The global adoption of AI is occurring at record speed. No one wants to miss the potential benefits that AI might bring to an organization, so companies are channeling significant resources into this sector at rapid speeds. According to a Goldman Sachs report, companies are estimated to spend $1 trillion on capex in the coming years including investments in infrastructure (power grid, data centers), hardware (chips, equipment), and software (Goldman Sachs, “Top of Mind,” Issue 129, June 25, 2024).
The impact of AI on the U.S. economy is expected to be profound, driving productivity, innovation, job creation, and competitiveness. According to a PwC study, AI will contribute $3.7 trillion to the North American economy in 2030, equating to 14.5 percent of GDP in 2030. China is expected to record the largest increase in both relative and absolute terms with an increase of $7.0 trillion during this same period or 26.1 percent of GDP. Of these global gains, 40 percent will come from improved productivity and 60 percent from product enhancement (PwC, “Sizing the Prize”), mainly in the banking, technology, health care, and consumer experience sectors.
Race for Talent
However, these growing AI companies are facing a shortage of skilled and qualified workers right now. The race is on for experienced data scientists, developers, and programmers. According to a recent McKinsey Technology Trends Outlook report, job postings for specific Gen-AI talent jumped 111 percent between 2022 and 2023, and these companies are prioritizing the recruitment of highly skilled individuals, including PhDs and data scientists, and are prepared to invest in competitive wage packages. (McKinsey Digital, “Technology Trends Outlook,” July 2024)
Tech clusters driving AI innovation in the Bay Area
The Bay Area is an interconnected and supportive environment, otherwise known as economies of agglomeration, where various elements—including talent, technology, resources, and infrastructure—come together to drive innovation. This ecosystem fosters innovation by providing the knowledge, tools, capital, access to information, and processing capacity that encourages creativity, risk-taking, and opportunities for start-ups in general, and AI start-ups specifically.
Converging forces: AI and venture capital in the Bay Area providing for both stability and a projected longer recovery
The convergence of massive AI investments and the region’s dominance in venture capital is fueling unprecedented growth in the AI industry. Global funding for AI firms is projected to surge 45 percent to more than $115 billion by the end of 2024, as investors scramble to find the next OpenAI. The U.S. has consistently dominated this market, capturing an average of 60 percent of all global AI investment in the past decade. AI funding accounted for 36 percent of all VC investment in the broader U.S., with the Bay Area alone representing 23 percent of total U.S. VC funding across all sectors. A total of $43.1 billion in VC capital was invested in the Bay Area in the first eight months of 2024, with roughly 60 percent, or $26.8 billion, targeted at AI companies directly.
This has led to AI companies being one of the key drivers of demand for commercial real estate in the Bay Area in recent years. In San Francisco alone, AI companies accounted for 20 percent of all leases during the past 18 months and together with other tech sectors, accounted for nearly 44 percent of all leasing activity. Traditional office using companies, on the other hand, accounted for 38 percent of all leases signed. In Silicon Valley, we tracked more than nine percent of all leases were from AI companies in both office and industrial properties. In office alone, AI accounted for more than 10 percent of leases whereas in industrial that figure ticked down to eight percent.
Corporate venture capital lowers tenant risk profiles
Another key trend in the investment cycle is the role of Corporate Venture Capital’s (CVC) deeper involvement in startups, from inception to exit, more directly benefiting commercial real estate. Beyond funding, CVCs actively manage growth, mitigate risks, and provide crucial support like mentorship, market access, and parent company resources. Start-ups are now seen as A-plus creditworthy tenants, with global parent companies acting as “co-signers,” reducing their risk profile and stabilizing Bay Area markets. Amazon’s $4 billion investment in Anthropic highlights this trend. Boosting growth, integrating AI technologies, and leveraging AWS for cloud infrastructure and for real estate reduces the perception of risk of Anthropic.
AI companies raising capital but delaying deployment
Hidden in plain sight, many of these AI companies are still in their infancy and are not showing up yet in the funding data. They are quietly building their futures, bootstrapping their operations from a home office, co-working space, or a virtual office.
Compared to previous tech bubbles, AI companies are taking a more measured approach to growth. Instead of rapid expansion, early-stage startups are scaling deliberately by hiring slowly, opting for shorter leases, and utilizing the glut of available sublease spaces. These strategies have both positive and negative implications for the market.
Potential benefits:
- Stabilized growth: AI companies are taking longer to commit to office space and are gradually expanding their footprints, which may help them avoid the large fluctuations in both rapid expansion and subsequent downsizings recorded in previous cycles.
- Leasing sublease space: 61 percent of AI leases signed since the start of 2023 in San Francisco are in sublease space; in Silicon Valley, nearly 25 percent of all leases were in sublease space. These spaces are already built out, so landlords and tenants can keep costs down; plus they help to reduce available space in the market, which is especially beneficial in San Francisco and other urban markets where oversupply has been an issue.
- Quality over quantity: AI companies are increasingly preferring highly amenitized office spaces, with robust IT infrastructure, high-speed connectivity, flexible workspaces, and additional conference rooms to enhance team productivity.
But this slower growth strategy also poses a challenge for the market:
- Slower leasing activity/longer absorption periods: Tenants are taking longer to expand, which could result in prolonged periods of vacancy that will negatively impact overall market performance and value.
- Committing to shorter lease terms: Signing up for only three-year terms versus the average of 5-plus years for non-AI companies creates stability that benefits the markets in the longer term; in the immediate future, however, it leaves vacancies in the market. This environment also creates challenges affecting the operator’s ability to operate their buildings and manage their assets effectively.
Growth trajectory of Bay Area AI companies
We dug into our data to understand what the trajectory of these AI companies looks like today. We found that companies in the “Early Stage” of development (accelerator/ incubator, seed round, and early-stage VC), raised an average of $30 million, employed 82 people, have been in business six years, and leased only 6,400 square feet (595 sq m) of space on average. As these companies mature into the growth phase, they raised an average of $165 million, employ 243 people, and have committed to 9,700 square feet (901 sq m). There was a notable jump by the time these companies hit the “Late Stage,” with $540 million in total raised, 750 employees, in business 17 years on average, and committed to 40,000 square feet (3,716 sq m). As these companies mature, they are expanding their footprints and have greater leasing requirements, which bodes well for owners in the medium to longer term.
Case study: then versus now
In addition to looking at market averages and general trends, we wanted to dig deeper into two specific AI companies. As an example of last cycle’s exuberance, Company A founded in 2013 started with $1.6 million of seed funding, raised $8 million in Series A, and then signed a 12,400-square-foot (1,152 sq m) lease with 11 employees, at that point having 1,130 square feet (105 sq m) per employee. Whereas in this cycle, Company B founded in 2022 raised $7 million during its seed round and then $26 million during Series A, committed to only 5,400 square feet (502 sq m) with 17 employees, putting the average square foot per employee at 318 square feet (29 sq m). Through its latest funding round (Series C), Company A raised a total of $76 million, ramped up staff to 127 people, and expanded into 25,000 square feet (2,323 sq m). Their square footage per employee ticked down to 200 square feet (18.6 sq m). In its second and latest round, Company B raised a total of $83 million and recently signed a 11,800-square-foot (1,096 sq m) lease, equating to 140 square feet (13 sq m) per employee. In looking at total money raised relative to the size of the latest lease ($/SF), Company A raised $3,000 for every square foot leased whereas Company B raised $7,000 for every square foot, considerably more moderate.
What does this all mean?
AI companies today are operating more conservatively in terms of office space usage relative to their funding and headcount, as illustrated by the comparison between Company A from the previous cycle and Company B from the current cycle. While Company B has raised more money overall, it has committed to significantly less office space compared to the aggressive expansion of Company A. This reflects AI companies’ shift toward maximizing resource efficiency in the current business cycle. While this may translate to a slower absorption of space, and thus a slower market recovery, it also indicates a more sustainable approach to growth. As companies consider their real estate and resource allocation more thoroughly, they help ensure that the region is better prepared for potential challenges ahead. In the event of the next economic downturn, this groundwork could prove invaluable, as a more resilient local economy may weather the storm more effectively, supporting jobs and maintaining consumer confidence even in difficult times. This cautious growth strategy bodes well for landlords and office returns in the medium to longer term.
AI is still a young industry and most of these companies are all very lean. However, as these AI firms mature and navigate the complexities of a competitive market, will they ultimately adopt a more traditional company mindset? This shift could materialize in various ways, particularly through the hiring of larger marketing, sales, and HR teams, and thus increase the need for more real estate. Or will the computing power of AI enable a transformation of conventional business models into more efficient structures that rely less on human labor and instead harness the capabilities of the digital world? It remains to be seen whether these AI companies will evolve to resemble traditional businesses over the next few years or if the power of their product will enable them to further change the real estate leasing paradigm.
