In November 2022 I published a post titled “generative AI will go mainstream when it goes from playful to useful“, and I think you will agree that this transition is in full swing. Not so long ago, we were all posting AI profile pictures on social media, but very quickly we’re now seeing an emergence of generative AI tools and LLM applications to assist almost every role in the organisation, across industries and verticals.
2023 was the year generative AI went mainstream
The pace of advancement in generative AI has been astounding. In just over one year since ChatGPT was introduced in November 2022, it’s been the fastest consumer adoption tool to cross 100 million users, OpenAI reached $1.6 billion in annualised revenue, an a myriad of new companies were born, from open source LLMs to application layer startups that were able to move fast and do much more with less.
According to Crunchbase data, generative AI startups pulled over $50 billion in funding in 2023, led by megarounds into OpenAI, Anthropic, inflection AI and others.
According to GitHub, the largest open source repository (owned by Microsoft), developers are flocking to open source generative AI projects, and some entered the top 10 most popular open source projects by contributor count in 2023.
As an example, consider the advancement of generative AI in video in 2023:
Enterprise adoption is still in the early innings
While individuals and startups were quick to adopt generative AI tools, adoption has been much slower on the enterprise side. As Viola points out, there are 5 main barriers to adoption: Security, privacy, intellectual property, performance, and cost.
With that being said, corporates have generally leaned in to the generative AI trend and are currently working out their own strategy to implement generative AI and automation into their products safely.
As Thomas Tunguz points out in his 2024 predictions, AI is no longer a standalone category but rather a component of every product. This shift is particularly significant in the realm of Large Language Models (LLMs).
Much like mobile technology became a de facto part of every startup, AI is no longer a category but the core or a component of every product. It’s still early days with LLMs, and there’s a lot of work to do; however, LLMs have already wholly transformed data in many ways, and innovations with data will continue to command VC investment. Likewise, venture dollars will still funnel into startups in the space. LLMs have driven an increased demand for data, caused a complete architecture change inside companies and changed how data is manipulated. As the technology evolves, we’ll continue to see an increase in new data products and data teams.
Theory Ventures 2024 predictions
Israel remains one of the major AI hubs globally
In terms of funding, it’s safe to say that Israeli generative AI startups have yet to live to their full potential. Overall, Israeli startups raised $7.3 billion in 2023, the lowest amount since 2018 and a 60% decrease from 2022. The judicial reform proposed by the government gave international investors jitters and international venture investment further slowed down due to the war against Hamas in Q4 2023. It’s important to put this in context, as US investments also declined by 35% and UK investments dropped by about 40%.
According to IVC online, Israeli GenAI companies raised $5.57 billion in 390 deals between 2014 and 2023. I think the numbers are a bit fuzzy, since there are companies that incorporated generative AI into their products, such as MyHerritage, which aren’t GenAI startups per-se but are included in the count.
Nevertheless, In the past 3 years, Israeli generative AI startups attracted over $2.2 billion in funding, placing Israel 3rd globally in generative AI capital raising following the US and China, based on a recent report on Generative AI startups in Israel, created by our colleagues at Viola Ventures.
This is consistent with our own findings at Remagine Ventures, published in Calcalist on September 2023, which calculated that Israeli generative AI startups raised over $2.5 billion in the last 5 years, and the number of Israeli startups more than doubled in this space more in the past 8 months.
2024 will be a sink or swim for generative AI startups
While generative AI continues to be red-hot, investors are deploying capital with caution. The fear of commoditisation, questions about copyright, impending regulation, platform dependencies, and costs remain valid concerns in 2024.
While it’s still anecdotal evidence, some of the early ‘GPT-wrappers’ application layer startups have flamed out relatively quickly, leaving investors with a bad taste. Others, may have raised funding at high valuations during peak hype, and might face difficulties raising up-rounds without significant traction. The latter is true for all startups, not just in generative AI.
In July 2023, I’ve outlined a number opportunities for Israeli GenAI startups. To a large extent, they are still valid, but the progress made by OpenAI and open source has shuffled the deck slightly and I would add the following 3:
- Tools for Open source LLM adoption – while OpenAI remains the ‘IBM’ of LLMs, the attractiveness of Open Source LLMs is on the rise. The downside of OpenAI is the cost, platform dependency and limitations of its API. Open source LLMs have been improving at a rapid pace, have an active developer community and can be an attractive alternative, if managed properly.
- Smaller, vertical models – we still measure LLMs by the number of parameters, but it’s unreasonable to expect one model to do it all. For example, a customer service oriented use case for a specific product can be done with a smaller model, trained specifically on that use case. It’s expected that smaller models will also suffer from less hallucinations. Startups enabling this while still providing security, privacy, copyright, etc might be a good candidate for enterprise adoption.
- Automation and workflow integration in niches – imagine if IFTTT was tailored for every manual, repetitive task in the company. For example, Jasper.com is a copyright tool serving marketers. GPT-4 can of course provide the copy, but the copy generation is only one of many steps in the job of a marketer. It needs to be formatted to ads, connect to hubspot, run A/B tests and choose the top performer, connect the text to relevant images, etc. OpenAI will compete in this space with its shiny new app store.
After a tough 2023, I expect 2024 to be a bumpy year for Israeli generative AI startups. But I’m generally optimistic. Companies can do more with less with generative AI, and Israel remains a major AI hub in terms of talent, capital and ideas.
For additional AI predictions and startup opportunities in 2024, I recommend the following resources:
- 10 predictions on the future of AI – Data Driven VC
- AI in 2024 by Sequoia
- Where VCs are betting in 2024
- The state of generative AI in the enterprise – Menlo VC
- What will happen in 2024 – Fred Wilson
- 10 AI predictions for 2024 – Radical Ventures
- 5 M&A observations for 2024 – Bessemer’s Jannelle Teng
- Optimistic Nihilism for 2024 and trends of consumer spending – Rex Woodbury
- Themes for 2024 – Wing Ventures
- Deloitte 2024 TMT predictions
I'm a former general partner at google ventures, head of Google for Entrepreneurs in Europe and founding head of Campus London, Google's first physical hub for startups.
I'm also the founder of Techbikers, a non-profit bringing together the startup ecosystem on cycling challenges in support of Room to Read. Since inception in 2012 we've built 11 schools and 50 libraries in the developing world.
- Weekly #FIRGUN Newsletter – November 1 2024 - November 1, 2024
- The Art of Non-Consensus Investing: Unlocking Venture Capital’s Hidden Gems - October 31, 2024
- Weekly #FIRGUN Newsletter – October 25 2024 - October 25, 2024
