Historians often trace the start of the California gold rush back to an entrepreneur, Samuel Brannan, walking through the streets of San Francisco shouting: “Gold! Gold! Gold from the American River!”
One hundred and seventy-six years later, engineers are shouting a similar verse: Just swap out “American River” for “machine learning Ph.D.”
Deep learning AI researchers and engineers are in higher demand than ever as corporations and startups compete for a dwindling pool of qualified talent.
Tech companies are taking drastic measures to win the AI talent wars—see Microsoft’s recent talent grab from Inflection AI and Elon Musk’s xAI poaching Tesla engineers. “Salaries are approaching, in some cases, seven digits. It’s becoming insane,” said Jonathan Userovici, a general partner at Headline.
Plenty of engineers are abandoning Big Tech for the startup world, not wanting to miss out on joining a transformative AI startup—where even a small employee-owned stake could pay extraordinary dividends.
But when it comes time for offer letter negotiations, founders are suddenly finding themselves on the back foot: Machine learning experts, aware of how in-demand their skill sets are, are asking for significantly more equity —even 3x more—than what an early employee would typically see.
“We’re seeing a lot of outrageous [job] offers because we’re in a gold rush right now,” said Rob Biederman, managing partner at Asymmetric Capital Partners, a pre-seed and seed-stage VC firm.
OpenAI in particular has a reputation for shelling out huge pay packages approaching the million-dollar mark. But the going market rate has risen across the board: The median total compensation for an AI engineer across all experience levels grew from $246,000 to $270,000 year-over-year, according to Levels.fyi, a salaries and compensation data provider.
Millionaire FOMO
Userovici, who led Headline’s investments in AI startups Mistral AI and Bioptimus, said it’s especially competitive if you’re one of the rare people with prior experience in building foundation models, the complex neural networks trained on huge datasets that perform tasks like natural language processing and image generation.
Indeed, one of the main reasons why companies like Anthropic and Mistral AI have priced their rounds so aggressively is to offer competitive equity packages. The median pre-money valuation of AI and machine learning startups more than doubled to reach $27.5 million between 2017 and 2023 and 44% of the newly minted unicorns in 2023 were in AI and machine learning, according to PitchBook data.
There’s a popular sentiment that the AI boom will mint a new generation of millionaires. And if you’re an engineer with experience developing LLMs, there’s a fear of missing out.
Alex Lim, general partner at IVP, said he’s seeing a flood of people leaving Big Tech. “A lot of the great technical AI talent is coming from places like Google and Amazon. They can offer large packages, but they can’t offer the same amount of upside as a startup can,” he said.
Of course, some foundation-layer startups won’t last: The combination of high energy bills, eye-watering GPU costs, and hefty compensation packages can rapidly burn through even extremely well-capitalized startups’ runways.
Top-tier machine learning experts know they’re a hot commodity that could make or break a startup’s success. Many are asking for large equity packages that are 2x or 3x what you might expect an early engineer to command, according to Lim.
These equity shares may still sound small as a proportion of a company—it’s still typical to offer a senior engineer less than 1% of the company’s total equity—but calculated as a proportion of a startup’s valuation, it quickly gets into the millions.
“If you join Mistral today, you join at a much lower valuation than if you join OpenAI, and so the leverage that you can get out of these shares is very big,” said Userovici.
Slicing up the pie
For founders, equity is everything. They’ll typically set aside just 10% of total equity for the employee option pool. But with such slim pickings of tier-one talent, there’s increasing pressure on early-stage founders to expand that bucket.
But broadening the employee option pool doesn’t come without sacrificing equity elsewhere. This can complicate future equity negotiations on term sheets or reduce a founder’s stake.
Parsing out shares over what might be a decade-long stint in the private markets over multiple venture-backed rounds is tricky, and VCs especially are pushing back on startups upping the budget for employee stock options.
“We still want founders that are playing for aggressive upside,” said Biederman.
Featured image by Joey Schaffer/PitchBook News
Learn more about our editorial standards.
