Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minutePC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Phaidra announced a $25 million Series A on July 15, 2022, led by Starshot Capital, with participation from Helena, Flying Fish, Ahren Innovation Capital, Section 32, Character and Mustafa Suleyman, a DeepMind co-founder investing personally. GeekWire reported the round on July 18 and put Phaidra’s total funding at $30.5 million, with no valuation disclosed. The Seattle company was building supervisory AI systems that use facility data to optimize energy-intensive operations—not a general-purpose chatbot and not a replacement for safety systems.
The financing was an important milestone, but it is now a historical round rather than Phaidra’s latest funding. The company announced another $12 million led by Index Ventures in July 2024 and a Series B of more than $50 million led by Collaborative Fund in October 2025. Its current product positioning is increasingly centered on data centers and “AI factories.”
What Phaidra raised in 2022
| Detail | Reported information |
|---|---|
| Round | $25 million Series A |
| Announcement | July 15, 2022; GeekWire coverage followed on July 18 |
| Lead investor | Starshot Capital |
| Other participants | Helena, Flying Fish, Ahren Innovation Capital, Section 32, Character and Mustafa Suleyman |
| Total raised at the time | $30.5 million |
| Valuation | Not disclosed |
GeekWire’s report described the business as AI for industrial efficiency, stability and sustainability. Phaidra’s contemporaneous announcements used similar language.
Suleyman’s participation was a personal investment. It did not mean that DeepMind or Google invested in Phaidra, and Phaidra was not a DeepMind corporate spinout.
Recommended Free Tools
#1 Best Overall
Who founded Phaidra?
Phaidra says its founders launched the company in 2019 after working on AI-based data-center cooling and seeing an opportunity to apply reinforcement learning to complex physical systems. The founding group combined machine-learning experience with industrial-controls knowledge.
Jim Gao
Gao is co-founder and chief executive. He previously led DeepMind Energy and worked on Google data-center operations.
Vedavyas Panneershelvam
Panneershelvam is co-founder and chief technology officer, a DeepMind alumnus and member of the AlphaGo engineering team.
Katherine (Katie) Hoffman
Hoffman is a co-founder and operations leader with industrial-controls experience at Trane and later Raytheon-related work.
That background matters: the company’s proposition depends as much on connecting to real equipment and operating procedures as on the underlying AI.
What Phaidra actually sells
In 2022, Phaidra marketed AI-powered control systems for facilities such as data centers, pharmaceutical and vaccine plants, paper and pulp mills, chemical facilities, refineries and steel mills. The system is intended to connect with existing industrial-control or building-management infrastructure, ingest sensor data, model how a particular plant behaves and adjust operating settings within configured limits.
Its current product pages describe a narrower commercial emphasis:
- Phaidra Factory: specialized AI agents for mission-critical data-center infrastructure.
- Phaidra Prism: an operational language-model product intended to help technicians identify, prioritize and troubleshoot issues.
The present focus is managing power, cooling and workloads in large AI data centers. See the company’s product descriptions.
How the control technology works
- Collect facility data. Sensors, historians and existing building or industrial-control systems provide temperatures, pressures, flows, electrical readings and equipment status.
- Build a site-specific model. Phaidra combines knowledge of facility components with models learned from that site’s sensor history.
- Evaluate operating choices. A reinforcement-learning agent estimates the effects of possible changes, such as cooling or electrical set-point adjustments.
- Act within constraints. The system can make supervisory-level changes subject to configured limits, escalation rules and manual override.
- Learn from results. Operational outcomes feed back into the model so it can refine future decisions.
This is not a claim that Phaidra replaces programmable logic controllers, emergency shutdowns, safety interlocks or human operators. It is an AI layer interacting with existing controls.
Customers and reported performance
Named customer
TechCrunch identified Merck as an early customer, with Phaidra deployed at a large vaccine-manufacturing facility. Public reporting also mentions applications in paper mills, data centers, chemical manufacturing, refineries, steel mills and other industrial sites. Those categories should not be read as a complete customer list.
Energy and operating claims
Phaidra said its systems could reduce plant energy consumption by up to 30%. Helena described target ranges of 15%–30% lower energy costs, process-stability improvements of up to 70% and as much as 50% less equipment runtime. Phaidra’s DeepMind-related data-center work was described as producing roughly 30% cooling-energy savings.
These are company, investor or project claims, not an independently audited average across all deployments. Results can vary with weather, production schedules, electricity prices, equipment condition, baseline efficiency and the quality of the control data. An “up to” figure is not a guaranteed saving.
Sources for the reported figures include GeekWire, Helena and Phaidra’s 2024 financing announcement.
Why the Series A mattered
Industrial AI is difficult to commercialize because every plant has different equipment, telemetry, operating rules and risk tolerances. The funding gave Phaidra capital for research and development, plant implementation, customer success, hiring and expansion into data centers.
The investor group also supplied credibility across several relevant domains: climate technology, industrial systems and AI. Suleyman’s involvement connected Phaidra to DeepMind’s founding generation, while the operating company’s credibility still depended on Gao, Panneershelvam, Hoffman and its deployment teams.
Rank #4
How Phaidra makes money
TechCrunch reported that Phaidra uses a SaaS-like annual subscription whose price depends on facility complexity and local energy prices. The commercial model can also involve implementation, integration and ongoing customer-success work. Phaidra does not publish a standard rate card; its product site directs prospective customers to request a demo.
Free tools Windows power users keep installed
One-click scans. No signup required.
That model creates a trade-off for buyers. A recurring fee may be justified when optimization cuts energy costs, improves stability or frees electrical capacity for additional computing. But integration with legacy controls can be slow and expensive, and the economics depend on each facility’s baseline, energy prices, downtime exposure and ability to provide reliable data.
What changed after 2022
| Date | Development | What it indicates |
|---|---|---|
| July 2, 2024 | $12 million round led by Index Ventures; Phaidra said total funding reached $60.5 million and its workforce was about 100 | Continued expansion, with data centers becoming more prominent |
| October 1, 2025 | Series B of more than $50 million led by Collaborative Fund, with Helena, Index Ventures, NVIDIA, Sony Innovation Fund and others participating | A larger push into AI-factory infrastructure |
Those announcements are available through Phaidra’s news archive, the 2024 release and the 2025 Series B announcement.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What a prospective customer should evaluate
Control compatibility
Confirm interfaces to the site’s building-management system, HVAC and cooling equipment, electrical infrastructure, liquid-cooling systems and workload controls. Protocol, historian and data-quality compatibility can determine whether a deployment is practical.
Safety and authority
Define which actions are recommendations and which can execute automatically. Require hard operating limits, escalation paths, manual override, alarm handling and a tested rollback plan.
Quick wins for a faster PC:
Clear out junk files and repair common Windows errorsFree Scan →Scan for outdated or missing drivers - takes under a minuteDriver Scan →Deployment workload
Expect site-specific commissioning, sensor cleanup, historical-data review and coordination with operators, engineers and existing controls contractors.
Measurement and economics
Agree on a baseline and normalize for weather, production and workload. Separate energy savings from improved throughput, stability, avoided capital expenditure and capacity freed for compute. Ask how subscription, integration and support costs affect payback.
Cybersecurity and resilience
Review network segmentation, cloud connectivity, identity and access controls, data retention, incident response and behavior during network or cloud outages. The buyer should be able to isolate or disable the AI layer without losing essential safety functions.
Risks and unanswered questions
- Poor, drifting or incomplete sensor data can degrade recommendations.
- Equipment changes can make a site-specific model stale.
- A system that optimizes energy may harm throughput, redundancy or equipment life if objectives and constraints are incomplete.
- Operators may disable automation after an unexpected event, reducing realized savings.
- Short measurement windows can confuse seasonal or production effects with AI impact.
- Models configured for one facility may not transfer cleanly to another.
- Public information does not establish contract values, deployment counts, retention, annual recurring revenue or profitability.
Bottom line for investors and industry readers
Phaidra’s 2022 Series A financed an ambitious attempt to apply reinforcement learning to real industrial infrastructure. The company had a named early customer, a technically relevant founding team and reported energy and stability benefits, but those benefits require facility-by-facility verification. The company has since raised substantially more capital and concentrated its public positioning on data centers and AI factories. Its significance is therefore less the celebrity of one investor than whether it can repeatedly integrate with complex controls, operate safely and prove durable economic value.
The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
