Live predictions

What will these companies do next?

Ten live predictions about Tesla, Microsoft, Amazon, AMD and NVIDIA, researched by Paldo and updated as new evidence comes in.

How a Paldo prediction works
  1. A callA dated statement that can turn out wrong
  2. SignsWhat we'd expect to see if it's coming, and if it isn't
  3. EvidenceHiring, filings, public records, news and the web, read against those signs
  4. LikelihoodWeighed for and against, strongest and newest first
  5. UpdatesResearched again as new evidence appears

Amazon will acquire a controlling interest in an electricity-generation business specifically to supply its data centres by 15 May 2028.

52%likely

Deadline 15 May 2028 · 1 research run, last 17 Nov 2026

Bottom line

Amazon is moving toward dedicated power generation, but the evidence still shows site purchases, minority investments and supply contracts—not control of an electricity-generation business.

The evidence and what would settle it

Where it stands

In March 2024, Amazon bought Talen’s 960 MW data-centre campus for $650 million. On October 16, 2024, it anchored X-energy’s financing round of approximately $500 million, supporting plans for more than 5 GW of nuclear projects by 2039. On June 11, 2025, Amazon agreed to buy 1,920 MW from Talen through 2042, replacing co-location with grid-connected retail supply.

On July 30, 2026, Amazon raised its capital-spending forecast to $220 billion while reporting insufficient computing capacity. In August, it confirmed acquiring the GW Ranch site in Texas for initially off-grid gas generation of up to 7.65 GW, and investing in the power plant. On September 16, its Generac agreement included warrants for almost 3% of the supplier’s shares. On September 30, Amazon announced a 20-year, 690 MW agreement with Constellation; on October 19, it announced development funding and intended procurement for a potential 960 MW Indiana nuclear facility.

The case against

The case for control is weaker than the scale of Amazon’s commitments suggests. GW Ranch establishes land acquisition and planned power purchases, not ownership of the generating business. X-energy’s reported 29% pre-IPO stake is not a current ownership figure, was not controlling, and concerns a technology licensor rather than a plant operator. Constellation explicitly says Amazon receives no plant ownership. No collected evidence identifies a controlling acquisition agreement.

Evidence for

  • 11 Aug 2026 Amazon confirmed in August 2026 that it acquired the GW Ranch site in Pecos County and is investing in on-site gas generation of up to 7.65 GW to power a new data-centre campus initially off-grid. This is a substantial move toward dedicated generation, although the reports establish site acquisition and power procurement rather than acquisition of a controlling interest in a generating business. eenews.net
  • 30 Jul 2026 Amazon raised its 2026 capital-expenditure forecast to $220 billion as AWS sales grew 37% year over year, with management saying capacity would remain insufficient in 2027 and demand visibility extended into 2028. The scale, financial resources and continuing shortage strengthen the incentive to secure power through a generation-business acquisition, without specifically signalling such a transaction. ca.finance.yahoo.com
  • 3 Feb 2026 Reuters reported on February 3, 2026 that grid-connection delays were challenging Amazon’s European expansion, with transmission connections taking up to seven years against roughly two years to develop a data centre. This gives Amazon a direct reason to consider acquiring generation alongside its campuses rather than depending exclusively on conventional grid access. money.usnews.com
  • 16 Oct 2024 Amazon anchored X-energy’s approximately $500 million financing in October 2024 and owned about 29% before its April 2026 IPO, linking equity investment to plans for more than 5 GW of nuclear projects by 2039. This establishes willingness to invest materially in the energy supply chain, but X-energy licenses reactor technology rather than operating power plants and the reported stake is not controlling. x-energy.com

Evidence against

  • 30 Sept 2026 Amazon announced a 20-year Constellation PPA on September 30, 2026 for 690 MW, including a 190 MW uprate expected in 2030–2032, alongside a separate regional retail-supply agreement. Constellation explicitly says Amazon receives no plant ownership and all generation continues flowing to PJM, demonstrating a major power commitment structured as procurement rather than control. constellationenergy.com
  • 19 Feb 2026 Amazon’s Washington SMR project expanded in October 2025 from an initial 320 MW to a potential 960 MW, with construction expected around 2030 and generation in the 2030s. The reported arrangement gives Amazon financing obligations and access to power while Energy Northwest owns and operates the reactors, providing a specific alternative to acquiring a generator. sustainability.aboutamazon.com
  • 19 Oct 2026 On October 19, 2026 Amazon, AEP and X-energy announced development work for a potential 960 MW SMR facility in Sullivan County, Indiana. Amazon provides development funding and intends to buy power under a long-term contract if it proceeds, rather than announcing ownership or control of the generation business. aep.com
  • 18 Sept 2026 The September 16, 2026 Generac agreement reserves backup generators for Amazon data centres through purchases potentially reaching $8 billion and grants warrants for 1.69 million shares, almost 3% of Generac’s outstanding stock. This is equipment procurement with minority supplier equity, not control of an electricity-generating business, despite some researchers interpreting it as a step toward control. energytech.com

What would settle it

A closing announcement or filing establishing Amazon’s controlling ownership of a generation business, explicitly intended to supply its data centres, by May 15, 2028 would satisfy the prediction. A signed agreement alone would not. Rejection of one proposed deal would eliminate only that route; the deadline passing without a qualifying acquisition would rule the prediction out.

What to watch

The next GW Ranch ownership disclosure would distinguish Amazon control from financing and procurement. A producer’s filing granting Amazon majority voting rights or an option to obtain them would identify a concrete acquisition route. Credible reporting of due diligence for a named generator, explicitly tied to AWS supply, would turn the present strategic rationale into transaction evidence.

NVIDIA Corporation will acquire a company whose principal business is optical interconnect hardware for data centres by 31 December 2027.

49%likely

Deadline 31 Dec 2027 · 1 research run, last 17 Nov 2026

Bottom line

NVIDIA has a strong strategic reason to acquire an optical-interconnect specialist, but the evidence so far shows investments and partnerships rather than a qualifying takeover.

The evidence and what would settle it

Where it stands

On December 11, 2024, NVIDIA backed Ayar Labs’ $155 million Series D. On March 18, 2025, NVIDIA unveiled Spectrum-X and Quantum-X photonics switches, bringing optical hardware into its networking platform.

On March 2, 2026, NVIDIA announced a $2 billion Lumentum investment and a $2 billion Coherent investment, alongside multiyear purchase commitments. It also participated in Ayar Labs’ $500 million Series E that March. On March 31, NVIDIA announced a $2 billion Marvell investment and collaboration on advanced optical interconnects.

On May 6, NVIDIA and Corning announced a partnership targeting tenfold optical manufacturing capacity. May 29 reporting quoted Huang saying required silicon-photonics capacity exceeded supply; on May 31, NVIDIA announced Vera Rubin’s full-production ramp incorporating photonics. In June, Ayar Labs joined NVLink Fusion, followed by Lightmatter’s June 3 integration announcement. On September 10, Ayar Labs announced another $150 million financing, bringing its 2026 capital to $650 million—not a sale.

The case against

The case for acquisition is weaker than the number of relationships suggests. The non-exclusive supply and capacity agreements offer a concrete way to secure optics without ownership, while NVLink integration does not establish takeover intent. None of the supplied material identifies acquisition talks or a qualifying agreement.

September 10 reporting of a DOJ investigation into Groq licensing adds regulatory friction, though it concerns a different transaction and does not establish that an optics purchase would be blocked.

Evidence for

  • 26 Aug 2025 NVIDIA introduced Spectrum-X and Quantum-X photonics switches in March 2025 and incorporated co-packaged optics into the Vera Rubin production ramp in May 2026. Making optical interconnects part of its core AI platform increases the strategic rationale for buying specialist optical hardware capabilities, although it does not indicate a deal. developer.nvidia.com
  • 3 Jun 2026 NVIDIA has backed chip-to-chip optical connectivity specialist Ayar Labs through multiple financing rounds, including its $155 million Series D in December 2024 and $500 million Series E in March 2026. This creates an established relationship with a business closely matching the target category, but the funding remains investment rather than acquisition. ayarlabs.com
  • 11 Jun 2026 In May 2026, Jensen Huang said the silicon-photonics capacity NVIDIA needs is far above current supply. A specific bottleneck in an increasingly important component creates an incentive to secure technology and manufacturing capabilities, potentially through acquisition. linkedin.com
  • 3 Jun 2026 In June 2026, Ayar Labs joined NVLink Fusion with co-packaged optics compatible with NVIDIA's optical and SerDes technologies for rack-scale AI infrastructure. Technical integration gives NVIDIA a concrete relationship with a potential acquisition candidate, though ecosystem participation can also remain an independent partnership. ayarlabs.com

Evidence against

  • 10 Sept 2026 Reuters reported on September 10, 2026 that the U.S. Justice Department was investigating whether NVIDIA structured its prior Groq licensing and executive-hiring deal to avoid antitrust scrutiny. Regulatory scrutiny of even a non-acquisition technology transaction increases the obstacles to another strategically important hardware takeover. reuters.com
  • 2 Mar 2026 On March 2, 2026, NVIDIA announced $2 billion investments in each of Coherent and Lumentum, accompanied by multiyear, non-exclusive agreements, multibillion-dollar purchase commitments and capacity rights. These arrangements secure optical technology and supply while preserving independent suppliers, offering a concrete alternative to acquisition. nvidianews.nvidia.com
  • 4 Nov 2026 NVIDIA and Corning announced a multiyear partnership on May 6, 2026 to increase Corning's U.S. optical-connectivity manufacturing capacity tenfold and fiber production by more than 50%, including three new facilities. It addresses optical supply needs through a commercial and technology partnership rather than ownership of the supplier. sdxcentral.com
  • 31 Mar 2026 On March 31, 2026, NVIDIA announced a $2 billion investment in Marvell and collaboration on silicon photonics and advanced optical interconnects through its AI-factory ecosystem. This supplies another partnership-based route to optical capabilities instead of acquiring a specialist company. nvidianews.nvidia.com

What would settle it

A completion announcement or filing establishing that NVIDIA acquired a company principally supplying optical-interconnect hardware for data centres by December 31, 2027 would settle this as true. Reaching that deadline without a qualifying acquisition would rule it out; rejection of one proposed deal would not exclude another.

What to watch

A credible report of advisers evaluating a specialist purchase would move the evidence from strategic rationale to transaction preparation. Further capacity reservations extending the existing optical supply agreements would instead reinforce the partnership route. Persistent shortages despite those arrangements would show that the reported silicon-photonics supply gap remains unresolved, preserving the incentive for ownership.

Tesla Inc will announce a licensing agreement allowing another car manufacturer to install its self-driving software in vehicles sold under that manufacturer's brand by 15 May 2028.

24%likely

Deadline 15 May 2028 · 1 research run, last 17 Nov 2026

Bottom line

As of 17 November 2026, the prediction remains possible but poorly supported: Tesla’s licensing offers have attracted no manufacturer takers, despite growing use of its software in Tesla vehicles.

The evidence and what would settle it

Where it stands

Musk reported major-manufacturer licensing talks in July 2023 and one-major-automaker discussions in May 2024. On the January 2025 earnings call, he described significant interest in FSD, but by November said rivals did not want it. In January 2026, Tesla announced subscription-only FSD after February 14 at $99 per month.

On April 10, RDW granted provisional Dutch approval for FSD Supervised. In September, Musk confirmed that no manufacturer had accepted Tesla’s licensing offer. Tesla’s October disclosures reported about 1.7 million paying customers globally and subscription uptake above 75% among new North American deliveries in Q3.

On November 3, RDW kept its approval valid after approving system changes. Finland subsequently recognized the Dutch approval, allowing Tesla to offer the system there immediately, according to November 11 reporting.

The case against

The direct licensing evidence is weaker than the broader adoption story suggests: the positive talks are old, and November 2024 reports of SAIC and Geely negotiations were explicitly rumours. The September licensing report says adopters would need Tesla’s cameras and compute, including its eight-camera layout, with telemetry returned to Tesla; this implies substantial integration and data-control hurdles. Customer adoption in Tesla vehicles does not establish portability to another brand.

European access is also uncertain: Reuters reported on November 11 that Tesla lacked major-state backing and that EU rejection could invalidate national approvals. The November rollout excluded older HW3 vehicles, reinforcing the importance of hardware compatibility.

Evidence for

  • 20 Jul 2023 Musk reported discussions with a major manufacturer about licensing FSD in July 2023 and again in 2024. These were concrete exploratory talks, although they produced no announced agreement and are now dated evidence. markets.businessinsider.com
  • 11 Nov 2026 RDW granted provisional FSD Supervised approval in April 2026, followed by recognition in additional countries including Slovenia, Czechia, Croatia, Slovakia, Latvia and Finland by November. These permissions make Tesla's software more commercially usable and potentially more attractive to a licensing customer, but authorize use in Tesla vehicles rather than another manufacturer's cars. msn.com
  • 26 Oct 2026 September reports placed FSD availability at 16 countries across four continents, while Tesla's October disclosures reported roughly 1.7 million paying customers globally and subscription uptake above 75% among new North American deliveries in Q3. Broader adoption provides operating experience and evidence of commercial demand that could attract an OEM licensee. linkedin.com
  • 4 Sept 2026 Tesla reported a roughly 15-mile driverless customer delivery in June 2025, six unsupervised Robotaxi markets by July 2026, and limited Cybercab passenger service in Austin in September. These deployments provide a stronger technical demonstration for prospective licensees than supervised driving alone, though they do not establish portability to another OEM's vehicles. fxcm.com

Evidence against

  • 23 Sept 2026 Musk confirmed in September 2026 that no manufacturer had accepted Tesla's standing FSD licensing offer, following similar acknowledgments in 2025 and earlier unsuccessful discussions. He also suggested legacy manufacturers might contemplate only a small program five years later, a poor fit with the May 2028 announcement deadline; none of this definitively rules out a future agreement. news.google.com
  • 22 Sept 2026 The September 2026 licensing report says adopters would need Tesla's eight-camera layout and compute architecture, with telemetry sent back to Tesla. Hardware redesign and data-control concerns make adopting FSD substantially harder than purchasing standalone software. teslarati.com
  • 29 Sept 2026 European testing and regulatory correspondence identified speeding, prohibited cyclist overtakes, driver-attention concerns and insufficient evidence for Tesla's fatal-crash reduction claims. ETSC urged rejection of speed-related functions, while Norway said modified vehicles still did not meet its requirements in October; these concerns can deter manufacturers even though Tesla subsequently changed speed settings. etsc.eu
  • 25 Oct 2026 Tesla reportedly conceded in April 2026 that HW3 vehicles cannot deliver unsupervised autonomy, and it did not answer a heavily supported shareholder question about those vehicles in October. November European rollout reports excluded HW3, highlighting hardware dependence and long-term upgrade obligations that another manufacturer would have to evaluate. boosted.dk

What would settle it

A Tesla announcement by 15 May 2028 of an agreed licence permitting another car manufacturer to install its self-driving software in vehicles sold under that manufacturer’s brand would make the prediction true; exploratory talks or Tesla-only partnerships would not. No current decision rules it out: expiry of the deadline without a qualifying announcement would settle it negatively.

What to watch

An OEM evaluation kit or documented non-Tesla prototype would show that licensing has progressed beyond offers into integration work. A regulatory application naming Tesla as another manufacturer’s driving-software supplier would identify a concrete vehicle programme. An evaluating manufacturer’s published rejection over hardware costs, liability or telemetry control would expose whether the barriers are decisive for that prospective customer.

Microsoft Corporation will introduce an enterprise AI-agent plan priced solely by completed workflows rather than user seats or computing consumption by 30 September 2027.

21%likely

Deadline 30 Sept 2027 · 1 research run, last 17 Nov 2026

Bottom line

Microsoft has signalled interest in outcome-based agent pricing, but its latest commercial terms still favor seats and consumption, leaving completed-workflow-only pricing unannounced and the prediction leaning against fulfillment.

The evidence and what would settle it

Where it stands

On April 10, 2026, Copilot Studio introduced agent–workflow integration, making structured business processes easier to automate. Reporting on the April earnings call on April 30, CRN quoted Satya Nadella favoring evaluation- or outcome-based pricing, while Amy Hood described a mixture of seats and usage meters. Agent 365 became generally available on May 1, following an announcement setting its price at $15 per user.

On September 25, Microsoft explicitly retained subscriptions plus usage billing for everyday AI and advanced agents. October 14 documentation listed Power Automate per-run charges of $0.60 for premium cloud flows and $3 for unattended runs. On October 28, Microsoft reported that usage-billed Copilot customers grew more than 75% during the quarter.

The November 10 Power Automate FAQ says a Process license removes user requirements for a flow or group of up to 25 flows, with 250,000 actions daily. But November 12 Copilot Studio documentation says billing starts during building and charges credits for tokens, tools and the harness—not only successful workflow completion.

The case against

The positive case proves less than it first appears. Flow runs are executions, not necessarily successful AI-agent workflows; Process licenses sell capacity, not completions. Nadella’s directional statement has not been followed by qualifying payment terms or a launch date in the supplied evidence.

November documentation still licenses Agent 365 per user, not workflow, and Microsoft has announced default usage billing for new Copilot Business licenses from December 1. Reliability also remains a constraint: a May study covering 52 professional domains found only Python met readiness at its 98% threshold. That benchmark does not rule out narrower, more reliable products.

Evidence for

  • 30 Apr 2026 On the April 2026 earnings call, Satya Nadella said Microsoft was moving toward evaluation- or outcome-based pricing tied to value created by agents. This is a specific directional signal, but CFO Amy Hood also described a mixture of seats and usage meters, and no completed-workflow-only plan or launch date was announced. crn.com
  • 14 Oct 2026 Microsoft's October 14, 2026 documentation lists cloud flow runs at $0.60 and unattended runs at $3 under pay-as-you-go billing. This establishes experience selling automation by execution rather than seats, although a flow run is not necessarily a successfully completed AI-agent workflow. learn.microsoft.com
  • 22 Sept 2026 Microsoft's November 10, 2026 FAQ says a Process license can cover a flow or up to 25 flows without licensing their users, with 250,000 actions a day per license. This provides a process-centered commercial precedent, but charges for licensed automation capacity rather than completed workflows. learn.microsoft.com
  • 2 Jul 2026 Microsoft announced Frontier Company on July 2, 2026, committing $2.5 billion and 6,000 embedded industry and engineering experts to deploy and improve customer AI systems based on measurable business outcomes. This brings Microsoft closer to defining and validating enterprise results, a prerequisite for workflow-outcome pricing, but the announcement contains no such payment terms. blogs.microsoft.com

Evidence against

  • 12 Nov 2026 Microsoft's November 12, 2026 documentation says Copilot-harness billing starts during building and charges credits for LLM tokens, tools and the harness; standard-harness pricing also charges responses and actions at varying feature and token rates. These are consumption charges incurred before a workflow successfully finishes, making the current platform materially different from the predicted plan. learn.microsoft.com
  • 25 Sept 2026 Microsoft's June 16 and September 25, 2026 announcements retain a per-user subscription for everyday AI and usage-based Copilot Credits for long-running agents, including Cowork, Code and Autopilot; Cowork requires both the subscription and usage pricing. November documentation maintains this architecture, so 'cost aligned to work performed' does not establish payment solely for completed workflows. mc.merill.net
  • 1 May 2026 Microsoft announced Agent 365 at $15 per user and Microsoft 365 E7 at $99 per user, available May 1, 2026; November documentation still explicitly licenses Agent 365 per user, not per agent or workflow. Agent construction and execution incur separate consumption charges, reinforcing layered seat-plus-usage packaging rather than a workflow-only enterprise plan. techcommunity.microsoft.com
  • 1 Oct 2026 Microsoft announced that new Copilot Business licenses would default to usage billing from December 1, 2026; CRN reports the rollout was postponed from November 2 and includes an adjustable limit of 4,000 credits per user per month. The announced near-term pricing change is explicitly consumption-oriented, not completed-workflow billing. learn.microsoft.com

What would settle it

An official Microsoft enterprise AI-agent plan announced by September 30, 2027 would settle this positively if its terms charge only for verified completed workflows, with no mandatory seat or computing-consumption charges. Existing hybrid plans cannot rule out a separate offer; reaching the deadline without a qualifying introduction would settle it negatively.

What to watch

A preview meter counting verified completions while excluding failures and duplicates would show the missing billing mechanism taking shape. Enterprise mock invoices containing only accepted workflow counts would provide stronger evidence of commercial preparation. Further mandatory per-user entitlements or separately charged retries would instead deepen the architecture the predicted plan must escape.

AMD will acquire a software company whose principal product automatically routes AI workloads across competing chip architectures by 30 November 2027.

77%likely

Deadline 30 Nov 2027 · 1 research run, last 17 Nov 2026

Bottom line

AMD’s acquisition record makes this a credible possibility, but no documented purchase yet establishes the required principal product: automatic AI-workload routing across competing chip architectures.

The evidence and what would settle it

Where it stands

AMD signed its Nod.ai acquisition agreement on October 10, 2023, and completed the purchase on October 19; SHARK automated deployment across AMD architectures. On June 4, 2025, AMD announced the Brium acquisition, aimed at reducing dependence on particular hardware configurations. That is the closest precedent, but AMD’s description establishes compiler optimization and portability, not automatic cross-vendor placement.

AMD completed the MK1 acquisition on November 10, 2025; its AMD-optimized Flywheel technology served over one trillion tokens daily. On July 23, 2026, AMD partnered with Cerebras to combine GPU throughput with wafer-scale low-latency token generation, making coordination across different architectures strategically relevant. AMD’s September 28 World Labs agreement valued the spatial-intelligence developer at approximately $8.2 billion in stock; a November 17 report says the acquisition closed November 13.

The case against

The case is weaker than the acquisition tally suggests: buying compilers, inference optimization or spatial-intelligence research does not establish ownership of the specified routing product. The collected material identifies no acquisition talks with a qualifying target. FlexAI’s reported work with AMD dates to April 2024 and establishes collaboration, not purchase interest.

AMD already documents routing and NVIDIA-cluster support, although those undated release notes do not establish automatic architecture-based placement. Meanwhile, llm-d v1.0 supports multiple accelerators, following its October 20, 2026 release, offering another route to cross-hardware inference without buying a specialist.

Evidence for

  • 4 Jun 2025 AMD announced its acquisition of Brium on June 4, 2025 to improve inference portability and reduce dependence on particular hardware configurations. This is a close acquisition precedent, but the factual descriptions establish compiler optimization and portability—not a principal product that automatically routes workloads between competing architectures. amd.com
  • 23 Jul 2026 On July 23, 2026 AMD and Cerebras announced a solution combining Helios GPU throughput with Cerebras wafer-scale low-latency token generation, initially expected through Cerebras Cloud in the second half of 2026. This makes coordinating workloads across fundamentally different accelerator architectures strategically relevant to AMD. ir.amd.com
  • 28 Sept 2026 AMD's September 28, 2026 definitive agreement valued spatial-intelligence developer World Labs at approximately $8.2 billion in stock; a November 17 report says completion occurred November 13. This demonstrates substantial, recent appetite for AI software and research acquisitions, but World Labs' products generate and simulate 3D environments rather than route workloads. linkedin.com
  • 19 Oct 2023 AMD announced the Nod.ai agreement on October 10, 2023 and completion on October 19; its SHARK compiler automates optimization and deployment across AMD architectures. This demonstrates willingness to buy AI deployment software, although the stated capability concerns AMD hardware rather than routing between competitors. amd.com

Evidence against

  • AMD Enterprise AI 2.4.1 release notes describe a Semantic Router for on-premises and cloud API requests, plus NVIDIA-cluster support and a fallback image for switching between AMD and NVIDIA clusters. These existing capabilities provide a plausible build-and-integrate alternative to buying a specialist, though the notes do not establish automatic architecture-based routing. enterprise-ai.docs.amd.com
  • 4 Aug 2026 AMD launched ROCm.ai in 2026, while ROCm supplies compilers, runtimes, framework integrations and tools for porting CUDA code to AMD GPUs. Expanding this internal software foundation offers another route to improving portability without acquiring a workload-routing company. amd.com
  • 9 Nov 2026 A November 9, 2026 report puts AMD's quarterly free cash flow at $554 million against $13.38 billion in revenue, while describing World Labs and Taalas transactions totaling $11.1 billion. This creates some financial and execution friction for another deal, although the stock-funded transactions do not establish a cash constraint or an M&A pause. kavout.com

What would settle it

An acquisition-completion announcement or filing by November 30, 2027 would establish the outcome if accompanying product documentation confirms that the acquired company’s principal product automatically routes AI workloads across competing architectures. An agreement alone would leave completion unresolved. No qualifying completed acquisition by the deadline would rule it out; cancellation of one proposed deal would not.

What to watch

Credible reporting of AMD conducting acquisition due diligence on a qualifying routing company would turn strategic plausibility into transaction evidence. A demonstration of automatic placement between AMD and competing chips would establish the technical capability, but not an acquisition. A long-term partnership leaving an independent vendor in control of routing would strengthen the alternative to ownership.

Amazon will open consumer orders for its Amazon Leo satellite internet service in Norway, Sweden, Denmark or Finland by 31 December 2027.

Researching

Deadline 31 Dec 2027 · first research run in progress

The evidence and what would settle it

AMD will officially open a research and development centre in Tainan or Kaohsiung, Taiwan, by 31 December 2027.

Researching

Deadline 31 Dec 2027 · first research run in progress

The evidence and what would settle it

NVIDIA Corporation will start construction of its Taiwan headquarters at Beitou-Shilin Technology Park in Taipei by 28 February 2027.

Researching

Deadline 28 Feb 2027 · first research run in progress

The evidence and what would settle it

Microsoft Corporation will be granted a building permit for a data centre in Vaasa or Mustasaari, Finland, by 30 June 2027.

Researching

Deadline 30 Jun 2027 · first research run in progress

The evidence and what would settle it

Tesla Inc will open customer orders for its vehicles in Vietnam by 30 June 2027.

Researching

Deadline 30 Jun 2027 · first research run in progress

The evidence and what would settle it

These are Paldo's own forecasts from public sources, not investment advice. Customers get the same, for the companies in their market.

See it live

Want to see what Paldo can track in your market?