Billionaire-Backed Data Center Plans Divide Oakland

Google News put Oakland’s data center dispute before a wider audience after three billionaire-linked companies disclosed plans involving 160,000 square feet of local property. Prometheus, Tesla, and the Behring Company describe different facilities, tenants, and development paths. Their projects have nevertheless merged into one political conflict over AI infrastructure, public oversight, and who benefits from Oakland’s next technology cycle.

The dispute is not simply between technology supporters and opponents. It is between project-by-project review and a precautionary effort to stop data center development before Oakland establishes specific rules. Councilmember Carroll Fife says no project has received final approval and is working on a possible moratorium. Activists argue that waiting for individual applications gives developers too much influence before residents understand the cumulative effects.

That tension has become especially sharp because the proposals differ greatly. One involves an existing West Oakland industrial property. Another would reuse a former federal supercomputing facility downtown. Treating every server installation like a vast hyperscale campus obscures those distinctions. Approving unfamiliar projects without verified energy, water, pollution, and employment data creates a different problem.

Google News Amplified Three Separate Oakland Projects

The Oakland controversy combines three proposals whose footprints, operators, and approval requirements remain materially different.

Prometheus has reportedly leased 100,000 square feet at American Steel Blocks on Mandela Parkway in West Oakland. The artificial intelligence startup was co-founded by Amazon founder Jeff Bezos and former Google executive Vik Bajaj. Public information about its intended computing capacity, cooling equipment, and electricity demand remains limited.

Tesla reportedly leased another 40,000 square feet at the same industrial property. Elon Musk leads Tesla, although the company has not publicly detailed the Oakland site’s purpose. Tesla and Prometheus did not respond to questions for the original Bay Area News Group coverage.

The third project comes from the Behring Company, led by developer Colin Behring. It concerns 415 20th Street, a four-story downtown building previously used by Lawrence Berkeley National Laboratory for supercomputing. The company has discussed allocating approximately 20,000 square feet to technical infrastructure for AI and robotics companies.

Together, those disclosed areas total 160,000 square feet. That number describes floor area, not computing load. It does not reveal the number of servers, electrical demand, cooling design, backup generation, or operating schedule.

Those missing measurements matter more than the combined real estate figure. A compact installation filled with high-density accelerators can demand substantially more power than a larger conventional office. Conversely, a research floor inside an existing building is not automatically comparable to a newly constructed hyperscale campus.

The initial project details show why residents are asking for additional disclosure. Prometheus and Tesla have revealed little publicly about their planned operations. Behring has offered more detail, but its site still lacks a named technology tenant.

Behring says the downtown property was selected for its existing energy capacity, previous computing use, transit access, and nearby housing. The company also says the building can support office, laboratory, and technical functions within one site.

According to the developer, the technical area would occupy one floor. It would use closed-loop cooling, which recirculates coolant instead of continuously consuming fresh water. Behring says this makes the proposal fundamentally different from a large campus using water-intensive cooling towers.

That distinction has not settled the argument. The cooling design has not been independently assessed in a published environmental review. The lack of an announced tenant also prevents residents from testing the developer’s description against a specific computing workload.

The event that changed Oakland’s debate was therefore not a city approval. It was the arrival of several proposals before the city had a shared framework for classifying them. Google News gave that local uncertainty national visibility, but aggregation did not resolve the underlying facts.

Oakland Must Decide What Counts as a Data Center

The central policy problem is classification, because the same label now covers a technical office floor and a power-intensive computing campus.

Data centers are buildings or dedicated spaces containing servers, storage, networking equipment, cooling systems, and backup power. That definition spans small enterprise rooms, research clusters, colocation facilities, and enormous AI campuses. Their neighborhood effects vary accordingly.

Behring has compared his proposed technical floor with a hypothetical 500-acre, gigawatt-scale campus. His point is that size, power demand, and infrastructure design should determine regulation. Activists answer that developers should disclose those measurements before asking the public to accept the comparison.

The downtown proposal offers a useful test. The building already supported federal supercomputing work, and the developer does not propose a new sprawling structure. Those facts can reduce land-use disruption. They do not establish the future project’s energy load or environmental performance.

The downtown proposal would use only part of the existing property, according to Behring. He also says he has given 15 tours to startups interested in the space. No disclosed lease identifies an operator or confirms what hardware would run there.

This uncertainty weakens both sweeping positions. The public record does not support calling the project a giant campus. It also does not support treating the technical floor as an ordinary office without examining its equipment.

Oakland’s zoning process will shape the next step. If a project exceeds the activities already allowed at its location, it can require further review and a City Council vote. Developers may argue that previous computing use or existing industrial permissions cover much of their intended work.

That creates a permit-based version of the larger conflict. Project supporters see adaptive reuse, which means assigning a new purpose to an existing property, as a faster route to downtown investment. Opponents see a potential gap that could let AI infrastructure advance without a full public assessment.

Councilmember Fife has said that nothing has been approved and that the proposals only reached the planning process. She has also said she is developing a possible moratorium. Her position places her between residents demanding a ban and developers seeking individual consideration.

That middle position has become politically difficult. At an August 21 community meeting, activists pressed for a clear halt to data centers. Fife argued that advocates should distinguish verified project facts from speculation and condemned threats connected with the dispute.

Councilmember Charlene Wang has called for more transparency before assessing the projects’ merits. Representative Lateefah Simon has also expressed concern about the growth of data centers and their community effects. Both positions focus on disclosure rather than immediate endorsement.

The most useful classification system would therefore begin with measurable thresholds. Relevant categories include electrical load, water source, cooling design, backup generator emissions, construction footprint, operating noise, and proximity to housing. Without those figures, “data center” remains a politically potent but technically imprecise label.

The Main Conflict Is Local Control Versus Case-by-Case Development

Oakland’s primary divide concerns who sets the terms before development advances, not whether computing infrastructure should exist anywhere.

The Party for Socialism and Liberation has organized a “No Data Centers in Oakland” campaign. Organizers say their petition has gathered more than 6,700 signatures. They cite electricity demand, water use, pollution, displacement, and limited public participation as reasons for a ban or moratorium.

Developers and some technology investors favor evaluating each proposal on its actual design. They argue that Oakland risks rejecting investment based on environmental problems associated with much larger facilities elsewhere. That approach requires credible project-level data to work.

Fife has not publicly adopted either absolute position. She says she supports responsible technology, sees potential economic value in downtown reuse, and wants safeguards. She also says she has been working on a moratorium proposal.

This positioning produced the reported rift. Some activists interpreted Fife’s public appearance beside Behring as support for the project. Fife denied that characterization and said her office was trying to lower tensions while establishing the facts.

The disagreement mirrors a broader Bay Area shift. San Jose spent years actively assembling utility capacity and a development pipeline for data centers. It now faces growing public pressure for stronger standards and greater community participation.

San Jose has approved more than one million square feet of data center development since Mayor Matt Mahan took office, according to regional development data. Thirty-four additional projects were reportedly under review in late August 2026.

San Jose also reached an agreement with Pacific Gas and Electric for upgrades intended to deliver more than 2,000 megawatts. City leaders framed that capacity as an economic development asset. Residents increasingly view it as a commitment made before the public fully examined the consequences.

Oakland is earlier in that cycle. Its officials can set disclosure rules before a large pipeline forms. However, a broad moratorium could treat a small research installation, an existing colocation facility, and a new hyperscale campus as identical uses.

Case-by-case review presents the opposite risk. If each project appears manageable alone, officials may overlook cumulative demand across several sites. Three moderate installations can still compete for grid capacity, require backup generation, and concentrate industrial effects in particular neighborhoods.

The West Oakland location adds an environmental justice concern. That term describes the unequal concentration of pollution and infrastructure burdens in lower-income communities or communities of color. Residents want developers to address Oakland’s history rather than discuss each property in isolation.

A workable review would therefore combine individual measurements with cumulative analysis. Each developer would disclose its expected peak load, water demand, generator operations, construction effects, and permanent employment. The city would then assess those figures alongside existing neighborhood burdens.

Local benefits also require definition. Construction work, property taxes, startup offices, and occupied buildings can support the economy. Permanent data center staffing is often more limited than the employment created during construction.

Behring has acknowledged this uncertainty. His company has said that any specific jobs estimate would be a guess because the building’s final use is undecided. That is more cautious than promising a large employment windfall, but it leaves a key public benefit unquantified.

Oakland’s choice is thus procedural before it becomes ideological. The city must decide whether developers carry the burden of proving limited impact or residents carry the burden of proving unacceptable harm. That decision will shape every proposal that follows.

What the Developers’ Claims Do Not Establish

Small floor area and closed-loop cooling address parts of the criticism, but neither proves that a project has low community impact.

Behring’s strongest argument is grounded in the property itself. The downtown building exists, previously supported advanced computing, and includes infrastructure that could support another technical use. Reusing it avoids the land disturbance associated with constructing a vast new campus.

The closed-loop cooling claim is also relevant. Such systems can consume less water during regular operation than evaporative cooling. Yet water efficiency does not reveal electricity demand, refrigerant use, emergency procedures, or the emissions from backup generators.

Computing density is the key unknown. AI accelerators can concentrate substantial electrical demand within a limited area. A square-foot comparison cannot determine that load because equipment configurations vary widely.

Prometheus and Tesla have disclosed even less about their West Oakland space. The public does not know whether the leases support model training, inference, research, vehicle data processing, offices, or mixed uses. Each activity carries a different infrastructure profile.

Claims about permit exemptions also require scrutiny. Previous use can simplify some approvals, but it does not automatically establish that a new operator’s equipment matches the old facility. Current zoning, electrical work, generators, cooling equipment, and exterior changes can trigger separate requirements.

The skeptical question is not whether every developer is concealing a hyperscale project. No available evidence establishes that. The question is whether Oakland can verify project descriptions before issuing approvals that become difficult to revisit.

Professor Hiu Yung Wong of San Jose State University told Bay Area News Group that little was known about the three proposed facilities. He noted that large data centers can use extensive power and water. His general examples should not be treated as measurements for the Oakland sites.

That distinction matters. Applying national worst-case numbers to an unmeasured 20,000-square-foot technical floor would overstate the evidence. Accepting a developer’s best-case description without engineering documents would understate the uncertainty.

Nationally, communities are challenging data center proposals after discovering that conventional zoning rules provide few answers. The community opposition includes concerns about utility infrastructure, water, noise, land use, and public access to negotiations.

The secrecy surrounding some developments has intensified distrust. Developers sometimes operate through limited-liability companies or request confidentiality while negotiating for land and electricity. Oakland’s known companies reduce some anonymity, but the future tenants and technical specifications remain unclear.

The projects also lack independently verified employment forecasts. Construction can create substantial temporary work, while operating facilities often rely on smaller technical teams. Office and research tenants could expand employment, but those tenants have not been secured publicly.

Tax benefits remain uncertain for the same reason. Property reuse can return vacant space to productive use. The fiscal effect depends on ownership, equipment valuation, applicable incentives, and the services required to support the facility.

Oakland should avoid converting these unknowns into either promises or accusations. A defensible process would require load studies, cooling specifications, generator permits, construction plans, and workforce estimates before officials characterize any project’s impact.

Public disclosure also needs a common format. Residents cannot compare three proposals when one developer discusses floor area, another discloses only a lease, and none provides the same infrastructure metrics. Standardized filings would turn a rhetorical dispute into a reviewable record.

Oakland’s Fight Has Become Part of National Data Center Politics

The Oakland dispute arrived as data centers shifted from an obscure land-use category into a national political liability.

AI companies need growing amounts of computing capacity to train and operate models. Developers are searching for buildings, land, grid connections, and communities willing to host that capacity. Local resistance increasingly determines whether those projects proceed.

A 2026 Annenberg survey found that 61 percent of Americans opposed a new data center in their community. Opposition included 54 percent of Republicans, according to the published public opinion results. That bipartisan resistance has changed how elected officials present data center policy.

Political leaders now stress ratepayer protections, water limits, community benefits, and public disclosure. Industry advocates increasingly emphasize closed-loop cooling, construction employment, taxes, and dedicated utility tariffs. A tariff in this context is a rate structure that assigns infrastructure costs to large electricity users.

The debate escalated further when President Donald Trump publicly defended data center development. His intervention tied local infrastructure disputes to national arguments about economic growth, AI leadership, and household costs.

An industry-aligned advocacy organization linked to a pro-AI political network reportedly prepared a $50 million campaign supporting data centers in several states. The national campaign shows that local permitting fights now affect electoral strategy as well as technology investment.

Oakland presents an unusual version of that national conflict. The disclosed projects are smaller than the vast campuses driving many national headlines. They are also connected with some of technology’s most recognizable billionaires through Prometheus and Tesla.

That association makes the proposals easier to frame as outside capital imposing costs on a working-class city. It also risks reducing distinct projects to the personal reputations of Bezos and Musk. Oakland still needs to judge the physical facilities, not only their financial backers.

The city’s economic position makes the calculation difficult. Downtown Oakland has struggled with vacant commercial property and reduced activity. Reusing an existing technical building can appear attractive when conventional office demand remains weak.

However, distressed property does not remove the need for public conditions. A city can welcome investment while requiring developers to disclose utility demands, fund infrastructure, limit emissions, and report employment. Those requirements become more credible when adopted before individual approvals.

The Bay Area comparison also challenges a simple growth narrative. San Jose deliberately pursued data center capacity, yet its leaders now face calls for stricter rules. Oakland can study that experience before it accumulates a similar project queue.

Another precedent comes from communities where officials negotiated projects before residents understood the operators or power requirements. Those cases turned confidentiality into a symbol of exclusion. Oakland can reduce that risk by publishing applications and technical studies early.

This is why the story traveled through Google News beyond its original local audience. Oakland’s argument captures the national transition from enthusiasm about AI infrastructure to scrutiny of where it goes, what it consumes, and who makes the decision.

The dispute also exposes a communications failure within the industry. “Data center” is too broad to tell residents what a proposal will do. Developers who want individualized treatment must provide individualized evidence.

Activists face a corresponding challenge. Their strongest case rests on cumulative burdens, enforceable safeguards, and unresolved project data. Claims based on megacampus examples become less persuasive when applied automatically to a reused downtown floor.

Three Signals Will Determine What Happens Next

The next phase depends on Oakland’s moratorium language, verified infrastructure disclosures, and the identity of each project’s operating tenant.

The first signal is Fife’s proposed moratorium. Its scope will show whether Oakland intends a temporary planning pause or a broad rejection of data center development. Definitions and exemptions will be more important than the measure’s label.

A temporary pause tied to rulemaking would strengthen the case that the city wants measurable standards before approvals. An indefinite ban covering every server-intensive use would support developers’ argument that materially different projects are being grouped together.

The second signal is the publication of comparable technical data. Prometheus, Tesla, and Behring should disclose expected peak electrical demand, cooling technology, annual water consumption, backup generation, operating hours, and construction requirements.

Low verified loads and limited generator use would strengthen Behring’s claim that the downtown project differs from a hyperscale campus. Large loads or missing documentation would strengthen calls for a moratorium and cumulative environmental review.

The third signal is tenant identification. Behring has not announced the operator that would use its technical space. Prometheus and Tesla have not fully explained their planned Oakland workloads.

A named tenant with an enforceable operating plan would give residents something concrete to evaluate. Continued ambiguity would keep the argument focused on billionaire backing and national worst-case examples rather than local engineering facts.

Employment commitments should accompany those disclosures. The city needs separate estimates for construction jobs, permanent facility roles, office workers, and research teams. Combining those categories can make a limited operating workforce appear larger than it is.

Oakland should also clarify which decisions require public hearings. Residents currently hear that nothing is approved, while developers discuss leases, tours, and existing-use rights. A published approval timeline would show where public input can still change each project.

That transparency would lower the temperature without predetermining an outcome. It would also help Google News readers distinguish a proposal, a private lease, a planning application, and a final authorization.

The strongest conclusion currently available is narrow. Oakland has several billionaire-linked technology projects, but it does not yet have enough standardized information to measure their combined impact. The projects should not be described as one giant facility, and their backers’ assurances should not replace verification.

Watch the moratorium text first, the engineering disclosures second, and the tenants third. Those signals will reveal whether Oakland is evaluating actual facilities or fighting over the symbolism of AI infrastructure.

For readers following this story through Google News, the useful question is not whether data centers are universally good or bad. Ask what each operator must disclose before receiving approval. Then ask whether Oakland will measure cumulative neighborhood effects alongside individual project claims. That standard offers a clearer test than billionaire names, promotional promises, or comparisons with distant megacampuses.