From AI to Absorption: Office Demand, AI Talent Concentrations, and What it Means for Data Centers
Animated publication
FROM AI TO ABSORPTION Office Demand, AI Talent Concentrations, and What it Means for Data Centers
JANUARY 2024
1
TABLE OF CONTENTS
01 02 03
04 05 06
07 08
AI COMPANY LANDSCAPE
FUTURE TRENDS & CONSIDERATIONS
KEY TAKEAWAYS
COMMERCIAL REAL ESTATE OVERVIEW
DATA CENTER OVERVIEW
METHODOLOGY
DEMOGRAPHICS & EMERGING MARKETS OVERVIEW
AI TALENT LANDSCAPE
2
Key Takeaways
3
Artificial Intelligence Sector Key Takeaways
AI Companies
Demand for Office Space
AI Talent
• AI Tenant Demand highest in SF Bay Area (2.5 msf). • Most active demand in Silicon Valley. • Austin is the market with the 2nd highest tenant demand, but it is largely concentrated with a single, large tenant requirement.
• AI talent demand is high with 5,550 unique job postings each month on average in the U.S. • California is a powerhouse for AI-related talent. Three California cities are in the top 10 MSA’s for AI talent. • AI-related degree conferrals in the U.S. have nearly tripled since 2010.
• California has the highest concentration of AI firms in the U.S., most of which are in the San Francisco Bay Area. • Spending for AI in horizontal and vertical applications is expected to increase 123% from 2022-2025. • VC funding for AI peaked in 2021 but remained level in 2023 while overall VC funding fell 38%. Horizontal platforms have been the most resilient.
Data Centers
Demographics
Future Trends
• The semiconductor market is expected to see significant growth through 2030. • Spending on data center systems, infrastructure and public cloud services is rising. • Electricity costs vary greatly across states and impact large language model (LLM) training costs. • Rack densities and cooling requirements are rising due to AI.
• Emerging AI markets have seen significantly higher percentage based growth in AI-related employment and degree conferrals than established markets. • Recent college degree holders are moving on a net basis to more affordable metros in the Sunbelt, though there remains an inbound movement to high-priced markets.
• Renewable energy production varies across states. Texas produces the most. • Data center site selection is multifaceted and considers energy, climate, water, regulations, waste and community impact.
4
Commercial Real Estate Commercial Real Estate
5
Austin has temporarily heightened demand San Francisco Bay Area Still an AI Hotspot
AI Tenants in the Market
• Silicon Valley
accounts for more than half of AI tenant demand in the Bay Area.
3,000
2,456 KSF
2,500
• Austin’s figure is anomalous, as one tenant constitutes more than half of AI tenant demand in the market. • The 410,000 sf of
2,000
1,801 KSF
1,729 KSF
1,388 KSF
1,500
Thousands SF
979 KSF
1,000
video game tenants in the Los Angeles market are not included, even though they often utilize AI in their games.
500
292 KSF
0
Greater Los Angeles
DC Metro
Greater Seattle
New York City
Austin
SF-Bay Area
Source: Cushman & Wakefield Research, 2023 Note: San Francisco includes pre-OpenAI lease total
6
The San Francisco Bay Area is most evenly split AI TIMs Composition by Market Sub-Area
AI TIMs Composition by Market
1,305 KSF
1,400
1,151 KSF
1,200
1,000
900 KSF
800
600
488 KSF
448 KSF
Thousands SF
333 KSF
400
198 KSF
200
-
San Francisco
Silicon Valley
Seattle
Puget Sound-Eastside
Washington, DC
Southern Maryland
Northern Virginia
SF-Bay Area
Greater Seattle
DC Metro
Max Target SF
Source: Cushman & Wakefield Research, 2023 Note: San Francisco includes pre-OpenAI lease total
7
The San Francisco Bay Area has the largest tenants on average, anomalies excluded. Median and Average TIM Size by Market Sub-Area
AI TIMs Composition by Market
140
131 K
129 K
120
100
80
72 K
64 K
64 K
60 K
53 K
60
50 K
Thousands SF
46 K
41 K
40 K
37 K
35 K
33 K
40
30 K
28 K
25 K
24 K
23 K
20 K
15 K
20
6 K
-
Southern Maryland
Northern Virginia DC Metro
Washington, DC Silicon Valley San Francisco Austin (With Outlier)
Austin (No Outlier)
New York City Greater Los Angeles
Puget Sound Eastside
Seattle
SF-Bay Area
Austin
Greater Seattle
Median AI TIM Average AI TIM
Source: Cushman & Wakefield Research, 2023 Note: San Francisco includes pre-OpenAI lease total
8
Class A CBD vacancy highest in San Francisco Bay Area High Vacancy Presents Opportunity for AI Tenants
Class A CBD Vacancy Rates - AI Cities
• Pandemic shutdowns lead to higher vacancy rates in most, if not all, Class A CBD office markets in the United States. • High vacancy rates give tenants: – Negotiating power – Options – Flexibility – Reduced competition
35%
30%
29.6%
25%
21.7% 22.8% 23.6% 24.2%
20%
17.6%
15%
10%
5%
– Room for growth – Potentially lower operating costs
0%
New York City
DC Metro
Southern California
SF Bay Area
Greater Seattle
Austin
Source: Cushman & Wakefield Research, 2023
9
Premium Space Availability Highest on West Coast
DC, Austin and NYC top tier space show resilience
Top Tier Non Lease-Up CBD Vacancy Rates - AI Cities
22.7% 23.6% 23.4%
25%
20%
15%
13.3% 11.8% 13.1%
10%
5%
0%
New York City
DC Metro
Southern California
SF Bay Area
Greater Seattle
Austin
Source: Cushman & Wakefield Research, 2023
10
San Francisco Bay Area and New York City down significantly Leasing Activity Still Down from Pre-Pandemic Level
Overall Leasing Activity - AI Cities
• Pandemic shutdowns lead to much lower leasing activity in most, if not all, CBD office markets in the United States. • Low leasing activity gives tenants: – Negotiating power – Options – Flexibility – Reduced pressure – Room for growth – Potential for incentives and concessions
18
16
14
12
10
8
Millions (SF)
6
5.6 MSF 5.6 MSF
4
2.2 MSF 3.0 MSF 0.7 MSF 0.6 MSF
2
0
New York City
DC Metro
Los Angeles
SF Bay Area
Greater Seattle
Austin
Source: Cushman & Wakefield Research, 2023
11
AI Talent Landscape AI Talent Landscape
12
Number of job postings below post-March 2020 average, but above pre-pandemic average AI Talent Is in High Demand but Has Cooled Recently
Unique Monthly AI Job Postings (Trailing 4-Month Average)
• Demand for AI talent began increasing in 2021 after experiencing slight downward pressure from the pandemic. • A 10-month hiring spree began in • Job postings have cooled since July 2022 and now more closely resemble early-2021 figures. September 2021 and ran until June 2022.
0 1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000 9,000 10,000
5,550
3,800
Jul 2019
Jul 2020
Jul 2021
Jul 2022
Jul 2023
Apr 2020
Apr 2021
Apr 2022
Apr 2023
Apr 2019
Oct 2019
Oct 2020
Oct 2021
Oct 2022
Jan 2019
Jun 2019
Jan 2020
Jun 2020
Jan 2021
Jun 2021
Jan 2022
Jun 2022
Jan 2023
Jun 2023
Mar 2019
Mar 2020
Mar 2021
Mar 2022
Mar 2023
Feb 2019
Feb 2020
Feb 2021
Feb 2022
Feb 2023
Nov 2019
Dec 2019
Nov 2020
Dec 2020
Nov 2021
Dec 2021
Nov 2022
Dec 2022
Aug 2019
Sep 2019
Aug 2020
Sep 2020
Aug 2021
Sep 2021
Aug 2022
Sep 2022
Aug 2023
Sep 2023
May 2019
May 2020
May 2021
May 2022
May 2023
Unique Postings (Trailing 4-Month Average)
Post-March 2020 Average
Pre-Pandemic Average
Source: Cushman & Wakefield Research, Lightcast, 2023
13
Three California MSA’s appear in the top 10 MSA’s with the most job postings Washington, DC Leads the Pack for AI Job Postings
Unique Job Posting Concentration by MSA, 2018-2023 YTD
MSA’s With the Most Unique AI Job Postings, 2018-2023 YTD
Houston, TX Detroit, MI Minneapolis, MN Denver, CO Charlotte, NC Baltimore, MD Philadelphia, PA San Diego, CA Phoenix, AZ Austin, TX Atlanta, GA Chicago, IL Seattle, WA Dallas, TX Los Angeles, CA Boston, MA San Francisco, CA New York, NY San Jose, CA Washington, DC
0 5,000 10,000 15,000 20,000 25,000 30,000 35,000
Source: Cushman & Wakefield Research, Lightcast, 2023
14
Despite hiring variances, top 5 states consistently absorb 37-40% of talent Hiring for AI-Related Occupations Trending Upward
Average Monthly Hires in AI Occupations by Year, 2018 - 2023
• California companies consistently hire more AI-related employees than companies in any other states. Hires are concentrated in San Francisco, Silicon Valley, Los Angeles and San Diego. primarily concentrated in the Texas Triangle: Dallas, Austin, Houston and San Antonio. • Texas hiring is
39.8%
40%
250,000
39.6%
39.5%
39.0%
39%
200,000
38.4%
38.1%
134,719
130,331
38%
150,000
129,258
115,729
115,088
109,196
37%
100,000
24,134 13,454 13,352 8,979
22,797 12,944 12,794 8,731
20,913 12,731 12,065 9,874
16,889 11,656 9,697 8,178
18,286 9,603 10,276 7,876
16,788 9,470 9,928 7,327
36%
50,000
29,144
28,796
28,226
27,567
24,827
24,663
35%
0
2018
2019
2020
2021
2022
2023
California Texas New York Florida Virginia All Other States Top 5 States Percent of Total
Source: Cushman & Wakefield Research, Lightcast, 2023
15
IT Consulting, hardware/software companies & consulting (broad) rank 1-3 MAMAA Companies Rank 4th in AI Job Postings
AI Job Postings by Industry Segment
AI Job Postings by Industry Sub-Segment
10,000 15,000 20,000 25,000 30,000 35,000
AI Job Postings by Industry Segment, 2018 – 2023 YTD
29.4
26.3
60,000
21.7
50,000
16.3
10.8
40,000
7.9
5.4
5.1
4.3
0 5,000
2.4 2.4 1.3
2.3 2.1
30,000
1.8
1.0 1.3
20,000
Government Banking
MAMAA
Telecom
10,000
Software
Hardware
Insurance
Recruiting
Automotive
IT Consulting
Cybersecurity
Health Insurance
0
Consulting (Broad)
Aerospace/Defense
Consulting Technology Gov't & Gov't Contractors
Financial Services
Healthcare Automotive
Artificial Intelligence
Healthcare Products
Hardware & Software Technology
Consulting
Gov't & Contractors
Fin Svcs
Health
Source: Cushman & Wakefield Research, Lightcast, 2023
16
Companies target AI employees skilled in AI, Machine Learning and Coding Software Devs and Data Scientists in High Demand
Occupations with the Most Unique AI Job Postings, 2018-2023 YTD
Most Desirable AI Skills from Job Postings, 2018-2023 YTD
Artificial Intelligence
Software Developers
Machine Learning
Data Scientists
Computer Science
Other Comp. Occupations
Python
Operations Research Analysts
Agile Methodology
Database Administrators
Java
Comp. Systems Analysts
Data Science
Web Developers
Database Architects
SQL
InfoSec Analysts
Software Engineering
Comp. and Info Rsch Scientists
Software Development
0 20 40 60 80 100 120
0 25 50 75 100 125 150 175 200 225
Thousands
Thousands
Source: Cushman & Wakefield Research, Lightcast, 2023
17
The 25 highest ranked programs are seeing a higher share of total completions than in the past Number of AI-Related Degrees Nearly Tripled from 2010
AI-Related Degree Completions, 2010-2021
Universities with most AI-related completions in 2021
80
16%
Univ. of Maryland Global Campus
14% 14% 13%
70
14%
13%
13% 13%
Univ. of Maryland-College Park
12% 12%
12%
11%
60
12%
USC
10% 11%
50
10%
Univ. of Illinois Urbana-Champaign
UC-Berkeley
40
8%
Thousands
Penn State
30
6%
UC-Irvine
20
4%
ASU
10
2%
Columbia
0
0%
UNC Chapel Hill
2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021
0 500 1,000 1,500 2,000 2,500
Total Completions Top 25 Total Completions Top 25 Percent of Total
Source: Cushman & Wakefield Research, Lightcast, 2023
18
The 25 highest ranked programs produce 3 in 10 AI-related PhD’s each year AI-Related PhD Completions Growing
Highly ranked universities with most AI-related completions in 2021 (all degree levels)
AI-Related PhD Completions, 2010-2021
34%
1,600
35%
33%
33%
32%
USC
31%
31%
30%
30%
29% 29%
1,400
28%
30%
28%
UC-Berkeley
1,200
Columbia
25%
1,000
Cornell
20%
M.I.T.
800
15%
Stanford
600
Carnegie Mellon
10%
400
UCLA
5%
200
Univ. of Chicago
0
0%
Duke
2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 PhD Completions Top 25 PhD Completions Top 25 Percent of Total
0 200 400 600 800 1,000 1,200 1,400 1,600
Source: Cushman & Wakefield Research, Lightcast, 2023
19
AI Company Landscape
20
A quarter of the world's top 100 AI Companies Are Headquartered in California Global Top 100 AI Co. HQ’s Concentrated in California
• Of the global top 100 AI companies, 65 are based across 15 U.S. states. • Cities in California with the most HQs include San Francisco (6), Palo Alto (3), Santa Clara (3), San Jose (2) and Campbell (2). • Of the nine AI companies headquartered in Texas, seven are in the Austin area.
Source: Cushman & Wakefield Research, Pitchbook, 2023
21
Horizontal platforms spending expected to grow 28% annually Vertical Platforms Expected to Grow 31% Annually
Horizontal Platforms – Total Spending ($B)
Vertical Platforms – Total Spending ($B)
$70
$70
50%
50%
44%
$60
$60
40%
40%
34%
$50
37%
32%
$50
30%
32%
30%
28%
30%
30%
$40
26%
$40
25% 28%
$30
20%
$30
20%
$20
$20
10%
10%
$10
$10
$-
0%
$-
0%
AI automation platforms*
AI core software Computer vision*Natural language technology* 2022 2025 CAGR
2022 2025 CAGR
Source: Cushman & Wakefield Research, Pitchbook, 2023
22
Autonomous machines spending expected to grow 12.3% annually AI Semiconductors Expected to Grow 19% Annually
AI Semiconductors – Total Spending ($B)
Autonomous Machines – Total Spending ($B)
45%
45%
$60
$60
40%
39%
40%
40%
$50
$50
35%
35%
30%
30%
$40
$40
25%
25%
22%
$30
$30
20%
20%
15%
15%
15%
15%
$20
$20
10%
10%
5%
$10
$10
5%
5%
0%
0%
$-
$-
GPU
FPGA Microcontroller
ASIC
Autonomous Vehicles
Intelligent Robotics & Drones
2022 2025 CAGR
2022 2025 CAGR
Source: Cushman & Wakefield Research, Pitchbook, 2023
23
Deal value and deal counts down after historic 2021 peak as high interest rates & economic uncertainty set in Venture Capital Activity Lower After 2021 High
Semiconductor Venture Capital Deal Activity
Autonomous Machines Venture Capital Deal Activity
500
500
$18,000
$18,000
458
$16,000
$16,000
416
400
400
$14,000
$14,000
386
341
340
316
311
$12,000
$12,000
316
284
300
300
$10,000
$10,000
249
271
$8,000
$8,000
188
200
200
175
$6,000
$6,000
129
104
$4,000
$4,000
76
75 100
67 100
60
52
$2,000
$2,000
23 27
11 22
$0
$0
2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 23Q1 Deal value ($M) Deal count (R)
2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 23Q1 Deal value ($M) Deal count (R)
Source: Cushman & Wakefield Research, Pitchbook, 2023
24
Data Centers
25
Semiconductors
Key Points
• Semiconductors address two crucial phases of the AI lifecycle: training and inference. In training, AI chips process vast amounts of existing data and perform complex calculations. In the inference phase, AI chips are used to make inferences on the data on which they’ve been trained. • Speed, efficiency and energy usage are top concerns for AI chip manufacturers. • The leading AI chip makers, by performance, are Nvidia, AMD, Intel and, most recently, Amazon AWS.
26
Overall semiconductor market grew 31% from 2020-2022 Long-Term Semiconductor Market Growth Expected
Semiconductor market size worldwide 2020-2030, by application
• By 2030, the leading application within the global semiconductor industry is expected to be servers, data centers and storage. • In 2020, the servers, data center and
1,200
1,000
249
82
800
160
136
600
79 93 93 62
100
149
71 63 73 53
storage market was valued at $76 billion, and it is expected to increase by 227% by 2030.
50 40 51 38 76
400
114
131
124
200
115
117 100
213
150
144
0 Worldwide Market Revenue (in billion USD)
2020
2022
2025*
2030*
Smartphone Automotive
Personal computing Industrial electronics
Consumer electronics
Wired and wireless infrastructure
Servers, data centers, and storage
Note: * denotes forecasted figure
Source: Cushman & Wakefield Research, ASML, 2023
27
Idaho, Utah and Nebraska offer commercial power under 9¢ Least Expensive States for Commercial Power
15 Cheapest States by Avg. Commercial Electricity Cost
12
11
10.41¢
10.25¢ 10.31¢
9.89¢
9.85¢
9.66¢
10
9.36¢
9.36¢
9.27¢
9.05¢
8.89¢
9.59¢
9.52¢
9
8.45¢
8.32¢
8.78¢
8
8.39¢
8.14¢
8.10¢
7
7.43¢
7.32¢
7.29¢
7.12¢
7.01¢
6.97¢
6.94¢
6.92¢
6.76¢
6.57¢
6
Average Price of Electricity (¢/kWh)
6.12¢
5
Idaho
Utah Nebraska North Dakota
North Carolina
Oregon Texas Wyoming Washington Virginia Missouri
Nevada South Dakota
Arkansas Oklahoma
Commercial (¢/kWh)
Industrial (¢/kWh)
Source: Cushman & Wakefield Research, FindEnergy.com, 2023
28
Open AI’s GPT-3 used more power than other leading models to train Large Language Models Use a lot of Power to Train
Energy consumption when training LLMs in 2022 (in MWh)
• Energy consumption in training is high; frequent retraining is required to maintain data relevance. Lifetime energy consumption is even higher than initial training usage. • Energy savings from
1,400
1,200
1,000
800
600
AI will be big. Mobile phone operators alone expect AI to reduce power consumption by 10-15%.
MWh Used in Training
400
200
0
GPT-3
Gopher
Bloom
OPT
Note: San Francisco includes pre-OpenAI lease total
Source: Cushman & Wakefield Research, Cornell University, 2023
29
Ranges from $107k to $530k Hypothetical Electricity Cost to Train GPT-3 by State
• GPT-3 is estimated to have used 1,287 megawatt hours of electricity during training. • The estimated cost to train GPT-3 varies by state based on average commercial electricity rates. • Training is not latency dependent, LLM operators can harvest significant savings in low-cost energy states.
Source: Cushman & Wakefield Research, FindEnergy.com, Cornell University, 2023
30
Hyperscaler and colocation providers offer greater PUE than those running older technology Power Usage Effectiveness Gaining Greater Efficiency
Innovations Most Likely To Deliver Better Data Center Efficiency Worldwide, 2022-2027
Data Center Average Annual PUE Worldwide
2.7
Software-defined power
50%
Artificial intelligence
45%
2.5
2.5
Multisite resiliency
38%
2.3
Fuel cells for primary power generation Transactive relationship w/ utility provider Direct liquid cooling
36%
23%
2.1
1.98
17%
1.9
Multiday battery storage
15%
Heat rejection into water
10%
1.67
1.65
1.7
1.59 1.57
Power Usage Effectiveness (PUE) Ratio
1.58
Metaverses
8%
1.55
Other
5%
1.5
0% 10% 20% 30% 40% 50% 60%
Source: Cushman & Wakefield Research, Uptime Institute, 2023 Note (rhs): Survey of 744 IT and Data Center managers worldwide in 2022
31
Data center developers have seen density and cooling expectations rise Rack Densities Increasing with Compute Requirements
New Project Est. Rack Density Ranges (kW per Rack)
• Both CPU and GPU oriented data centers are requiring higher densities • Cooling requirements
140
120
100
will likewise need to increase as higher densities lead to higher temperatures. Expect liquid-to-chip cooling to become more commonplace, with many current plans designed to be
80
60
40
20
flexible between different cooling regimes.
0
2021
2022
2023
2024
2025
Source: Cushman & Wakefield Research, estimates based on discussions with data center developers and operators
32
Demographics & Emerging Markets
33
AI-Related jobs grew by 33% on average in the markets outlined below Established AI Hubs Seeing Long-Term Job Growth
AI-Related Jobs & Growth In Established Markets
450
120%
104%
400
100%
350
300
80%
65%
65%
250
60%
51%
200
Thousands
150
40%
28%
25%
21%
100
17%
15%
20%
11%
50
0
0%
New York City Washington, DC Los Angeles San Francisco
Seattle San Jose/Silicon Valley
Chicago
Boston
Philadelphia
Austin
2012 Jobs 2023 Jobs 2012 - 2023 % Change
Source: Cushman & Wakefield Research, Lightcast, 2023
34
Emerging Markets Primarily Located in The Sunbelt
With the exception of Portland and Salt Lake City
AI-Related Jobs & Growth In Emerging Markets
94%
250
100%
84%
90%
81%
200
80%
66%
70%
63%
61%
57%
150
60%
51%
49%
45%
50%
Thousands
100
40%
30%
30%
50
20%
10%
0
0%
Dallas/Fort Worth
Atlanta Denver/Boulder
Phoenix
Houston
Miami
Raleigh/Durham Salt Lake City/Provo
Portland
Charlotte
Nashville
2012 Jobs 2023 Jobs 2012 - 2023 % Change
Source: Cushman & Wakefield Research, Lightcast, 2023
35
All emerging markets have seen significant growth in AI-related degree completions Emerging Markets Typically Investing In Education
AI-Related Degree Completions in Emerging Markets, 2012 - 2021
1,800
600%
564%
1,600
500%
1,400
428%
400%
1,200
1,000
321%
272%
300%
294%
800
164%
205%
200%
600
183%
132%
400
65%
100%
200
0%
0
Dallas/Fort Worth Atlanta
Denver/Boulder
Phoenix
Houston Raleigh/Durham Salt Lake City/Provo
Portland
Charlotte
Nashville
All Completions - 2012 All Completions - 2021 % Change in Completions
Source: Cushman & Wakefield Research, Lightcast, 2023
36
Emerging Markets Offer Model Training Savings
Estimated Electric Cost to Train GPT-3 in Emerging Market States
180
18.00
160
16.00
140
14.00
12.98¢
12.74¢
12.20¢
11.75¢
120
12.00
10.82¢
9.36¢
9.36¢
9.27¢
100
10.00
8.45¢
80
8.00
60
6.00
40
4.00
Average Cost of Electricity per KWh (¢)
20
2.00
Estimated Electric Cost to Train GPT-3 ($000's)
$164K
$157K
$151K
$139K
$120K
$120K
$119K
$109K
$167K
-
-
Georgia
Tennessee
Colorado
Arizona
Oregon
Texas
North Carolina
Utah
U.S. Average
Total Electricity Cost to Train GPT-3
Commercial (¢/kWh)
Source: Cushman & Wakefield Research, Statista, FindEnergy, 2023
37
Future Trends & Considerations
38
Washington tops the chart for large production states Top 10 States by Renewable Production Mix
Top 10 States by Percent of Renewable Energy Production
• The majority of Washington’s
250
120%
renewable energy production comes from conventional hydroelectric production. • Solar, conventional
99.6%
100%
200
83%
80%
72% 71%
67% 65%
150
62%
60%
50%
hydroelectric, and wind account are major sources of renewable electricity production in California.
46% 45%
100
40%
50
20%
Energy Production - Millions (KWh)
Renewable Energy Percent of Total
0
0%
Vermont
South Dakota
Washington Idaho
Maine
Oregon
Iowa
Montana Kansas California
Renewable Production (MWh)
Non-Renewable Production (MWh)
Renewable % of Total
Source: Cushman & Wakefield Research, FindEnergy.com, 2023
39
It isn’t even close Texas is #1 in Total Renewable Energy Production
Top 10 States by Total Renewable Energy Production
• Renewable production in Texas is higher than the total production in 44 U.S. states combined. • Wind and solar are the prominent renewable energy sources in The Lone Star State. • NRG, Austin Energy, and CPS Energy’s South Texas Project produces more power than the total of 10 other states.
600
80%
72%
65%
70%
62%
500
60%
400
46%
45%
50%
44%
300
35%
40%
29%
27%
30%
200
20%
14%
13%
Energy Production - Millions (KWh)
Renewable Energy Percent of Total
100
10%
0
0%
Texas California Washington Iowa Oregon Oklahoma New York Kansas Illinois Colorado North Carolina
Renewable Production (MWh)
Non-Renewable Production (MWh)
Renewable % of Total
Source: Cushman & Wakefield Research, FindEnergy.com, 2023
40
Environmental Considerations in Data Center Site Selection
Water Availability & Conservation
Energy Efficiency & Renewables
Climate Resilience
• Prioritize locations with access to reliable and affordable renewable energy sources. • Look for regions with a robust grid infrastructure and proximity
• Assess susceptibility to natural disasters and select sites with lower risk profiles. • Consider climate change projections to ensure long-term viability and resilience.
• Evaluate water availability and potential restrictions, especially in arid or drought-prone areas. • Implement water-efficient cooling systems and explore reuse/recycling options.
to renewable energy generation facilities.
Regulatory Compliance & Sustainability Standards
Community Engagement & Stakeholder Relations
Waste Management & Recycling
• Ensure compliance with local, state and federal environmental regulations and codes. • Seek locations with established sustainability standards to align with industry best practices.
• Seek areas with robust waste management infrastructure and recycling programs to minimize environmental impact. • Consider opportunities for repurposing or recycling data center components and materials.
• Engage local communities to address concerns and demonstrate a commitment to environmental responsibility. • Consider partnerships with local organizations or initiatives focused on sustainability and environmental conservation.
41
Methodology
42
Glossary of Key Terms
Term
Definition
Term
Definition
AI
Artificial Intelligence
KSF
Thousand Square Feet
An integrated circuit that is dedicated to training and executing neural networks
AI Chips
kWh
Kilowatt-hour
MSA
Metropolitan Statistical Area
A free large language model for large-scale public access created by AI researchers from HuggingFace, Microsoft, Nvidia, IDRIS/GENCI and PyTorch
Bloom
MSF
Million Square Feet
MWh
Megawatt-hour
CBD
Central Business District
Open Pre-Trained Transformer developed by Meta for natural language processing researchers
OPT
A measure of computational resources required for AI systems to perform tasks Synthetic media that’s been digitally manipulated to replace one person's likeness with that of another A 280 billion parameter large language model developed by Google subsidiary DeepMind Generative Pre-Trained Transformer 3; a large language model released by OpenAI in 2020
Compute
Power Usage Effectiveness: ratio measuring energy used by a data center to the energy delivered to computing equipment in the data center
Deepfake
PUE
Gopher
TIMs
Tenants in the Market
GPT-3
Companies involved in vertical applications, horizontal platforms, autonomous machines or AI & ML semiconductors, as rated by Pitchbook.
AI Company
GPU
Graphics Processing Unit
YTD
Year-to-date
IT
Information Technology
43
Occupations & Universities included in survey results Job Postings Parameters
Occupations Analyzed
Highest Ranked AI-Related Universities
• Computer And Information Research Scientists • Computer And Information Systems Managers • Computer Hardware Engineers • Computer Network Architects • Computer Occupations, All Other • Computer Programmers • Computer Systems Analysts • Computer User Support Specialists • Database Administrators • Database Architects • Data Scientists • Information Security Analysts • Network And Computer Systems Administrators • Operations Research Analysts
• University of
• California Institute Of Technology • University Of California-Berkeley • University Of California-Los Angeles • Harvey Mudd College • University Of Southern California • Yale • Georgia Institute of Technology-main Campus • University of Chicago • Northwestern University • Harvard • M.I.T. • Northeastern University
Michigan-Ann Arbor • Dartmouth College • Princeton • Columbia University • Cornell • Duke • Carnegie Mellon • University of Pennsylvania • Brown University • Vanderbilt University • Rice University • The University of Texas at Austin • Stanford
44
John McWilliams Senior Research Analyst john.mcwilliams@cushwake.com
Kevin Waldman Executive Director kevin.waldman@cushwake.com
Jacob Albers Head of Alternatives Insights jacob.albers@cushwake.com
David C. Smith Head of Americas Insights david.smith4@cushwake.com
About Cushman & Wakefield Cushman & Wakefield (NYSE: CWK) is a leading global real estate services firm that delivers exceptional value for real estate occupiers and owners. Cushman & Wakefield is among the largest real estate services firms with approximately 52,000 employees in over 400 offices and approximately 60 countries. In 2022, the firm had revenue of $10.1 billion across core services of property, facilities and project management, leasing, capital markets, and valuation and other services. To learn more, visit www.cushmanwakefield.com or follow @CushWake on Twitter.
©2023 Cushman & Wakefield. All rights reserved. The information contained within this report is gathered from multiple sources believed to be reliable. The information may contain errors or omissions and is presented without any warranty or representations as to its accuracy.
45
Made with FlippingBook flipbook maker