Inflection AI Raises $1.3 Billion in New Funding Round

Microsoft bets big on Inflection

Inflection AI announced it has raised $1.3 billion in a new funding round led by Microsoft, NVIDIA, Reid Hoffman, Bill Gates, and Eric Schmidt. This represents another big bet by Microsoft, which invested over $13 billion in OpenAI and today is believed to have effective control of the leading generative AI foundation model creator.

Total funding for Inflection AI since its launch is $1.525 billion. The company raised $225 million in 2022, led by Greylock. According to the announcement:

Along with its partners CoreWeave and NVIDIA, Inflection AI is building the largest AI cluster in the world comprising 22,000 NVIDIA H100 Tensor Core GPUs.

…

The funding will support the company’s continued work to build and design Inflection’s first product, launched in May. A new class of AI, Pi is designed to be a kind and supportive companion offering text and voice conversations, friendly advice, and concise information in a natural, flowing style.

Personal AI Studio

Inflection describes itself as an “‘AI Studio’ specializing in creating personal AIs.” The company launched Pi, a personal chatbot assistant, earlier this year. It is designed to be emotionally intelligent and personalized to the user and is currently available for free in a beta release.

Whereas Alexa, Siri, and ChatGPT are general tools that show little differentiation in performance from user to user, Pi is positioned as a personal companion. Mustafa Suleyman, Inflection AI co-founder, and CEO, commented:

Pi is a new kind of AI, one that isn’t just smart but also has good EQ. We think of Pi as a digital companion on hand whenever you want to learn something new, when you need a sounding board to talk through a tricky moment in your day, or just pass the time with a curious and kind counterpart.

We have a lot to learn and a long way to go, but we are excited to bring this first version of Pi to people around the world. We really are at an inflection point. AI is going to be the most transformational tool of our lifetimes. What motivates me to work on Pi is that we will soon be able to provide a smart and helpful personal AI to millions of people around the world.

Foundation Model Competitor

While Inflection entered the market with the Pi chatbot, it expanded the product offering to include API access to the large language model (LLM) behind the personal assistant two weeks ago. According to Voicebot.ai’s Eric Schwartz:

Inflection AI is making the foundational large language model powering its generative AI chatbot Pi, called Inflection-1, as an API for use by other businesses. The startup is pitching its LLM as a rival to those built by Meta, OpenAI, and Google and ran tests claiming to prove that Inflection-1 generally outperforms those other models.

…

Inflection tested their LLM with the Massive Multitask Language Understanding (MMLU) benchmark for examining knowledge in 57 subjects and difficulties. The results have Inflection-1 beating some versions of other major LLMs, including Meta’s LLaMA, OpenAI’s GPT 3.5, and Google’s PaLM 540B, particularly when it comes to trivia questions.

Inflection has entered a very competitive LLM foundation model market led by the OpenAI-Microsoft partnership and Google PaLM. Nvidia NeMo is also a competitive offering, as are Antrophic, Cohere, and Hugging Face models. It is interesting to see Microsoft and NVIDIA provide more funding to a competing product, but this is now common. Google has invested heavily in Anthropic, and NVIDIA has funded Cohere.

The tech giants are all hedging their bets and want access to new technology and products that may become breakout hits. That access translates into investing in several companies, including direct competitors, to ensure they have a seat at the table as the companies grow. This approach appears to have paid off handsomely for Microsoft’s investments in OpenAI and helped it keep the leading LLM models away from its tech giant competitors.

Accessing New Revenue

In addition, Bloomberg Intelligence forecasts the generative AI market will grow to $1.3 trillion in revenue by 2032, up from $40 billion in 2022. The potential upside for investing early in the winners in this market is substantial. It may also help some tech giants protect their product dominance in several market segments.

Of course, companies such as Microsoft, Google, and NVIDIA have other reasons to invest in hot generative AI startups that are training and operating foundation models. These companies spend a lot on hardware and computing for training and inference. Microsoft wants the computing to run on Azure, and Google would like to see it done on Google Cloud. In many cases, it appears that Microsoft and Google investments are partially or wholly in-kind contributions to cloud computing services. Of course, there could be cash exchanged and then paid to the cloud hosting companies when the computing resources are consumed.

NVIDIA generally benefits no matter what cloud provider is used. Note that Inflection AI has deployed 22,000 NVIDIA N-100 Tensor Core GPUs. That might translate into as much as $800 million in hardware costs. With this kind of revenue opportunity, NVIDIA has an interest in these startups becoming successful.

AI Pedigrees

Aside from the big name investors such as Gates, Schmidt, and Hoffman, Inflection AI’s founders have notable AI industry pedigrees. Suleyman was a co-founder and the first chief product officer at DeepMind. More recently served as VP of AI product management and AI policy at Google. Karén Simonyen, Inflection AI’s chief scientist and co-founder, was a principal research scientist for seven years.

Talent, timing, technology, and funding are critical elements for any company competing for customers, partners, and mindshare in generative AI. Inflection AI may have all four at this point. The company definitely has momentum, given its announcements since May. Whether that will translate into users, customers, and revenue remains an open question. The generative AI market has hit an inflection point. I suspect Inflection AI’s inflection point may take some time to materialize.