One Year Anniversary of ChatGPT Release: ChatGPT Growth Timeline
-
One year ago today, ChatGPT was released, reaching 1 million registered users in just 5 days, setting an internet history record. Two months later, in January 2023, ChatGPT's monthly active users reached 100 million.
The rocket-like surge in user numbers foreshadowed the rapid development of generative AI this year. Marc Andreessen, founder of a16z, proposed in 2011 that "Software is eating the world." Twelve years later, this statement might need to be changed to "AI is eating the world."
In just one year, many companies have ridden the wave of generative AI, achieving two rounds of financing and valuations exceeding $1 billion without issue. Meanwhile, other companies have been impacted by ChatGPT, experiencing declines in stock prices/valuations... and resorting to layoffs to survive.
ChatGPT's first anniversary, let's take a look at how many billions we missed in 2023.
OpenAI released ChatGPT powered by the GPT-3.5 model.
Prior to this, AI marketing content generation company Jasper and AI image generation company Stability AI completed new rounds of funding at valuations of $1.5 billion and $1 billion respectively. These stood out in the sluggish primary market of the second half of 2022, igniting American investors' enthusiasm for generative AI.
OpenAI's investment appeal skyrocketed. Well-known tech media writer Eric Newcomer wrote that "some prominent investors in Silicon Valley have been considering buying shares from OpenAI shareholders. Meanwhile, according to insiders, Microsoft has been in talks with OpenAI about the possibility of direct investment."
Microsoft has confirmed its continued investment in OpenAI with an additional $10 billion, increasing its stake to 49% and valuing OpenAI at $29 billion. Further deal details reveal that OpenAI can reclaim its shares if it generates $105 billion in returns for Microsoft and $150 billion for other investors.
The rivalry between Google and Microsoft begins to emerge.
Microsoft launched Bing Chat powered by GPT-4, a chatbot-integrated search engine, catching Google off guard. In response, Google hastily announced its Bard conversational search engine, but a demo error caused its stock to plummet by 9%, wiping out $100 billion in market value. This sparked prolonged debates about whether Google Search can maintain its dominance in the generative AI era.
Additionally, Google partnered with Anthropic, a large language model company founded by former Google and OpenAI researchers. Google not only exclusively invested $400 million but also formed a strategic alliance, subtly challenging the Microsoft+OpenAI combination.
Tech giants and startups are rushing into the field.
Bloomberg released its large-scale model BloombergGPT with 50 billion parameters; at Nvidia's launch event, Jensen Huang declared that 'the iPhone moment of AI has arrived'; Bill Gates, through a blog post, heralded 'the arrival of the AI era,' calling it a 'revolutionary, disruptive product, much like the advent of smartphones and the internet.'
The authors of the foundational paper on generative AI, 'Attention is All You Need,' have founded AI Agent company Adept AI and AI social company Character AI, each securing new rounds of funding and achieving unicorn status. Elon Musk quietly established the large-scale model company xAI.
The capabilities demonstrated by artificial intelligence are causing alarm among people.
Musk leads over 1,000 Silicon Valley entrepreneurs and scientists in calling out the dangers of AI, urging an immediate halt to all large-scale AI research.
The competition in large language models has reached a fever pitch.
In response to Microsoft's aggressive moves, Google has merged Google Brain and DeepMind into Google DeepMind to present a united front. Amazon has joined the large model wave by launching Bedrock, offering third-party model hosting services and its own large language model, Titan FM.
The 'chip shortage' issue is gradually emerging.
Elon Musk has stockpiled approximately 10,000 GPUs. Corweave, originally a blockchain mining company, has successfully pivoted to providing computing resources for AI, completing a new funding round with a valuation reaching $2 billion.
Foundation models lower application development barriers, triggering an explosion of generative AI applications.
In Y Combinator's winter batch of 268 projects, AI-related companies account for 50% of the cohort. The new AI sensation AutoGPT has emerged, amassing nearly 80,000 stars on GitHub—surpassing PyTorch (65,000)—with its AI Agent development direction drawing significant attention.
Additionally, Rewind AI, a personal historical search engine/memory assistant, became so sought-after by investors that its founder publicly shared funding details. Its trajectory and rapidly growing ARR (reaching $700,000 in months) drove investor frenzy, resulting in 170 term sheets. NEA ultimately led its new round at a $350 million valuation.
Undercurrents beneath the AI boom persist, with issues like privacy and data security. Italy's privacy regulator has temporarily banned ChatGPT, citing improper methods of collecting and storing information.
Google and Microsoft's developer conferences this month showcased fierce competition.
At the Google I/O event, AI took center stage with the unveiling of the PaLM2 language model, Vertx AI platform, one-click photo editing tool Magic Editor, office assistant Duet AI, and an upgraded Bard. Meanwhile, Microsoft's Build conference introduced Windows Copilot, integrating AI as core infrastructure into the Windows operating system.
Chip shortage drives Nvidia's earnings beyond expectations, with its stock price surging to $380 after hours, surpassing its historical high and making Nvidia the fifth company with a market capitalization exceeding one trillion dollars, following Apple, Google, Amazon, and Microsoft. Cloud providers like Microsoft, Google, and Amazon are racing to deploy AI chips and plan to increase supply in the second half of the year.
Companies offering large models and AI services to enterprises are gaining attention. AI Agent startup Adept AI's CSO & CTO (author of the Transformer paper) left to found Essential AI, helping enterprises better utilize large models, and announced funding at a $50 million valuation. Another author of the "Attention is All You Need" paper, Aidan Gomez, founded Cohere, an enterprise large model company that has become a unicorn with strategic investments from Nvidia, Oracle, and Salesforce.
AI Concerns Further Escalate—AI-generated images of an explosion near the Pentagon went viral, temporarily causing the S&P 500 to decline. Sam Altman attended a hearing to discuss AI risks and regulation with lawmakers.
Large models continue to demonstrate strong fundraising capabilities.
Chips have become the oil of the AI era.
Former GitHub CEO Nat Friedman and prominent angel investor Daniel Gross have purchased 2,512 H100 GPUs to build a $100 million server farm, prioritizing compute resources for startups backed by their AI Grant fund.
It's reported that the US Department of Commerce is preparing to update AI chip export controls implemented last October, which will impact NVIDIA and AMD. According to WSJ, the Biden administration plans to restrict Chinese companies' access to US cloud computing services, potentially affecting providers like AWS and Microsoft.
The AI competition has extended to cloud data platforms. Databricks and Snowflake have escalated tensions by scheduling their annual conferences in the same week, transforming from frenemies into major competitors in the AI arena. In this context, Databricks acquired MosaicML (an ML model deployment startup) for $1.3 billion after just 1.5 years of operation.
Major AI companies reveal their strategic focuses.
OpenAI is forming a new team led by its Chief Scientist Ilya Sutskever to develop methods for guiding and controlling "superintelligent" artificial intelligence systems. The team has access to 20% of OpenAI's computational resources.
Anthropic has released the latest version of its AI chatbot, Claude 2, and has begun offering it to a wider user base for free. Currently, Claude 2 supports file uploads and can perform tasks such as document analysis and data organization.
Musk leads 12 people to establish X.AI.
Hugging Face raised $200 million, valuing the company at $4 billion.
Meta continues to advance the open-source large model ecosystem by releasing Llama 2, which is freely available for commercial and research use. It includes three parameter variants: 7 billion, 13 billion, and 70 billion. The model was trained on 2 trillion tokens and fine-tuned using 1 million human-annotated data points.
In the generative AI era where Apple has remained silent until now, news has emerged that the company has developed an LLM framework called "Ajax" and created an "AppleGPT". Insiders claim that AppleGPT is a replication of Bard, ChatGPT, and Bing AI.
Google DeepMind has focused on embodied intelligence, spending seven months to release the robotic model RT-2, which possesses symbolic understanding, reasoning, and human recognition capabilities. This model integrates PaLI-X, PaLI, and PaLM-E.
Despite ChatGPT's traffic declining for three consecutive months, new players and directions continue to emerge from foundational models to applications.
Llion Jones, co-author of "Attention is All You Need", has left Google and co-founded Sakana AI with the former head of research at Stability AI, aiming to build the next generation of artificial intelligence.
A strong competitor to Midjourney has emerged—text-to-image company Ideogram AI has completed a $16.5 million funding round led by a16z and Index Ventures. The founding team includes contributors to renowned papers like Diffusion and Imagen.
OpenAI appears to be actively seeking solutions to its traffic decline—announcing its first-ever acquisition, Global Illumination, the developer of a popular online role-playing game. This move has sparked multiple interpretations, such as strengthening its capabilities in visual AI, expanding into the gaming sector, or training AI Agents.
In contrast to this uncertainty, OpenAI is decidedly focusing on the enterprise client segment. According to The Information, OpenAI's ARR has surpassed $1 billion, with roughly half coming from ChatGPT subscriptions and the other half from B2B clients. OpenAI has announced the launch of ChatGPT Enterprise, tailored for large corporations, offering generative AI services while ensuring data privacy and security. ChatGPT Business for small and medium-sized companies will also be released soon.
This month's keywords are open-source and computing power.
LLaMA and Falcon Open-Source Models Maintain Strong Positions. Meta announced plans to develop an open-source large language model comparable to GPT-4, with parameters several times larger than LLaMA 2. Training is expected to begin in early 2024. Meanwhile, Falcon-180B, a commercially free open-source model launched by Abu Dhabi's leading technology center Tll, outperformed Llama 2 in benchmark tests, approaching GPT-4's performance. Upon release, it immediately topped Hugging Face's leaderboard.
Compute Shortages Remain a Persistent Headache. According to an exclusive report from The Information, Meta's internal AI teams have experienced divisions due to competition for computing resources. More than half of the original 14 authors of the first-generation LLaMA model have left the company. Reuters also reported that OpenAI is exploring manufacturing its own AI chips and has even evaluated potential acquisition targets.
OpenAI still leads the development of large models, with a valuation of $90 billion. ChatGPT has undergone multimodal upgrades, introducing voice and image capabilities, while DALL·E 3 generates highly detailed images, significantly lowering the barrier to entry for users.
Google's large models are still catching up. Google has released an early version of Gemini to a select group of companies, aiming to compete with GPT-4. The model uses five times more training compute power than GPT-4, with significantly enhanced multimodal capabilities. Bard has been updated to integrate with Google's suite of apps, allowing it to fetch information from Gmail, Google Docs, Google Maps, YouTube, and other applications. Additionally, Bard now includes a model self-check feature.
As the year draws to a close, it's time for all parties to present their performance reports.
In September, ChatGPT achieved 15.6 million downloads globally on iOS and Android, generating nearly $4.6 million in revenue. However, its growth rate slowed to 20%, showing a decline compared to the previous month.
GitHub has surpassed 1 million paid users.
Microsoft's cloud computing business has exceeded growth expectations. As of September, the growth rate of Azure cloud business reached 29%. In contrast, Google's cloud business growth fell short of expectations.
AI search engine unicorn Perplexity has an ARR of $3 million, with approximately 15,000 paying users, and plans to raise funds at a *50x valuation multiple, targeting a $500 million valuation.
Embodied Intelligence and AI Agents Make Significant Progress.
Google DeepMind has open-sourced the Open X-Embodiment dataset, which includes over 500 skills and 150,000 tasks demonstrated in more than 1 million scenarios. This is expected to become a cornerstone for embodied intelligence.
Adept.ai has open-sourced its smaller version model Fuyu-8B on HuggingFace, which powers its products. Compared to other multimodal models, Fuyu-8B is easier to understand, scale, and deploy, specifically designed for AI Agent.
Imbue is one of the few companies with sufficient funding to develop foundation models, aiming to create a specialized large model for AI Agents. It has completed a $12 million financing round with participation from Amazon Alexa Fund.
AI regulation sees major updates.
The US has issued its first executive order on artificial intelligence, requiring new safety assessments, fairness and civil rights guidance, and research on AI's impact on the labor market.
2023 was the year of 'AI'. The term 'AI' was named the most notable word of 2023 by dictionary publisher Collins, with its usage quadrupling over the past year. The 2022 word of the year was 'NFT'.
OpenAI staged a corporate drama. In the first half of the month, it held its first developer conference, announcing major releases like the ChatGPT App Store, GPT-4.5, and Assistant API, leaving AI entrepreneurs on edge. In the second half, Sam Altman faced a 'backstabbing,' with 90% of employees threatening to quit. OpenAI nearly became a 'zero-cost acquisition' for Microsoft. Ultimately, Sam Altman staged a 'redemption arc,' returning as CEO and resolving the crisis.
Musk's xAI bears fruit. xAI released its first large model product, Grok. With 33 billion parameters, 8K context length, and trained in just two months, Grok-1 outperformed models like ChatGPT-3.5 and Inflection-1 in benchmark tests.