Small print.
Wide open.

Table of Contents

infrastructure initiatives requires us to recruit personnel with relevant experience and manage complex multi-
year deployment timelines and relationships with developers, contractors, and service providers.

As  we  continue  to  build  out  our  own  compute  and  infrastructure  and  further  vertically  integrate  across  the 
compute  supply  chain,  these  multi-party  structures  will  become  more  prevalent  and  will  expose  us  to  a 
structural  timing  and  liquidity  risk  in  our  payment  obligations  that  differs  from  our  historical  compute 
arrangements.  Historically,  we  have  generally  obtained  compute  from  a  limited  number  of  CSPs  under 
arrangements in which our compute and infrastructure expenses are paid as compute becomes available to us. 
In  the  future,  we  expect  we  will  increasingly  enter  into  separate  arrangements  with  a  greater  number  of 
different  counterparties,  each  responsible  for  discrete  services  or  inputs  across  the  compute  supply  chain. 
Under  many  of  these  arrangements,  we  may  be  obligated  to  pay  a  counterparty  once  it  performs  its  discrete 
obligation, even if delays or failures by others prevent us from assembling those discrete services or inputs into 
usable  compute  we  can  deploy  to  generate  revenue.  In  addition,  counterparties  may  deliver  their  services  or 
inputs later than scheduled, or may fail to deliver them at all, and any such delay or non-delivery by a single 
counterparty could postpone or prevent our deployment of the associated compute. As a result, we could incur 
significant cash outflows without obtaining usable compute on a timely basis, which could adversely affect our 
liquidity, financial condition, and results of operations.

Certain of our data center infrastructure arrangements involve complex multi-party structures in which chips 
are procured from or through our CSPs, leased to us by third-party financing entities, installed in data center 
facilities  procured  by  or  through  other  third-party  partners,  and  operated  by  yet  another  set  of  third-party 
service providers. The complexity of these arrangements increases execution risk, as delays or failures by any 
participant in the chain could affect the availability or performance of the resulting compute. 

In addition, to meet the substantial energy demands of large-scale AI workloads, we have entered into, and may 
in  the  future  enter  into,  arrangements  involving  dedicated  power  generation  assets  and/or  the  development, 
operation,  or  control  of  power  stations  or  other  energy  infrastructure.  Such  arrangements  expose  us  to  new 
operational, regulatory, and environmental liabilities, and require expertise and capabilities that differ from our 
core  competencies.  Successfully  executing  on  these  power  infrastructure  projects  requires  us  to  recruit 
personnel  with  relevant  experience  and  manage  complex  multi-year  deployment  timelines  and  relationships 
with developers, contractors, service providers, and regulators, including aligning project completion schedules 
with  the  data  centers  such  projects  are  expected  to  serve  in  order  to  avoid  timing  delays  that  constrain  our 
ability to utilize this compute and to generate revenue and recoup costs from such investments.

Deployment of our data center and power infrastructure projects involves execution risks, including delays due 
to permitting issues, change in law, force majeure, lack of network connectivity between facilities, supply chain 
disruptions,  labor  shortages,  or  contractor  performance;  cost  overruns  due  to  equipment  pricing  changes, 
design  modifications,  or  unanticipated  requirements;  and  technical  risks  that  facilities  may  not  perform  as 
specified  or  achieve  expected  reliability  or  efficiency  levels.  Certain  of  our  arrangements  are  also  subject  to 
conditions precedent that must be satisfied before construction can commence, and completion guarantees that 
may  terminate  upon  a  lender  foreclosure  before  construction  is  complete,  leaving  us  without  contractual 
recourse  to  compel  completion.  Certain  of  these  arrangements  also  depend  on  the  continued  performance  of 
credit  support  provided  by  a  third  party  that  is  also  a  competitor,  the  loss  or  impairment  of  which  could 
jeopardize the underlying arrangements. Such credit support provided to third-party data center developers or 
hardware  providers,  from  which  we  derive  no  direct  contractual  rights  or  obligations,  may  also  affect  the 
availability  and  cost  of  data  center  capacity  and  hardware  generally.  Additionally,  new  tariffs  or  import  fees 
enacted after the effective date of certain arrangements may be passed through to us as additional costs. Any 
delay in or limitation on the amount of compute delivered to us could adversely affect our business.

The capital-intensive and long-term nature of leasing and equipping data center capacity creates financial risk, 
as  large  upfront  investments  and  long-term  payment  commitments  must  be  made  months  before  facilities 
become fully operational. If our compute requirements or technology needs change, we may find ourselves with 
leased  or  dedicated  infrastructure  that  is  poorly  suited  to  our  actual  needs,  and  technological  changes  in  AI 
hardware  or  training  methodologies  could  render  such  infrastructure  less  efficient  than  newer  alternatives. 
Unlike  CSP  arrangements  where  compute  can  be  adjusted  relatively  dynamically,  long-term  leases,  reserved 
compute,  and  dedicated  infrastructure  investments  are  much  less  flexible  and  create  long-term  financial 
obligations  regardless  of  whether  the  capacity  continues  to  meet  our  needs.  For  example,  certain  of  our 
dedicated  power  and  data  center  infrastructure  arrangements  have  initial  terms  of  20  years  with  annual 

75