The past year has seen an unprecedented AI hype wave triggered by the launch of OpenAI’s ChatGPT. Analysis abounds on whether the hype is real, where value will accrue and whether generative AI-first product builders have a real shot at category disruption or creation. As frenzied R&D and market activity continue unabated, market maps and take after take continue to drop hot.
But what about revenue? Is there evidence that AI products and services will drive significant revenue to existing companies and new startups? Early signs point to yes. We may be at the start of an AI revenue wave.
The two big early winners are obvious: NVIDIA and OpenAI. NVIDIA, whose H100 GPUs are the spice of the AI universe, announced all time record revenue, more than tripled YoY. ChatGPT itself, no doubt a top consumer of NVIDIA GPUs, has reached 100M weekly active users who are rocket fueling OpenAI’s reported $1.3B ARR, up from only $28M last year.
But an AI revenue wave is starting to form in the application layer, too, in two key different ways:
- Incumbents incorporating new AI products or product features for upsell.
- Start-ups offering new AI-first products while drawing massive investment from big tech and venture capital firms.
In the first category we see big tech and established vendors moving relatively quickly to unlock new AI features. The most popular AI delivery model is in the form of AI copilots, or LLM-powered AI assistance embedded natively in existing product experiences. Copilots promise to enhance existing applications by giving users a tireless assistant that helps answer questions, accelerate workflows, and enrich experiences.
Having blazed a trail with Github Copilot, the original fruit of its investment in OpenAI, Microsoft was early to announce its 365 Copilot for 365 applications, and recently rolled out its integrated Microsoft Copilot experience across Windows, 365 apps and Edge browsing. Notion AI made its remarkably agile debut, and AI features dropped quickly with HubSpot’s ChatSpot and Canva’s Magic tools. Not to be outdone, Salesforce flexed hard with Einstein and Adobe with Firefly, while Google followed Microsoft with Duet AI for Workspace apps. In the FinTech realm, Inuit recently announced Intuit Assist for its TurboTax, Credit Karma, QuickBooks and Mailchimp products.
On the startup side, a long tail of products applying the power of LLMs to reimagine software for creative and knowledge work has emerged. A number are attracting substantial investments from industry giants and venture capital firms:
Disruptive AI-first products like Jasper and Midjourney have taken hold for creative content use-cases, in addition to newer players commanding attention and investment like Runway and Ideogram. Hitting the LLM-powered RevStack are reimagined prospecting tools like Clay and Ignition, targeted at product marketers. Vertically integrated tools like Harvey are commanding back-to-back funding rounds and a heavy hitting waitlist.
This all sounds impressive, but does revenue follow? While most companies are keeping their revenue numbers close to their vests, we can ballpark some estimates using metrics that have been released.
The AI copilot cash register
Call them copilots, sidekicks, or uniquely branded something-or-others – LLM-powered AI assistants are making their way into all kinds of SaaS products, and vendors are preparing to cash in. Recent news details the incremental subscription revenue that players like Microsoft, Google and Salesforce stand to gain as they roll out add-on subscription pricing for access to new AI features embedded in their core productivity suites. While the street is paying attention, SaaS leaders everywhere might, too, as chatGPT-like product experiences start popping up in product after product.
Microsoft 365 Copilot: ARR, but make it plus $14B (at least)
Microsoft's trail blazing 365 Copilot integrates generative AI in the 365 application suite to give users AI superpowers when using Excel, PowerPoint, Outlook, Teams and more. Priced at an up-charge of $30 per month, 365 Copilot is also super-powering Microsoft’s top line revenue, potentially adding $13.75B in revenue within its first year alone, assuming a modest 10% adoption rate among its 382 million Office 365 subscribers. Indeed Microsoft’s latest results are making good on an upward, AI demand-driven trend, and if customers continue to feel that “once they use Copilot, they can’t imagine work without it”, even loftier projections may ring true.
Salesforce Einstein: Igniting a buying cycle, pouring on gas
Salesforce is also touting lucrative AI product add-ons, sharing its recent 11% revenue growth and other sunny, street-loving KPIs attributed to customer uptake of its AI Cloud with Einstein. By offering new AI features for an additional $50 per user per month, Salesforce can capitalize on its belief in increased demand for AI investment sparked by ChatGPT. Starting at Salesforce’s over 150,000 customers, a mere 10% adoption rate of new generative AI features, and an average of 100 seats per customer, we arrive at nearly $1B in incremental revenue. Salesforce is also rapidly evolving its conversational CRM tools with the introduction of Einstein Copilot, a more feature-rich AI assistant integrated across its CRM and app suite, along with Einstein Copilot Studio, customization capabilities that let Salesforce customers deploy AI assistance via web chat or messaging to their own customers – so they can cash in, too.
GitHub Copilot: Happiness sells
“I’d like to spend more time on repetitive tasks”, said no software developer, ever. While Microsoft hasn’t shared complete details on how well it’s going, commercially speaking, with GitHub Copilot, it has provided hard evidence of just how much developers are loving AI assistance in their IDE. GitHub Copilot for Individuals, priced at $10 per month, launched last summer clenching 400,000 users in its first month, and Copilot for Business, priced at $19 per user per month, is consistently growing quarter-over-quarter customer counts — 37,000 of them, at last count. With a reasonable assumption that Copilot for Individual subscribers have reached the low millions, plus a confirmed 1 million users among 37,000 Copilot for Business customers, it’s a clear path to $0.5B in additional annual revenue, on top of the recently achieved $1B ARR, through the GitHub Copilot extension. But wait, there’s more! That’s not including more revenue to come from GitHub’s new Copilot Enterprise service announced at GitHub Universe, to be priced at $39 per user per month when made generally available in February 2024.
Google Duet AI: Price-matched to rake it in
Dropping at an additional $30 per month per user, like Microsoft’s 365 Copilot, Google is also primed to cash in on AI assistance add-ons with its Duet AI for Google Cloud Workspace apps including Gmail, Drive, Slides, Docs, and more. It’s an immediate opportunity to upsell into an installed base of 10 million paying Google Workspace customers. With a conservative estimate of 10% uptake, that’s an additional $3.6B in annual Google Cloud Workspace revenue.
The new AI product cash cow
In the camp of brand new products that deliver a completely new, LLM-powered way of doing something, companies are also seeing dollar signs, much in the form of big checks coming from big tech and venture capital. While detractors point out hype-fueled FOMO, companies in this category are demonstrating major user demand and even revenue growth. And the “demand is elastic”, Martin Casado of a16z argues, referring to the sheer scale of automation and democratization of knowledge and creative work these products provide. Whether these disruptors eventually die off, get acquired by giants, or find their way to market dominance, today they are commanding users and dollars.
Jasper: Early wins for the AI marketing tool chest
It’s still early, and competitors abound – from incumbents embedding native AI assistance to a host of other AI writing point solutions – but Jasper has proven an early success story of new AI products taking hold. Jasper reached 100,000 paying customers and touted 2X revenue growth YoY, to $75M in 2022. Funded by its $125M Series A, Jasper offers over 50 AI writing templates including blogs, ad copy and product descriptions as well as campaign and image generation capabilities, all which can be grounded in brand and domain knowledge. Dinged slightly by news of layoffs and slowing growth, it will be interesting to see how Jasper’s traction fares as the AI wave continues to ripple through both new AI-first and incumbent marketing tools alike.
Midjourney: /Imagine bootstrapping to $200M
Dropping in stealthily just after OpenAI announced its DALL-E 2 text-to-image model, Midjourney emerged to become one of the most popular AI image generation tools and a proper member of generative AI’s first killer app club. With a small team, zero initial venture funding, and distribution exclusively through its Discord server, Midjourney quickly gained traction through viral sharing of visually stunning and surprisingly realistic images. Reportedly profitable and on track to realize $200M in revenue, today Midjourney’s Discord server has reached over 15.5M registered users. Midjourney is rapidly releasing new features including its recently released Pan option (perhaps while glancing at Adobe’s Firefly over its shoulder).
Ideogram: Let them have text
One of the nagging shortcomings of DALL-E, Midjourney and other text-to-image generation services has been the inability to include accurate, stylized text in generated images. Ideogram launched with $16M seed funding as an alternative, positioning the unique ability to include typography in generated images. Presumably in support of its mission to boost creativity with more accessible AI tools, Ideogram’s responsive web UI includes shortcuts for applying recognizable styles and use-cases like ‘painting’, ‘illustration’, ‘poster’ or ‘photo’. And it’s working: in the week following Ideogram’s late August 2023 launch, at least 90,000 users were able to generate over 3 million images.
Runway: A moving AI picture is worth $1.5B
When it comes to creative content, that generative models can make stuff up is indeed the value prop, at least for some. Among the artistic community embracing generative AI tools are Runway’s founders who met as NYU art students in 2016, mutually inspired by applying machine learning in the artistic process. Today their 5 year old company offering generative AI tools for content creation, including text-to-video generation, is reportedly valued at $1.5B following a $141M series C extension round of funding from Google, NVIDIA, and Salesforce Ventures, among others. While Runway’s revenue is comparatively small, reportedly around $7M annually, it counts logos like CBS and New Balance alongside hundreds of thousands of individual creatives as users (who are likely using its ‘free’ plan to explore). In any case, the promise of demand is clearly calling big tech to its coffers to draw Runway’s workloads close.
Harvey: AI super powering professional services, starting with legal
Rounding out the new AI-native products category (in this non-exhaustive list of examples) is Harvey, an ‘AI copilot for lawyers’. Since launching in August 2022 Harvey has landed major channels through Allen & Overy and PwC in addition to a $21M Series A led by Sequoia. Law firms interested in securely applying the magic of chatGPT to effectively automate time-consuming, text intensive legal tasks like research, contract analysis, due diligence and compliance are invited to join the 15,000 firms on Harvey’s waitlist.
Ride the AI copilot revenue wave
Hopefully this sampling of how both established tech giants and heavy hitting disruptors are rallying users and dollars around new AI product experiences is enough to convince anyone on the sidelines that it’s time to take action. While the market is still early, there are already signs that users and companies are willing to fork out real money for AI solutions that enhance existing or new applications. Our bet is that 2024 will be the year of the AI copilot that ushers in an AI revenue wave.
