Generative AI - A Strategy Blueprint for Capital-Efficient Entrepreneurship
Sramana Mitra
Founder and CEO of One Million by the One Million (1Mby1M) Global Virtual Accelerator
I publish this series to discuss the nuances of bootstrapped entrepreneurship. Please subscribe to my Best of Bootstrapping series to never miss an article and?learn what to expect from 1Mby1M.
Ever since ChatGPT was made available to the general public, everybody is talking about how it will impact the world as we know it.?
ChatGPT is a generative pre-trained transformer (GPT) built on top of OpenAI’s GPT-3 family of large language models (LLM). Generative AI is trained on vast amounts of data that enable it to understand and respond like a human. It helps create new content from previously created content. It can write and debug computer programs, compose music, plays, and student essays, and answer test questions.?
OpenAI’s Journey and Financials
OpenAI was originally founded in 2015 with a stated goal of promoting and developing friendly AI in a way that benefits humanity as a whole. Sam Altman, Elon Musk, Greg Brockman, Reid Hoffman, Jessica Livingston, Peter Thiel, AWS, Infosys, and YC Research pledged over $1 billion to the venture.?
In 2019, OpenAI entered into a $1 billion deal with Microsoft. For using Azure Cloud Platform, it would give Microsoft the first opportunity to commercially leverage early results from OpenAI’s research. Microsoft has invested an additional $2 billion and also become a key backer of OpenAI’s Startup Fund, OpenAI’s AI-focused venture and tech incubator program.
OpenAI charges developers licensing its technology about $0.0004 to $0.02 to generate 750 words of text and about $0.016 to $0.020 to create an image from a written prompt. It expects revenue of $200 million in 2023 and $1 billion by 2024. It has been valued at $20 billion in a secondary share sale.
OpenAI’s Competitor Landscape – Bard, Cohere, Stability AI
Within a month of its launch, ChatGPT had answered queries for over 1 million users. This raised questions about the threat to Google’s dominance in the $200 billion online search market. Google has recently introduced Bard, a conversational AI service powered by Language Model for Dialogue Applications (LaMDA). It expects to soon have AI-powered features in Search to process multiple perspectives into easy-to-digest formats.?
Google will be releasing Bard initially with a lightweight model to make sure Bard’s responses meet a high bar for quality, safety and is grounded in real-world information. Next month, it will start onboarding individual developers, creators, and enterprises so they can try its Generative Language API.
There are several lesser-known startups working on varied use cases. Cohere, which has a Google Cloud Partnership, plans to introduce a new dialogue model to let enterprise users generate text and engage with the model to refine the output. It has raised $175 million so far from investors including Index Ventures, Tiger Global and AI luminaries Geoffrey Hinton, Fei-Fei Li, and Pieter Abbeel. It is looking to raise more funds at a valuation of $6 billion. I like Cohere for its B2B strategy and its stated focus on supporting developers who want to piggyback and build applications on their platform.
领英推荐
Piggybacking is a capital efficient strategy of entrepreneurship. Generative AI is going to be a great platform to piggyback on and build your startup in a capital efficient manner. Learn how you can build startups by piggybacking on cutting edge technology AI platforms by taking this Udemy course on Bootstrapping by Piggybacking.
Stability AI released its open source text-to-image AI generator called Stable Diffusion last year. Stable Diffusion is the main competitor for Open AI’s Dall-E, a text-to-image AI program. In October 2022, Stability raised $101 million in a seed round led by Coatue Management and Lightspeed Venture Partners at a $1 billion valuation.
The first step to building a successful AI startup is to stay on top of this fast-paced industry and build an effective AI application around your own domain knowledge. One way to go about learning how to do this would be to take our How to Build an AI Startup course on Udemy.?
Global investment in AI surged from $12.75 billion in 2015 to $93.5 billion in 2021. The AI market is expected to grow at 39.4% CAGR to $422.37 billion by 2028. This decade, clearly, is going to be about AI, especially LML (large language models) and Generative AI platforms.
However, the AI platform vendors realize that the best way for them to get to accurate, actionable applications is by opening up to developers who bring specific domain expertise to the table, train their models on domain-specific data sets, and build domain-specific workflows. This is why Platform-as-a-Service (PaaS) will become the standard operating procedure of this industry. Developer ecosystems piggybacking on platforms will become the norm.
If you are a developer / entrepreneur playing this game, please beef up your knowledge of how to bootstrap by piggybacking and how to leverage domain knowledge to build AI startups.
Looking For Some Hands-On Advice?
For entrepreneurs who want to discuss their specific businesses with me, I’m very happy to assess your situation during my free online?1Mby1M Roundtables, held almost every week. You can also check out our free?Bootstrapping Course, our?Udemy courses,?YouTube channel,?podcast interviews?with VCs and Founders, and, to?follow my writings, click "Follow" from?here.
Founder and CEO of One Million by the One Million (1Mby1M) Global Virtual Accelerator
1 年Santanu Bhattacharya you may want to have a look at the comments
Founder and CEO of One Million by the One Million (1Mby1M) Global Virtual Accelerator
1 年> Even with paid API, would a company make a bet for it to generate 100% accurate search results (which at its heart the ChatGPT is all about). The answer is No. I myself have been part of some such conversation by virtue of my industry interaction. Say you have a $100 million contract, would you or any CEO/CIO/CTO risk it? Simple answer No. SM: You are missing a layer. A PaaS on top of ChatGPT would require a training tool for the model within a certain domain. Without the ability to train on vocabulary and workflow, a PaaS strategy will not work.
Founder and CEO of One Million by the One Million (1Mby1M) Global Virtual Accelerator
1 年> Why then the hype around ChatGPT or even more around Midjourney or DALL-E2 is that they bought the power or sense of power and amazing wonder out of Academic world to hands of common man who can do these amazing use cases which seems like magic, as it been productized and a simple interface and use case has been given to common man to play with which reduces human effort! Again not a PaaS. Not something I am sure enterprises can use right now, with safety. SM: The hype is because this is the first time consumers have been given the "toy" to engage with AI. However imperfect and useless, it IS a new "toy".
Founder and CEO of One Million by the One Million (1Mby1M) Global Virtual Accelerator
1 年> What is working and will work what IBM has done with Watson which is a mainstay for almost most AI development in standard SI companies be likes of Top IT services companies like TCS, Infy or Wipro or Accenture or HCL, etc which provides services across top fortune 500 companies Or even Google AI APIs which is used by Developers to build AI apps And those are PaaS. ChatGPT unfortunately is a tool and almost a system if you want to even call that, which a AI expert would even call it a GUI for their LLM GPT3 model, a clever one SM: I didn't say that ChatGPT is PaaS. Cohere is going in the PaaS direction. My prediction is that ChatGPT will also end up going there, as will the other efforts in this field.
Founder and CEO of One Million by the One Million (1Mby1M) Global Virtual Accelerator
1 年>And I am sure Microsoft now effectively owns it (You of course know Elon Musk left this org long time back) it will have the proprietary side to it own product integration Yes it's used it for CodePilot a brilliant tool but MS tool There is no particular problem with Microsoft owning a PaaS. They already own Azure. Amazon owns AWS. Salesforce owns one. Google owns one. Strongly backed PaaS platforms enable a lot of capital-efficient entrepreneurship, and the business models are well known. Often they entail 30% platform royalty. I suggest you go do my Bootstrapping by Piggybacking courses ASAP before challenging my premises and offering to advise me. Your confidence in your own knowledge is a bit overrated.