scarce resource.
    • The area of biology, specifically genetic data, has great potential for AI applications.

    Corporate VC Impact and Domestic AI Industry

    25:36 - 32:25

    • Corporate VC arms of large corporations impact the VC ecosystem in terms of making money and exiting investments.
    • Regulatory scrutiny makes it difficult for tech giants to make big acquisitions.
    • Destroying risk in early-stage companies creates value for later investors.
    • Focusing on preventing acquisitions and failures in the domestic AI industry may benefit peer adversaries like China.
    • The US needs to support domestic competitiveness in technology companies to compete globally.

    Investing in AI and Hiring Challenges

    32:01 - 38:32

    • Corporate VCs should pay a higher price and demand a lower quantum of return for taking less risk.
    • Early stage risks in companies help validate and create competition.
    • Investing in AI is highly competitive, but great founders want to work with top firms.
    • There is a five-year psychological bias in investing where people want to invest where others were five years ago.
    • Evaluating credibility of founders and their academic publications is important when assessing AI investments.
    • Hiring challenges arise when highly sought-after candidates can also raise significant amounts of funding for their own ventures.
    • A collective craze in tech funding has caused talent dispersion and misallocation of resources in the past decade.
    • The hiring data shows a spike in hiring activity related to AI algorithmic design and low-level jobs for data processing and cleaning.
    • Incumbents are currently the winners in the AI space, but small companies with novel approaches may emerge as interesting players.

    AI Models and the Future of Communication

    38:15 - 45:42

    • Communications will increasingly be generated by AI, leading to a flood of emails, texts, and tweets that are not written by humans.
    • This chat phishing phenomenon will make people yearn for in-person communications and private clubs where human interaction is guaranteed.
    • The influx of money and talent in AI may ironically bring us closer to valuing human communication more.

    Venture Capital Landscape and Investment Opportunities

    45:25 - 52:08

    • The disappearance of SoftBank and Tiger as major players in the venture capital industry has led to a more rational and scrutinizing market.
    • Down-rounds in companies, declining morale, and employees with underwater stock are some of the consequences of the disappearance of SoftBank and Tiger.
    • The Megas (giant funds) and Minos (small sub-hundred million dollar funds) have been squeezed out, resulting in a smaller base of capital available for investment.
    • LPs have pulled back, causing many funds to downsize and struggle to raise capital.
    • Investing in AI is still attractive due to the potential for big winners to emerge, but caution is advised when everyone agrees on certain companies being winners.
    • Cloudflare, a company involved in edge computing infrastructure, is seen as an interesting investment opportunity that is often overlooked.

    Big Companies' Revenue Generation and Future Focus

    51:38 - 58:48

    • CloudFlare's infrastructure is likely to succeed in the market.
    • The core business for big companies like Alphabet (Google) will continue to generate revenue and income.
    • Inference may be costlier than a typical search query, but big companies can find ways to sell something profitably from AI technology.
    • Good strong leaders are crucial for big companies to succeed in producing revenue and income.
    • Google needs to refocus on its roots and stop focusing on social issues internally.
    • YouTube is a successful acquisition for Google, generating significant revenue.
    • Ancillary product categories within big companies like Facebook and Google are being explored to generate income.
    • Google's prominence in search is likely to persist and extend into other domains, less threatened by AI advancements.
    • Meta (formerly Facebook) needs to double down on trust and embrace open-source as a way to regain trust from users and regulators.

    AI Accessibility and the Future of Tech Industry

    58:18 - 1:01:34

    • The excitement in AI is currently focused on chip makers and incumbents like Microsoft, rather than insurance companies.
    • Having accessible models like OpenAI has drawn additional interest in AI, similar to how crypto allowed normal people to participate.
    • Crypto had a learning curve with wallets and coins, but AI is powerful and easy to grasp.
    1